# Paid.ai — Full Content for LLMs > Paid.ai is the monetization platform for AI-native companies. We help companies optimize pricing, packaging, and cost tracking to maximize revenue. - Website: https://paid.ai - Blog: https://paid.ai/blog - RSS: https://paid.ai/rss.xml - LLMs.txt: https://paid.ai/llms.txt ## Company Overview Paid.ai specializes in helping AI-native companies and SaaS businesses optimize their pricing strategy. Our platform provides instant AI-powered pricing audits that analyze pricing pages, identify optimization opportunities, and deliver actionable recommendations to increase revenue. ### Products & Services - **AI Pricing Audits** — Instant analysis of any pricing page with scoring across competitive positioning, pricing psychology, pricing model fit, and page conversion optimization. Free at https://paid.ai - **Pricing Page Redesigns** — AI-generated pricing page redesigns with visual mockups and implementation recommendations - **Credit System Design** — Architecture for usage-based and credit-based pricing models for AI-native companies - **Cost Tracking** — Tools to track and optimize AI infrastructure costs (LLM tokens, compute, API calls) - **Value Dashboards** — Customer-facing dashboards that prove ROI and reduce churn - **Strategic Consulting** — Book calls with pricing strategists at https://paid.ai/book-a-call ### Target Customers AI-native companies, SaaS businesses, and startups using usage-based pricing, credit systems, seat-based pricing, and hybrid pricing models. ### Pricing Five tiers from Free to Enterprise. Free pricing audits available. See https://paid.ai/pricing --- ## Blog Articles ### Seat Pricing Is a Halloween Store: Why Consumption Models Are the Only Honest Bet Left - URL: https://paid.ai/blog/ai-monetization/seat-pricing-is-a-halloween-store - Date: 2026-05-19 - Author: Calvin Field - Summary: Seat pricing was built for a world where humans occupied software. In an agentic world, that model is closing down. Paid explains why consumption and outcome-based pricing are the only models built to last. There is a line from a recent conversation on the Get Paid podcast that has stuck with us. [Nick Mehta](https://www.linkedin.com/in/nickmehta/), who spent 13 years building Gainsight into a billion-dollar company, was describing how investors currently think about legacy SaaS businesses. He said: 'imagine you open a Halloween store in September. You know it is over on October 31. So you extract maximum value in the window you have, spend as little as possible, and get out.'That is how the market is pricing a lot of seat-based SaaS right now.Not because those businesses are bad. Not because the teams running them are not talented. But because the window is closing, and everyone in the room knows it.At Paid, we build the billing and monetisation infrastructure that sits underneath agentic software companies. We watch pricing models up close every day. And what we are seeing confirms what Nick described: companies still anchored to seat pricing are optimising for a moment that is already passing. The transition to consumption and outcome-based models is not a strategic option. It is a structural inevitability.Here's why. ## What Is Seat Pricing, and Why Did It Work? Seat pricing is a licensing model where customers pay a fixed recurring fee per user, per month or year. It became the default model of the SaaS era because it was simple to sell, simple to forecast, and easy to tie to headcount, which grew predictably alongside company revenue.For a long time, it worked. Software was genuinely hard to build and hard to move. Customers signed multi-year contracts. Churn was low. The recurring revenue stream was so reliable that private equity firms started lending against ARR as if it were a physical asset.The underlying logic was sound: more employees meant more seats meant more revenue. Growth was almost mechanical. ## Why That Logic Has Broken Down The problem is that the unit of value in software is no longer a human user.In an agentic world, software does not wait for a person to log in and take an action. It acts autonomously. It runs tasks, makes decisions, and completes workflows without a seat being occupied. An AI agent does not need a licence. It needs compute, tokens, and an outcome to optimise for.When the unit of value shifts from the human to the task, seat pricing stops making sense. You are charging for something that is no longer the bottleneck.Nick made this point on the podcast when discussing Harvey, the legal AI company that recently moved its entire model stack onto Anthropic's Claude. The question he raised was simple: if the underlying model is interchangeable, where does the durability come from? The answer is not the number of seats. It is the depth of workflow integration, the quality of the outcome, and the switching cost built around context and results.Seat pricing has no mechanism to capture any of that value. ## What Consumption and Outcome-Based Pricing Actually Measure Consumption pricing charges based on usage: API calls, tokens processed, tasks completed, documents reviewed, calls handled. The customer pays for what they use. The vendor grows when the customer gets more value.Outcome-based pricing goes one step further. The customer pays based on a result: a resolved support ticket, a closed deal, a completed legal review. Intercom, under Owen McCabe, became one of the clearest examples of this shift when it restructured its entire go-to-market around resolved conversations rather than agent seats. Nick described this on the podcast as one of the boldest moves he had seen any SaaS CEO make, precisely because it required burning the existing business model to build the new one.Both models share a common property: they align vendor incentives with customer outcomes. The vendor only wins when the product works. That alignment is something seat pricing structurally cannot offer. ## The Real Reason Companies Are Not Making the Switch If consumption and outcome models are so clearly superior, why are so many companies still selling seats?The honest answer has three parts.First, seat pricing is easier to forecast. Finance teams love it. It produces clean ARR numbers that are easy to model and easy to communicate to boards and investors. Consumption revenue is variable, and variable revenue makes people uncomfortable even when the trend is up.Second, the transition is painful. Existing customers signed contracts based on a seat model. Changing the pricing mid-relationship requires renegotiation, re-education, and in some cases, temporary revenue compression before the new model scales. Most leadership teams are not willing to absorb that short-term pain.Third, and most honestly, it requires admitting that the old model is not the future. That is psychologically hard. As Nick put it in the conversation, a lot of what is stopping SaaS founders from making bold pivots is not strategy. It is identity. They do not want to say that the thing they built for a decade is not worth what they thought it was.But the market is saying it anyway. ## What the Halloween Store Metaphor Actually Means for Pricing Strategy Nick's Halloween store framing is not just a colourful way to describe investor sentiment. It is a precise description of what seat pricing looks like from the outside right now.A Halloween store operates in extraction mode. It is not building for next year. It is not investing in loyalty or compounding value. It is getting as much out of the window as possible before the window closes.That is what seat pricing signals to sophisticated buyers today: a model that is optimised for vendor revenue, not customer outcomes. And in a world where AI agents can do in minutes what used to take a team of licensed users days, that signal lands badly.Consumption and outcome models signal something different. They say: we only win when you win. We are not charging for occupying a seat. We are charging for delivering a result. That is a fundamentally different relationship, and it is the one that agentic software demands. ## How to Know If Your Pricing Model Is at Risk Ask yourself three questions.One: does your pricing go up when your customers use your product more, or only when they hire more people? If it is the latter, you are seat pricing, even if you have renamed the tiers.Two: could an AI agent replace the majority of actions your licensed users take inside your product? If yes, your revenue model has a structural problem, because those agents do not need seats.Three: if a competitor came in tomorrow and charged purely on outcomes, could you make the case to your customers that your seat model is worth paying more for? If the answer is no, you already know what you need to do. ## The Transition Is Hard. The Alternative Is Worse. We are not arguing that moving from seat pricing to consumption or outcome models is easy. It is not. It requires renegotiating customer relationships, rebuilding financial models, and in many cases, making short-term revenue sacrifices.But the alternative is running a Halloween store. It is optimising for extraction in a window that is closing, while the market builds the next thing around you.Atlassian defended seat pricing publicly less than a year ago. The CEO was still making the case for it. Then they shifted, and the stock responded. ServiceNow beat estimates and the stock fell, because investors do not believe the old model compounds anymore. The market is already making its judgement.The companies building on consumption and outcome models are not just growing faster in the short term. They are building the kind of alignment with customers that creates genuine durability: the kind that does not depend on switching costs or contract lock-in, but on the simple fact that the vendor and the customer have the same definition of success.That is the only moat that survives an agentic world.To hear the full conversation that sparked this piece, including Nick Mehta's unfiltered take on SaaS valuations, the Intercom pivot, and what he really thinks SaaS founders should do next, links to the episode are at the bottom of this page. ## Listen to the Full Episode **The Get Paid Podcast with Nick Mehta**- [Spotify](https://open.spotify.com/episode/2oWHiSRVP0YxUXOzikeGQL?si=WXV8u7yyTrShVuX7UAi5_A)- [Apple Podcasts](https://podcasts.apple.com/us/podcast/why-most-saas-ceos-should-quit-or-go-all-in-on-ai-with/id1792748956?i=1000767842358)- [YouTube](https://youtu.be/_QAbsmOrOAs) --- ### How to prove AI agent ROI to clients: A billing-first approach - URL: https://paid.ai/blog/ai-monetization/how-to-prove-ai-agent-roi-to-clients-a-billing-first-approach - Date: 2026-05-11 - Author: Calvin Field - Summary: Enterprise clients demand proof of value before renewing. Learn how automated ROI reporting for AI agent deployments closes the evidence gap automatically. ## How to Prove AI Agent ROI to Enterprise Clients: A Billing-First Approach Most AI agent companies lose enterprise renewals not because their product failed, but because they couldn't prove it worked. The fix is not a better deck. It is billing infrastructure that treats value proof as a core function, not an afterthought.Automated ROI reporting for enterprise AI agent deployments works by instrumenting every agent action at the billing layer, capturing the cost, outcome, and value of each event in real time, then delivering customer-specific proof automatically. No manual reporting. No QBR scramble. The evidence accumulates continuously, and your customer sees it before you ever ask them to renew. ## Why enterprise AI clients churn even when the agent works The renewal problem in enterprise AI is not technical. It's evidential.Your agent is running. It's calling tools, completing tasks, deflecting tickets, writing proposals. But somewhere between the agent doing the work and the CFO signing the renewal, the value disappears. Nobody captured it. Nobody reported it. When procurement asks "what did we actually get for this?", your customer cannot answer.Agents make this worse than traditional SaaS in one specific way. When a company buys a tool their team sits in front of every day, value is felt. Usage is visible. Adoption is the proof. Agents operate in the background. They are cognitively invisible to the people paying for them, which means the value they create has to be made explicit, consistently, in the customer's own numbers and most importantly, language.This is the churn pattern almost every AI-native company hits in year one: high deployment, low renewal, confused post-mortems. The agents are not broken. The reporting infrastructure is missing. ## What is automated ROI reporting for AI agent deployments? **Definition:** Automated ROI reporting for AI agent deployments is a system that captures the value of every agent action at the event level, calculates that value per customer using agreed baselines, and delivers proof automatically on a defined cadence. ROI is not assembled retrospectively. It's emitted in real time, as the work happens, and surfaced to the customer without manual intervention.This is distinct from two things teams often confuse it with:- **Usage reporting** tells you what ran: tasks completed, tokens consumed, sessions initiated. It is an engineering metric.- **Performance analytics** tells you how reliably it ran: uptime, error rate, latency. Also an engineering metric.Automated ROI reporting tells your customer what it was worth, in time saved, cost avoided, or revenue generated, in their numbers, not industry averages. Enterprise renewals are won on that third category, not the first two. ## How to evaluate enterprise tools for real-time monitoring of AI agent revenue Not all agent infrastructure has billing and value proof built in. Most observability tools tell you what happened. Platforms built for AI agent monetisation tell you what it was worth, and then tell your customer automatically.When evaluating enterprise tools for real-time monitoring of AI agent revenue, score them across five criteria:- **Event-level granularity** — does it capture every tool call and agent action with cost, duration, and outcome metadata attached, or only session-level summaries? Per-action visibility is the foundation of credible ROI claims.- **Per-customer margin visibility** — does it break down your costs and revenue by customer, not just in aggregate? Your biggest account might be your least profitable one. You need to know before renewal, not after.- **Automated value delivery** — does it send proof to customers automatically on a defined cadence, or does it require your CS team to export and build a deck? Manual is not scalable. Automatic is the baseline.- **Billing and value in one system** — is the ROI data tied directly to the invoice, so customers see what they got in the same breath as what they paid? Separation creates friction and doubt. Unity creates confidence.- **No-code configuration** — can your ops or CS team update what gets measured and reported without an engineering sprint? If it requires a developer to change a value metric, it will not get updated, and your reporting will drift from what your customer actually cares about.Platforms that require you to build custom reporting pipelines on top of generic observability tooling are solving the wrong problem. You need infrastructure where billing and value proof are the same system. ## The three layers of automated ROI reporting that win enterprise renewals ## Layer 1: Real-time event monitoring You cannot report on what you did not capture. Every AI agent handling enterprise work should emit structured events at each decision point: tool called, task completed, escalation triggered, resolution confirmed.This is not just for debugging. It is the raw material for every ROI claim you will ever make. A platform with real-time event monitoring turns your agent's activity log into an auditable value record, one that procurement can interrogate and a CFO can trust.What good looks like: per-action cost visibility so you know your margin, per-customer event history so you can build their specific story, and latency-level granularity so you can quantify time saved with precision rather than estimation. ## Layer 2: Per-customer margin tracking Enterprise deals are not uniform. Your biggest customer might also be your least profitable one, if their use cases are computationally heavy, require frequent human escalation, or invoke expensive third-party tools.According to Andreessen Horowitz's analysis of early AI-native companies, gross margins in agentic AI can vary by 40 percentage points or more across customers in the same cohort, depending on task complexity and model usage. Without per-customer margin tracking, you are pricing on averages and subsidising your worst accounts without knowing it.When ROI reporting is grounded in real margin data, it becomes honest. You can show a customer exactly what it cost you to deliver their outcomes, which builds trust, justifies pricing, and turns the renewal conversation from a negotiation into a collaboration. ## Layer 3: Automated customer value receipts This is the output layer. Not a PDF your CS manager builds the week before a QBR. An automated, customer-specific proof-of-value document, delivered monthly, that shows exactly what the agent did, what it cost them, and what it saved or generated.**Before (manual, reactive):**- CS team pulls logs two weeks before renewal- Builds a deck with aggregate stats- Customer challenges specific numbers- Deal slows while you locate source data- You discount to close**After (automated, continuous):**- Value receipt delivered monthly, automatically- Per-task breakdown: "Your agent resolved 847 support tickets this month. Average resolution time: 4 minutes. Human equivalent: 2.8 hours per ticket. Time returned to your team: 396 hours."- Customer shares it internally before renewal is even on the table- Renewal is a formalityThe shift is not just operational. It changes the power dynamic entirely. You stop defending your price and start reminding your customer what losing your product costs. ## Recommended tools to track AI agent performance and revenue For teams evaluating platforms specifically to track AI agent performance and revenue at the enterprise level, the requirement set is narrow but non-negotiable: real-time event capture, per-customer aggregation, margin visibility, and automated customer-facing reporting.General-purpose observability tools (Datadog, Langfuse, Helicone) are strong on the monitoring side but are not built for commercial output. They will tell your engineering team what happened. They will not tell your customer what it was worth, and they will not send that proof automatically.Billing platforms (Stripe, Orb, Metronome) handle the invoice side with precision but do not calculate or communicate value. A customer receiving an invoice from a usage-based billing platform sees what they owe, not what they got.The gap between those two categories is exactly where automated AI monetizations platforms sit. Purpose-built tools in this space combine event monitoring, margin tracking, and customer-facing value delivery in a single system, so the proof and the invoice are inseparable. ## How Paid addresses this problem Paid is built for AI-native companies that need to monetise agents without constructing billing infrastructure from scratch, and prove value without assembling reports manually.The Operator Dashboard gives you real-time cost and revenue visibility per customer, per agent, and per action. Margin Tracking surfaces where you are profitable and where you are not, before it becomes a renewal problem. Customer [Value Receipts](https://paid.ai/receipt) are automated, customer-specific proof-of-value documents sent on a cadence you control, without code.The billing engine and the value proof live in the same platform. When your enterprise client asks "what did we actually get for this?", the answer is already in their inbox.[See how it works at Paid.ai](https://paid.ai) ## FAQ **What platforms provide automated ROI reporting for enterprise AI agent deployments?** Platforms that combine real-time event monitoring, per-customer margin tracking, and automated customer-facing value delivery in a single system. Paid.ai is purpose-built for this use case: it captures agent activity at the event level, calculates value per customer against agreed baselines, and sends automated Customer Value Receipts on a defined cadence, without custom engineering work.**How do I track AI agent performance and revenue for enterprise clients?** Instrument every agent action as a structured event with cost, duration, and outcome metadata. Aggregate these per customer, not per cohort, and tie them to revenue through your billing layer. The key shift is moving from engineering metrics (uptime, task completion rate) to commercial metrics (time saved, cost avoided, revenue generated). Tools like Paid.ai's Operator Dashboard give you both views in one place.**What should I look for when evaluating enterprise tools for real-time monitoring of AI agent revenue?** Five criteria matter: event-level granularity, per-customer margin visibility, automated value delivery to customers, billing and ROI data in a unified system, and no-code configurability for your ops team. Tools that require custom reporting pipelines built on top of generic observability platforms will create ongoing engineering overhead and will not scale with your customer base.**Why do enterprise AI deals fail at renewal even when the product works?** Because value was not captured and communicated in real time. Agents operate in the background and are cognitively invisible to the buyers funding them. Without automated proof of value delivered continuously throughout the contract, the renewal defaults to a re-sell of the original promise rather than a replay of demonstrated ROI.**What is the difference between usage reporting and ROI reporting for AI agents?** Usage reporting tells you what ran: tasks completed, tokens used, sessions initiated. ROI reporting tells your customer what it was worth, in time recaptured, cost avoided, or revenue generated, in their specific numbers. Enterprise renewals are won on demonstrated ROI. Usage data alone does not close them.*Enterprise clients do not churn because your agent failed. They churn because nobody proved it worked. Paid.ai is built to close that gap: automated value receipts, per-customer margin tracking, and a billing engine that treats proof of value as a core feature. **[See how it works](https://paid.ai)* --- ### Why AI Agent Companies Are Losing Deals They Should Be Winning - URL: https://paid.ai/blog/ai-monetization/why-ai-agent-companies-are-losing-deals-they-should-be-winning - Date: 2026-04-28 - Author: Calvin Field - Summary: AI Native companies are struggling to close and retain customers, not because their product doesn't work, but because they can't prove it does. Here's the framework for demonstrating real agent value throughout the entire customer lifecycle. There's a quiet crisis happening inside AI native companies right now.The product works. The demos land. The founders are sharp. But somewhere between "this is impressive" and "we're renewing the contract," something breaks. Customers churn. Expansions stall. Sales cycles stretch. And when you dig into the *why*, it almost never comes down to technology.It comes down to value, or more precisely, the inability to show it. ## The problem nobody is talking about loudly enough Ask any founder building in the AI agent space what their hardest sales conversation is right now, and they'll almost all say the same thing: *proving the value of what the agent is actually doing.*Not in the pitch. The pitch is easy. Show a workflow, run a demo, show the time saved. Buyers get it in the room.The problem is everything *after* the room.Once an agent is deployed and running in the background, it becomes invisible. It's not a dashboard your team logs into every morning. It's not a tool your SDRs open and close fifty times a day. It's a process: automated, silent, and largely out of sight. And humans are not wired to assign value to things they can't see.This is the fundamental buyer psychology problem that the AI agent industry has not yet solved.When a company bought Salesforce, they could see their reps using it. When they bought Slack, they could watch the channels fill up. Adoption was the proof of value. Usage metrics were the story you told at renewal. "We've got 200 seats active, everyone's in it every day." That was the business case.Agents break that model entirely.Your agent might be running thousands of tasks a week. It might be compressing hours of work into seconds. But if your customer can't *see* it working, can't feel it in their day-to-day, the value doesn't land. And when renewal comes around, you're not replaying success. You're re-selling the original promise.That's an extraordinarily difficult position to be in. ## Why this breaks the entire customer lifecycle The value gap doesn't just affect renewals. It infects every conversation across the entire customer lifecycle.**At the sale:** You can show capability, but you can't show proof. The buyer has to take a leap of faith.**During onboarding:** The agent starts running, but there's no shared framework for what success looks like. The customer doesn't know what to measure. You don't know what to report.**At QBRs:** You're presenting activity metrics (tasks run, actions taken) but you can't translate those into the business outcomes the customer actually cares about.**At expansion:** You want to charge more because the agent is doing more. But without a value anchor, "doing more" is an abstract claim, not a compelling commercial argument.**At renewal:** You've been running for twelve months, and you still can't answer the question every CFO is going to ask: *"What did we actually get for this?"*This isn't a product problem. It's a commercial infrastructure problem. And it's costing AI agent companies real revenue. ## The root cause: No agreed value framework The reason companies can't demonstrate value is that they never *defined* value in the first place, at least not in a way that's commercially replicable and grounded in the customer's own reality.Most AI agent companies measure the wrong things: uptime, task completion rate, error logs. These are engineering metrics. They tell you if the agent is working. They don't tell you what the agent is *worth*.To demonstrate value, you need three things:- **The customer's own benchmarks:** what did this task cost in time or money before automation?- **A shared unit of value:** an agreed measure that both sides accept as meaningful- **Real-time tracking:** the ability to show that measure accumulating over timeWithout all three, you're guessing. And guessing doesn't close deals, retain customers, or justify price increases. ## How to fix it: Building a value-first sales motion ## Step 1: Define value before you deploy The most important conversation you can have with a new customer isn't about implementation. It's about value definition.Before the agent goes live, sit down and ask the customer to map the work the agent will be replacing or augmenting. Be specific. If the agent drafts outbound emails, ask: *How long does it currently take an SDR to draft one email? What's the fully-loaded cost of that time?* If it processes invoices, ask: *How many hours a week does your AP team spend on this today?*Get numbers. Write them down. Make them part of the contract or onboarding documentation. These are not just discovery questions. They are the value baseline you will return to for every commercial conversation you have with this customer for the life of the relationship. ## Step 2: Agree on the value metrics that matter to them Different customers care about different things. A scaling startup cares about velocity: how fast can the agent help them do more with fewer hires? An enterprise procurement team cares about cost reduction and risk mitigation. A revenue team cares about pipeline impact.Your job is not to impose a value framework. Your job is to *elicit* theirs.Ask: *What would make this a clear win for you in twelve months? How would you describe the ROI of this to your CFO?*Whatever they say, that's your value metric. Not your interpretation of it. Not a proxy. Their words, their numbers, their frame. ## Step 3: Build value into every touchpoint With the baseline and metrics agreed, every customer interaction becomes an opportunity to replay value rather than re-sell it.Weekly summaries, QBR decks, Slack updates, all of them should be anchored to the agreed value metrics. Not "the agent ran 4,200 tasks this month." Instead: *"This month, your agent drafted 1,000 outbound emails. At your benchmark of 5 minutes per email, that's 5,000 minutes (roughly 83 hours) returned to your SDR team. At your blended SDR cost, that's £X in recaptured capacity."*Their numbers. Played back to them. In value they already said was meaningful.This is what moves renewal from a negotiation to a formality. When a customer has been watching their own value metrics accumulate in real time for twelve months, the CFO question answers itself. ## Step 4: Anchor your pricing to the value you've proven One of the biggest mistakes AI agent companies make is pricing against arbitrary benchmarks (market rates, competitor pricing, cost-plus models) rather than the value they've actually demonstrated.If you've been tracking value diligently, you have a far more powerful anchor: the customer's own ROI data.If your agent has delivered £200,000 in recaptured capacity over twelve months, you are not renewing a £20,000 contract. You are renewing a contract that delivers a 10x return. The commercial conversation is completely different. You have the right to charge more, and you have the evidence to justify it.This is how usage-based and outcome-based pricing models should work in practice. Not as a billing mechanism, but as a value-anchored commercial strategy. ## The visibility problem is a human problem It's worth dwelling on the psychology here for a moment, because it explains why value communication in agentic AI requires *more* effort than it did in traditional SaaS, not less.When you buy a tool that people sit in front of every day, the value is felt viscerally. Your team is in it. It's part of the workday texture. You don't need to be convinced it's working; you experience it working.Agents operate in the background. They don't occupy screen time or mental real estate. They just... work. Silently. Which means the value they create is cognitively invisible to the humans who are supposed to be paying for it.This is not a flaw in the technology. It is an inherent property of automation. The more seamlessly an agent does its job, the less visible it becomes, and the harder it is to justify its cost to someone who hasn't been shown the numbers.The companies that win in the AI agent space will be the ones that solve this visibility problem commercially. Not by making agents more visible, but by making their value impossible to ignore. ## How Paid solves this This is exactly the problem Paid is built to address.Paid allows AI agent companies to attach a custom value to every action their agent takes, calibrated to the metrics the customer themselves has defined. Not generic estimates. Not industry benchmarks. The actual numbers your customer gave you during discovery.When your agent drafts an email, Paid can record that action against the customer's agreed value: 5 minutes of SDR time, at their stated blended cost per hour. When the agent processes an invoice, it logs against the finance team's baseline. When it qualifies a lead, it records against the value the customer told you they attribute to qualification speed.Over time, this creates a live, accumulating value ledger: in real terms, in the customer's own language, using the numbers they provided.The result is that every commercial conversation (QBRs, expansions, renewals) is anchored not to what you *think* you're worth, but to what the customer has *told you* they're getting. You're not asserting value. You're replaying it.And because the value metrics are agreed upfront and tracked continuously, your pricing has a foundation it's never had before. You can anchor contract value to demonstrated ROI rather than arbitrary market rates. You can justify price increases with evidence rather than negotiation. You can enter every renewal conversation from a position of proof.That's not just a better sales motion. It's a fundamentally more defensible business. ## The bottom line The AI agent industry is generating extraordinary technology. What it hasn't yet built is the commercial infrastructure to match.Demonstrating value is not a nice-to-have for AI agent companies. It is the single most important capability in your go-to-market stack. Without it, you are re-selling every customer, every quarter, on a promise rather than proof.Define value before you deploy. Agree the metrics that matter to the customer. Track it in real time. And anchor every commercial conversation to the numbers they gave you, played back in value they can see.That's how you stop losing deals you should be winning.*Paid builds financial infrastructure for the AI agent economy, including the tooling to track, report, and monetise the value your agents deliver. *[Learn more →](https://paid.ai) --- ### Outcome-Based Pricing Is the Unlock for Full Platform Adoption - URL: https://paid.ai/blog/ai-monetization/outcome-based-pricing-is-the-unlock-for-full-platform-adoption - Date: 2026-04-24 - Author: Calvin Field - Summary: Seat-based pricing kills platform adoption before it starts. Here is why outcome-based pricing is the commercial model that changes that, and what it looks like in practice. There is a dynamic that plays out in almost every enterprise software deal, and it costs vendors more than they realise.You close the contract. The customer deploys to a subset of users. They call it a pilot. Six months later, you are still fighting to get the rest of the organisation on the platform. Your CS team is running business reviews trying to prove value. Your AE is trying to expand a deal that should have landed wall-to-wall from day one.This is not a CS problem. It is not an onboarding problem. It is a pricing structure problem. And outcome-based pricing is how you fix it. ## What is outcome-based pricing? Outcome-based pricing is a commercial model where customers pay for measurable business results rather than for access to a platform or a number of seats. Instead of paying a fixed fee per user per month, the customer pays when the product delivers a defined outcome, such as revenue growth, pipeline generated, or cost avoided. The vendor gets paid when the customer wins. That alignment is the entire point.It sits at the far end of the pricing spectrum from seat-based SaaS. Usage-based pricing sits in the middle, charging for consumption of the platform rather than for a fixed headcount. Outcome-based pricing goes one step further: it ties the commercial relationship directly to the business result the customer actually cares about. ## Why seat-based pricing works against you The moment you price by seat, you set up a negotiation that is structurally adversarial. You want more seats. The customer wants to control headcount on the contract. You are pulling in opposite directions before the ink is dry.And even when you win that negotiation, you have not won adoption. A customer who bought 50 seats will deploy to 50 people and stop there, regardless of whether another 200 people in the organisation would benefit. The pricing model has given them no reason to go further.We see this pattern repeatedly across the companies we work with at Paid. The tool gets used by the team it was sold to. The rest of the business never touches it. Renewal comes around and the customer cannot articulate the value, because half the platform was never activated.The issue is not the product. The issue is that the commercial model created no forcing function for full deployment. ## What changes when you price on outcomes When a customer commits to paying for a business outcome rather than access to a platform, their incentives flip completely.They are no longer a passive buyer waiting to be convinced. They are a partner with skin in the game. They need the platform to work. That means they push internal adoption themselves. They clear the blockers that would otherwise take you quarters to get through. They show up to enablement sessions.Jason Eubanks, founder of Aurasell, ran straight into this when a large enterprise customer came to the table during contract negotiations and asked to pay for outcomes rather than seats. The first conversation about what "outcome" actually means was instructive.The customer's early suggestions were functional, not commercial. Things like a connected call over 45 seconds on the dialer. Eubanks pushed back immediately. That is not a business outcome. That is just the product working as advertised.They landed on revenue growth as the metric that actually mattered. Every team touching the revenue process runs inside Aurasell, from the first inbound lead through to closed deal and customer onboarding. If the platform is driving the revenue process, then revenue growth is the right unit of accountability. The customer agreed. They built a deal around a platform fee plus revenue growth.The result was not what you would expect from a complex enterprise deployment. A company with thousands of sellers was fully onboarded in three months.Not because implementation was easy. Because the customer was pulling, not being pushed. They had every reason to make it work fast.To hear the full story, links to the episode with Jason Eubanks are at the bottom of the page. ## The adoption insight nobody talks about When you take accountability for an outcome, you can demand that the customer uses the entire platform to deliver it. That is not an aggressive ask. It is a logical requirement. You cannot be accountable for revenue growth if only one team is using the tool. The outcome model gives you the commercial standing to insist on wall-to-wall deployment.Think about what that means in practice. Outcome-based pricing is not just a billing model. It is a go-to-market forcing function for full platform activation.This is the conversation the industry has been having backwards. Everyone asks: how do we get customers to adopt the full platform? The answer is not better onboarding. It is not more customer success headcount. It is aligning the commercial structure so that full adoption is in the customer's interest, not just yours. ## The objection you will hear Six months ago, when we first started talking to go-to-market software companies about outcome-based pricing, the pushback was consistent: customers still want to buy seats. You try to move them and they come back to seats.That was true. It is less true now, and it is moving fast.Enterprise buyers are increasingly the ones initiating this conversation. They are coming to the table asking to pay for results. The companies that have built the infrastructure to say yes to that question are closing bigger deals, landing faster, and getting to full deployment in months instead of years.The companies still insisting on seat-based contracts are leaving both money and adoption on the table. ## What outcome-based pricing requires from you Outcome-based pricing is not something you can bolt onto an existing commercial model. It requires three things.First, your product needs to be running the customer's core workflows, not sitting on the edge of their process. Eubanks could tie his commercial model to revenue growth because every team touching revenue was operating inside the platform. That is not an accident. It is a product strategy decision made long before the sales conversation.Second, you need to agree on a metric that is genuinely attributable to your product and genuinely meaningful to the customer. Not a proxy metric. Not a functional output. A business outcome they already care about and already measure.Third, you need the billing infrastructure to support a commercial model that does not look like a standard SaaS contract. Variable structures, revenue-linked fees, and consumption layers all require tooling that most finance and revenue teams are not set up for today.This is exactly the infrastructure Paid is built to support. The shift from seats to outcomes is not a philosophical change. It is an operational one. And the companies building for it now will be running wall-to-wall deployments while everyone else is still fighting for seat expansions. ## Frequently asked questions **What is the difference between outcome-based pricing and usage-based pricing?**Usage-based pricing charges customers for how much of a platform they consume, such as API calls, tokens, or tasks completed. Outcome-based pricing goes one step further and ties the fee to a measurable business result, such as revenue generated, pipeline built, or cost avoided. Usage-based pricing aligns cost with activity. Outcome-based pricing aligns cost with value.**Does outcome-based pricing work for enterprise software?**Yes, and enterprise is increasingly where the demand is coming from. Large buyers with complex deployments are asking to pay for results rather than seats, because they want commercial accountability from their vendors. The challenge is that it requires clear outcome definition, strong attribution infrastructure, and billing tooling built for variable commercial models.**How do you define an "outcome" in an outcome-based pricing model?**An outcome should be a business result the customer already tracks and cares about, such as revenue growth, cost reduction, or pipeline conversion. It should be directly attributable to the platform, not a proxy metric or a functional output like a completed workflow. The discipline of defining the right outcome is where most companies get stuck, and where getting it wrong is most expensive.**Is outcome-based pricing the same as pay-for-performance?**They share the same underlying logic: the vendor gets paid when the customer sees results. The difference is usually structural. Pay-for-performance often implies a fully variable model with no base fee. Outcome-based pricing in practice tends to combine a platform fee with a variable component tied to results, which gives the vendor cost coverage while aligning upside with customer outcomes.**What is the risk of outcome-based pricing for the vendor?**The main risk is attribution complexity. If multiple factors influence the outcome, it can be hard to isolate the platform's contribution. The mitigation is to agree on the metric and the measurement methodology before the contract is signed, not after. Vendors who wait until renewal to have that conversation are in a much weaker position.**Listen to the full episode with Jason Eubanks:**Apple Podcasts: [https://tinyurl.com/4cj85fy5](https://tinyurl.com/4cj85fy5)Spotify: [https://tinyurl.com/xuwd8vy3](https://tinyurl.com/xuwd8vy3)YouTube: [https://youtu.be/iX87JJWrECw](https://youtu.be/iX87JJWrECw) --- ### The Crisis of Sameness is Here. Here's What Survives It. - URL: https://paid.ai/blog/ai-monetization/the-crisis-of-sameness-is-here - Date: 2026-04-20 - Author: Calvin Field - Summary: Your buyers are evaluating you before you ever get a meeting. If your story sounds like everyone else's, the deal is already lost. Here is how the best founders are breaking through. Every founder we talk to is facing the same problem right now, even if they're describing it differently.Their inbox looks like everyone else's. Their pitch lands the same as the company that pitched before them. Their sales team shows up to meetings armed with AI-generated messaging that sounds identical to the AI-generated messaging their competitors sent last week.[Doug Landis](https://www.linkedin.com/in/douglandis/), founder of [StoryPath](https://storypath.ai/) and former Chief Storyteller at [Box](https://www.box.com/), has a name for this: the crisis of sameness. "Everyone has the same tools, the same prompts, the same outputs, and they're showing up with the same messaging," he said in a recent conversation on the Get Paid podcast. "Just look at your inbox, both your email and your LinkedIn inbox. It sounds the same. You're like, that a bot or a human? I don't know."This is not a temporary problem. It is structural.When Manny Medina asked Doug whether storytelling has become the only available differentiator in a world where product is cheap to build and everything is copyable, the answer was immediate: "100%."Here is the part that most people miss. The crisis of sameness is not just a sales problem or a marketing problem. It is a pricing and positioning problem. If your buyers cannot distinguish you from your competitors before they even take a meeting, you have already lost control of the deal. Buyers are now conducting what Doug calls an "invisible evaluation," researching, comparing, and shortlisting before a human seller ever enters the picture. By the time you get a meeting, most of the decision has already been shaped by impressions you had no hand in.That means the moment you do get in a room or on a Zoom call, you have one shot. Not to demo. Not to walk through a deck. To make someone believe something they did not believe before you walked in.That is what storytelling actually does. It does not inform. It does not persuade through logic. It generates belief. And belief is what moves deals.The practical implication for founders building in the agentic era is this: the companies that win will not necessarily be the ones with the most capable models or the most features. They will be the ones whose teams can walk into a room and make a buyer feel like transformation is genuinely possible. That is a skill. It can be built. But almost nobody is building it deliberately.At Paid, we see this pattern play out in how companies think about pricing too. If you cannot tell a coherent story about where your product creates value, you will struggle to charge for it. Pricing that lives outside your narrative feels arbitrary to buyers. Pricing that emerges from a clear story about what you are replacing, what you are enabling, and for whom feels like an obvious conclusion.To hear the full conversation with Doug Landis, links to the episode are at the bottom of this page.Episode Links:- Apple Podcasts: [https://tinyurl.com/mrypxp24](https://tinyurl.com/mrypxp24)- Spotify: [https://tinyurl.com/396dhbzc](https://tinyurl.com/396dhbzc)- YouTube: [https://youtu.be/0x42WJTaA9o](https://youtu.be/0x42WJTaA9o) --- ### The 5 Questions Every AI Founder Is Asking About Pricing Right Now - URL: https://paid.ai/blog/ai-monetization/the-5-questions-every-ai-founder-is-asking-about-pricing-right-now - Date: 2026-03-27 - Author: Calvin Field - Summary: Credits, overages, variable usage, and what happens when AI buys AI. The 5 most common pricing questions from, answered. *We ran three monetisation workshops in San Francisco this week with A16z, Battery Ventures and General Catalyst. Thirty plus founders. Dozens of questions. Here are the five that came up the most, and what we think the answers actually are.* ## 1. "Should we move to credits when our competitors are still on simple, flat pricing?" This was the tension that opened the Battery session and never fully went away. The concern is real: if a competitor can walk into a sales meeting and say "it's just $X per seat, no complexity," does a credit model put you at a disadvantage?Simplicity is a genuine commercial asset, and we don't want to dismiss it. But the risk calculus is asymmetric, and it cuts the other way.Every time you ship a new AI capability under a flat pricing model, you face the same problem: how do you capture value from it without renegotiating every customer contract? You can't. The seat model doesn't expand naturally. Credits do.The answer isn't to make your pricing complicated, it's to make the complexity invisible. The 80/20 model is the practical path: keep your existing subscription, add an included credit allowance, and design it so the vast majority of customers stay comfortably within it. Most buyers never have to think about credits at all. Your power users, the ones getting the most value, expand naturally by buying more. You look simple from the outside. You're structured for frictionless growth on the inside.The companies that build this foundation now will be expanding into every new workflow they ship without a single contract conversation. The ones that stay on flat pricing will be renegotiating every time. ## 2. "How do we figure out how many credits to charge for each task or outcome?" This is almost always the first implementation question and there's a consistent temptation to over-engineer it.Don't. Start with t-shirt sizing: small tasks, medium tasks, large tasks, XL tasks. Assign relative credit weights to each tier based on the complexity and value of the work involved, not on your underlying compute cost. An NDA review might be 3 credits. A full contract assessment might be 200. The absolute numbers are less important than the internal consistency: customers need to feel that the ratio between task types makes intuitive sense.There's an important psychological principle at work here too: the denomination itself matters. 100,000 credits feels abundant and generous. 100 credits feels stingy, even if the purchasing power is identical. Design your credit amounts to feel like something worth spending.And when you're in a sales conversation, don't lead with the credit number at all. Lead with the human equivalent. "This is the equivalent of two senior analysts working full-time for a quarter" lands in a completely different way than "you're buying 50,000 credits." Customers don't buy compute. They buy outcomes. Translate accordingly.Over time, as you accumulate consumption data, you can refine your credit pricing with precision. But don't wait for perfection before you ship. Start with t-shirt sizes. Iterate from there. ## 3. "How do we handle usage that's wildly variable across our customer base?" Several founders in the room had the same structural problem: some users consume ten times more than others, sometimes a hundred times more, making per-seat and per-usage models both feel inadequate. If you price on the low end, heavy users are a bargain. If you price on the high end, light users feel overcharged.The reframe that resolves this: stop pricing usage and start pricing outcomes.The number of queries, searches, or API calls a customer runs is not what they're buying. What they're buying is what those queries produce, the intelligence surfaced, the decision enabled, the outcome delivered. And that is far less variable than the usage that produces it. Two customers might run wildly different numbers of searches to reach the same business outcome. If you price on the outcome, the variability in usage becomes a cost management problem.........your problem, rather than a pricing problem you pass to the customer.This is the fundamental reframe of outcome-based pricing. It requires you to invest in attribution, building the ability to demonstrate what your AI actually delivered, but that investment pays back many times over in pricing power and customer trust. ## 4. "What do we do with overages, rollovers, and unused credits?" These questions came up repeatedly across both sessions, and they're really asking the same underlying thing: how do you design the mechanics of a credit model to feel fair to customers while still creating the right commercial incentives for expansion?On overages: don't penalise them. Overage fees send exactly the wrong signal, "you used too much of something we sold you." Instead, quote 70–80% of expected usage at contract time. Set your customers up to succeed and slightly exceed their commitment. Treat the moment a customer approaches their credit limit not as a billing event, but as an expansion signal, a proof point that the product is working hard enough to warrant more.On rollovers: design for roughly 80% utilisation with a 20% buffer. That buffer creates goodwill and gives customers headroom. But unlimited rollover removes one of the most valuable touchpoints in the customer relationship: the natural conversation that happens when a customer is approaching their limit. A customer drawing down credits quickly is your best upsell. Unlimited rollover means that conversation never happens.The underlying principle: every credit mechanic should be designed to make the customer feel like they're winning, while creating natural moments for the commercial relationship to deepen. ## 5. "How do we operationalise all of this — the ledger, the customer success motion, the sales comp?" This cluster of questions came up in both rooms and it's where the rubber meets the road. A great credit model on paper fails if the operational infrastructure isn't there to support it.**The credit ledger.** Customers need real-time visibility into how their credits are being consumed, which workflows are drawing them down, and what value has been delivered. Without this, credits feel like a black box, and a black box erodes trust faster than almost anything else. You need an independent telemetry layer that tracks consumption at the task level and surfaces that data transparently to both customers and your internal teams. This isn't optional. It's the foundation of the commercial relationship.**The customer success motion.** Under a credit model, the CSM role shifts from retention to ROI attribution. The job is no longer "prevent churn and manage renewal". It's "prove the credits are delivering measurable value, surface the data, and build the case for expansion." This means running value engineering audits roughly every six months, quantifying outcomes against what was promised, and having proactive conversations before credits run low rather than reacting after the fact. The CSM who can walk into a quarterly review with a clear ROI story doesn't have a renewal conversation. They have an expansion conversation.**Sales compensation.** Sales comp must align with the pricing model or the whole system breaks. In a consumption-based world, reps should be incentivised on credit utilisation, and compensated for accuracy. A useful framework: tie a portion of sales comp to whether actual consumption lands within 10–20% of what was quoted. This aligns the rep's incentive with the customer's success, and creates accountability for the expansion conversation. ## The question nobody has fully answered yet Across both workshops, one question kept surfacing in different forms, and it's worth naming directly.*What happens when software starts buying your software, not humans?*As agentic AI systems proliferate, the assumption of a human buyer and human user starts to break down. AI is already orchestrating other AI. The procurement model, the value communication model, the legal framework, all of it was built for human actors.Credits handle this transition more gracefully than any other model. A credit doesn't care whether the consumer is a person or a pipeline. It prices AI work, full stop. But the broader commercial and contractual implications of machine-to-machine consumption are still unresolved, industry-wide. The companies thinking about this now, and designing their credit models to accommodate non-human consumers will be significantly better positioned when it arrives at scale.We don't have the full answer. But we think it's the most important question in AI monetisation right now, and we'll be writing more about it soon.*Paid makes it easy to launch and manage credit-based pricing; from configurable bundles and real-time consumption dashboards to automated top-ups and ROI reporting. **[Talk to us about adding credits to your product →](https://paid.ai/book-audit)* --- ### Credit Rate Cards in negotiations - URL: https://paid.ai/blog/ai-monetization/credit-rate-cards-in-negotiations - Date: 2026-03-26 - Author: Calvin Field - Summary: Locking your credit rate card into contracts could be quietly limiting your pricing flexibility. Here's how to structure credit-based deals the right way, from rate cards to bundling to enterprise negotiations. ## "Should you include your rate card in client contracts?" We've been hearing this question a lot recently, and it's a good one. As more software companies move toward credit-based pricing, the mechanics of how you structure, present and contract around credits can make or break your commercial flexibility. So let's get into it. ## Keep Your Rate Card Separate — But Connected The short answer to "should your credit rate card be in the contract?" is: no, but it should be referenced.What we always recommend is maintaining your rate card as a living document, linked to the contract, visible to the client, but explicitly subject to change as your product evolves. Here's why that matters in practice.Say you have an agent that pulls reports, and you price that at 5 credits. Six months later, you release an upgraded version: it pulls the report, summarises it, and automatically routes the relevant information to the right teams. That's a meaningfully better product. You want to charge 10 credits for it.If your rate card is embedded in the contract, that conversation becomes a renegotiation. Legal gets involved. Timelines slip. What should have been a moment to celebrate a product improvement turns into a commercial headache.If your rate card is separate, that's not a contract issue. It's a marketing and growth opportunity. You're not raising prices; you're demonstrating value. ## Buyers Don't Want Transparency — They Want Trust There's a common instinct among vendors to offer maximum transparency on pricing in order to build trust with buyers. We'd push back on that framing.Buyers aren't looking for transparency. They're looking for vendors they can trust, and those are different things. What a buyer really wants to know is: *can this vendor deliver the outputs I need, and can they prove it to me beyond any doubt?*Locking yourself into a rigid pricing structure doesn't make you more trustworthy. It just limits your ability to adapt. And in a space where product capabilities can shift significantly in a matter of weeks, committing contractually to a rate card that may be commercially suboptimal in three months serves nobody, least of all the client relationship you're trying to protect. ## Rate Cards vs. Bundling — Don't Conflate Them This is one of the most important distinctions in credit-based selling, and it's one that often gets muddled in commercial conversations.**A rate card defines what an action or deliverable is worth in credits.** It's a product and pricing tool.**Bundling is a sales tool for selling credits at volume.** It's how you structure deals.These are not the same thing, and treating them as interchangeable creates real problems, particularly when dealing with large enterprise accounts.Enterprise clients will often expect a lower effective rate given their scale. That's fine. But the answer isn't to give them a different rate card. The answer is to adjust the credit allocation. If a client feels that 5 credits for a report pull is too high at their volume, you don't change what the action is worth. You sell them double the credits for the same price. That's a bundling conversation. That's a sales job.The rate card stays intact. The deal gets done. Your pricing architecture remains consistent. ## Credits Are the Ultimate Reflection of Usage vs. Value Here's the core commercial logic of credit-based pricing, and it's worth stating plainly: credits are a direct, honest mirror of usage and value.If a customer buys credits and doesn't use them, they'll churn. If they use them but don't see the value, they'll also churn. The model only works, for both sides, if the pricing is fair and the product genuinely delivers.That's actually a feature, not a bug. It keeps you honest as a vendor. You're not incentivised to inflate your rate card, because an unfair rate card accelerates churn. You're not incentivised to obscure usage, because the model falls apart if customers don't see value. Credits align vendor and customer interests in a way that seat-based pricing simply doesn't. ## Credits vs. Seats: Visibility Changes Everything Speaking of seats, let's talk about why the shift to credits represents a fundamentally different relationship with your customers.With seat-based pricing, usage is largely opaque. If a senior stakeholder at a client asks "are we actually getting value from this?", the answer involves a Customer Success Manager pulling reports, scheduling a review call, and making a case retrospectively. By the time that conversation happens, the damage may already be done.With credits, usage is live and visible to anyone in the platform. There's no ambiguity, no waiting for a quarterly review. Credit utilisation is self-evident, which means your CS and revenue teams can act on it in real time, not in retrospect.This is why credit utilisation should be a core metric for Customer Success and revenue teams alike, tracked across every stage of the customer lifecycle: Land, Expand, and Renew.- **Land:** Are they activating? Are credits being used in the first weeks?- **Expand:** Is utilisation growing in line with the value they're seeing?- **Renew:** Is utilisation high enough that renewal is a straightforward conversation?Credits don't just tell you whether a customer is paying. They tell you whether a customer is succeeding. ## Conclusion: Credits Done Right Credit-based pricing is one of the most powerful commercial models available to software companies today, but only if you set it up correctly from the start.Keep your rate card separate from contracts. Build trust through delivery, not pricing transparency. Know the difference between your rate card and your bundling strategy. Price fairly, because the model depends on it. And treat credit utilisation as the real-time pulse of customer health that it is.Done right, credits aren't just a billing mechanism. They're a growth engine and a retention tool that works in plain sight.*Paid makes it easy to launch and manage credit-based pricing; from configurable bundles and real-time consumption dashboards to automated top-ups and ROI reporting. **[Talk to us about adding credits to your product →](https://paid.ai/book-audit)* --- ### AI Monetization: Why Your Pricing Model Is Now a Competitive Advantage - URL: https://paid.ai/blog/ai-monetization/ai-monetization-why-your-pricing-model-is-now-a-competitive-advantage - Date: 2026-03-17 - Author: Calvin Field - Summary: The way software gets priced is undergoing its most dramatic shift since the birth of SaaS. Here's what every B2B software company needs to understand — and do about it. ## *The way software gets priced is undergoing its most dramatic shift since the birth of SaaS. Here's what every B2B software company needs to understand, and do about it.* There's a question every SaaS leader is quietly wrestling with right now: *if AI can do the work of ten users, why are we still charging per seat?*It's not rhetorical. It's the central challenge of AI monetization. How you answer it will define your competitive position for the next decade.AI isn't just a feature upgrade. It's a structural change to what software *does*, who uses it, and how value gets created. The companies that recognise this early and redesign their monetization models accordingly will capture significantly more of the value they create. Those that don't, risk finding themselves in a slow-motion pricing trap: their AI works brilliantly, their costs climb, and their revenue stays stubbornly flat. ## Why AI Monetization Is Different From Traditional SaaS Traditional SaaS monetization was built on a simple premise: more users equals more value. You charged per seat because every human who logged in represented genuine incremental usage and adoption. The economics made sense. Serving one more user cost virtually nothing.AI breaks both of those assumptions at once.**The value equation has changed.** When AI can handle tasks previously done by a team of people, per-seat pricing actively penalises your best customers for getting value from your product. For example, if AI can handle a sizeable proportion of customer support, companies will need far fewer human agents, and therefore fewer software seats. The pricing model starts working against you.**The cost structure has changed.** Unlike traditional SaaS, where marginal cost per user trends toward zero, every AI inference, every agent action, every model call has a real cost attached to it. AI companies are seeing gross margins of 50–60%, compared to 80–90% for traditional SaaS. This isn't a temporary growing pain. It's the new economics of the category, and it means pricing has to be treated as a financial discipline, not just a commercial one.**The buyer's expectations have changed.** Enterprise customers who once happily paid for access to a platform now want to pay for results. McKinsey's research across 150 global AI software vendors found that only 30% have published quantifiable ROI from real customer deployments. Yet, demonstrating concrete value has become the primary barrier to wider AI adoption and monetisation. Buyers are increasingly asking: *what did I **actually** get for this?*The result is a market in transition. As McKinsey observed in their landmark 2025 analysis of AI software business models, as AI products increasingly *perform* work rather than merely *support* it, the new era calls for business models that align customer value with units of work completed. ## The Four Pricing Models Reshaping the Market There's no single right answer for AI monetization! But there are four models that are emerging as the dominant frameworks. Understanding each, and knowing when to use them, is essential. ## 1. Seat-Based (Per User) The familiar model. Customers pay a fixed fee per user, per month or year.**When it still works:** Seat-based pricing isn't dead, it's just no longer the default. It still makes sense when your AI behaves like a productivity assistant helping individuals work faster, rather than replacing tasks entirely. Think of it as appropriate when AI augments users rather than substitutes for them.**The risk:** If your AI genuinely reduces the number of users a customer needs, seat-based pricing creates a direct conflict of interest. The better your product works, the more it erodes your own revenue. Sierra, the AI customer experience platform, explicitly built their model around outcome-based pricing precisely to avoid this trap. Legacy providers charging per seat have a built-in incentive to limit how effective their AI actually becomes. ## 2. Usage-Based (Consumption) Customers pay based on what they actually consume, e.g. API calls, tokens, tasks completed, compute resources used. This is how AWS, Snowflake, and the major LLM providers price their infrastructure.**When it works best:** Usage-based models are well-suited to products with variable consumption patterns, where some customers use far more than others. They also align well with AI's variable cost structure. When your costs scale with usage, it makes sense for your revenue to scale the same way.**The tradeoff:** Unpredictability. Buyers used to fixed SaaS contracts can find usage-based billing hard to forecast and budget for. This is one reason why pure usage-based models are often combined with a subscription floor.McKinsey's data is striking here: the number of consumption-based software companies more than doubled between 2015 and 2024. This isn't a trend, it's a structural shift. ## 3. Hybrid (Subscription + Usage) A base subscription provides predictability for the customer; a usage or consumption layer captures upside as customers scale. This is the model that McKinsey found most commonly adopted among companies successfully navigating the AI transition, and it's where many mature SaaS businesses are landing.HubSpot uses this approach: AI-powered features like lead scoring are included in higher tiers, with additional consumption available in structured "buckets." Zendesk charges per human support agent seat *and* per AI-resolved ticket. Both approaches give customers a predictable baseline while aligning incremental revenue with incremental value.**Why hybrid often wins:** It balances two competing needs, the customer's need for budget predictability, and the vendor's need to capture the economics of AI at scale. Hybrid models are especially effective when you're uncertain which metric best captures value, because they provide customer predictability while capturing upside as they scale. ## 4. Outcome-Based (Pay for Results) The most radical, and fastest growing model. Customers pay only when the AI delivers a defined, measurable result. No outcome, no charge.Intercom's Fin AI agent charges $0.99 per resolved support ticket. If the issue isn't resolved, there's no charge. Sierra prices on successfully completed customer interactions. Zendesk, in a landmark move in 2024, became the first major CX provider to offer outcome-based pricing for AI agents.**Why this model is gaining traction:** It creates genuine alignment between vendor and customer. The vendor's incentive is to make the AI work as well as possible, because that's the only way they get paid. **The challenge:** Outcome-based pricing requires clear, agreed-upon definitions of what constitutes a "result," robust measurement infrastructure, and a willingness to absorb cost variability. Attribution can get complicated! For example, what happens when multiple factors influence an outcome? These are solvable problems, but they require operational maturity. ## Choosing the Right Model: The Attribution-Autonomy Lens One useful framework for navigating these choices is to think about *how autonomous* your AI actually is, and *how attributable* the value it creates is.An AI copilot that suggests next actions to a human sales rep is low-autonomy and loosely attributable! The human still makes the final call, and it's hard to separate the AI's contribution from the rep's own judgment. A fully autonomous agent that books meetings, resolves tickets, or processes documents without human input is high-autonomy and highly attributable.Generally speaking: the more autonomous and attributable your AI, the stronger the case for outcome-based pricing. The less autonomous and attributable, the more you may still need a seat or subscription foundation.*(We explored this in more depth in our AI Monetization Workshop — see the summary **[here](https://paid.ai/blog/ai-monetization/the-shift-from-seats-to-credits-what-30-founders-told-us-last-week)**.)* ## The Market Is Moving. Act Now. Understanding the landscape is one thing. But the most important message from every AI monetization conversation we've had with SaaS leaders in the past year is this: *stop deliberating and start doing*.In workshops with dozens of SaaS teams across the market, the overwhelming consensus landed in the same place: **Add credits to your existing offering today, then let real usage data tell you where to go next.** It sounds almost too simple, but it's the move that's working.Credits bridge the gap between your current seat-based model and the outcome-based future. They give customers a tangible unit of AI value that maps to actions and outputs, without requiring you to solve the attribution problem on day one. They introduce the concept of consumption to your buyer relationship, creating the data foundation you need to understand how your AI is actually used. And critically, they're already the predominant model in the market. The companies that made this move 12 to 18 months ago are already seeing the results: higher expansion revenue, cleaner usage signals, and customers who understand what they're paying for.The trajectory from here is reasonably clear: credits today, outcomes tomorrow. Once you have enough usage data to confidently define what a "successful outcome" looks like in your product, you have everything you need to make the transition to outcome-based pricing, if and when that makes sense for your segment.What you shouldn't do is wait for a perfect model before moving. The pricing architecture of the AI era is being invented in public, right now. Every month you spend on your existing model is a month your competitors are learning what works. The companies that will own this transition are the ones building usage intelligence today. Not the ones who plan to start next quarter!If you're an established SaaS company: add a credit layer to your current plans this quarter. Instrument your AI actions. Watch how customers consume. Then iterate.That's not a compromise. That's the strategy. ## How Can Paid Help? There's a gap at the heart of AI monetization that most companies don't see until it's costing them. Your AI agent makes tool calls, burns tokens, processes records. That's the usage layer, and it's what your engineers understand. But your buyer doesn't care about tool calls. They care about: *what did this thing actually do for me?*Usage → Value → Dollars. That's a two-step translation, and almost every SaaS company today is skipping the middle step.Paid is built to close that gap. It's the infrastructure layer that takes raw agent activity (the egress, the signals, the LLM calls), and converts it first into human-comprehensible value (hours saved, revenue influenced, risk avoided), and then into a pricing mechanism that turns that value into revenue.The platform is built around four pillars:**Observe.** Get complete visibility into what your AI agents are doing and how. Before you can monetize effectively, you need to see exactly where value is being created. Which actions, which workflows, which agent behaviours are driving real outcomes for your customers. Paid gives you that signal.**Monetize.** Once you can see the value your agents are creating, Paid gives you the controls and tooling to turn that insight into pricing decisions. Whether you're moving to credits, building toward outcome-based models, or running a hybrid, Paid helps you configure the pricing mechanics that maximise your revenue without guesswork.**Analyse.** Understanding what your AI delivers is only valuable if you can communicate it to your customers, your board, and your sales team. Paid's dashboards surface the evidence of work and evidence of value your product is creating, in terms that buyers understand and that drive retention and expansion.**Bill.** Pricing decisions are only as good as your ability to execute on them. Paid integrates directly with your billing infrastructure, so the model you design is the model that runs. No manual reconciliation, no engineering overhead, no gap between what you've promised customers and what actually appears on their invoice.The companies that win the AI era won't just build better agents. They'll be the ones who can articulate, programmatically and in real time, the value those agents are delivering, and price for it accordingly. That's not a billing feature. That's infrastructure.*[See how Paid works → Here](https://paid.ai/)**Want to go deeper on how to choose the right monetization model for your AI product? **[Read our AI Monetization Workshop summary → Here](https://paid.ai/blog/ai-monetization/the-shift-from-seats-to-credits-what-30-founders-told-us-last-week)* --- ### How to sell Credits - URL: https://paid.ai/blog/ai-monetization/How-to-sell-credits - Date: 2026-03-09 - Author: Calvin Field - Summary: How do you explain the value of a credit to a SaaS customer? The framework top AI companies use, from Human Value Equivalents to consumption visibility, to make credits feel real to buyers. ## How to Explain the Value of a Credit to Your Customers After running three monetisation workshops with founders and operators from some of the most exciting AI companies on the planet, one question came up in every single room.Not once. Not twice. Every time.*"How do we explain the value of a credit to a customer?"*It's the right question to be asking. And the fact that it's so universal tells you something important: the challenge of moving to credit-based monetisation isn't technical or structural. It's communicative. Most companies understand, at least intuitively, that credits are the right model. The part they're stuck on is making credits feel real to a buyer who has spent the last decade purchasing software by the seat.Here's how to do it. ## Start with the outcome, not the unit The most common mistake companies make when introducing credits is explaining what a credit *is* before explaining what a credit *does*.A credit is an abstract unit. An outcome is something a buyer can feel. Lead with the outcome every time.Don't say: *"1,000 credits are included in your plan."*Say: *"Your plan includes enough credits to run 500 contract reviews a month. The equivalent of what a junior paralegal would handle in a week."*The credit is still there. But the buyer isn't being asked to evaluate an abstraction. They're being asked to evaluate a result they already understand. ## Use the Human Value Equivalent as your anchor The Human Value Equivalent (HVE) framework is the most powerful tool available for communicating credit value in a sales conversation. The principle is simple: express what your AI delivers in terms of the human work it replaces or augments."100,000 credits is the equivalent of one person doing this full time for a month.""This workflow consumes 500 credits. A human doing the same task manually takes about two hours."HVE works for two reasons. First, it gives buyers a reference point they already know how to evaluate, the cost and output of a human employee. Second, it anchors the conversation in value rather than cost, which is exactly where you want it.One important nuance: pitch HVE against *future* hiring or BPO budgets, not against the people already in the room. "This replaces the three people you'd need to hire to scale this function" lands very differently from "this replaces the three people currently doing this job." The first is a growth story. The second is a threat. ## Borrow Superhuman's playbook Superhuman built one of the most effective value communication frameworks in modern SaaS with a single sentence: *"$30 a month is a dollar a day to get four hours of productivity back."*It does four things at once. It reframes the price into a daily unit that feels trivial. It expresses the value in time, something everyone understands. It implies a clear ROI without requiring the buyer to do any maths. And it makes the abstract (a software subscription) feel like a concrete daily exchange.Every AI company building on a credit model should be able to construct their equivalent.The formula: take your credit bundle price, express it as a daily or per-use cost, and translate it into a felt, tangible outcome. "X credits a month is Y pence per [task] → the cost of a coffee to [specific result your customer cares about]."If you can't construct that sentence for your product, your pricing communication needs work before your pricing model does. ## Make consumption visible One of the biggest sources of credit anxiety for buyers is the fear of the unknown. If they can't see what they're spending in real time, credits start to feel like a black box, and black boxes make procurement teams nervous.The solution isn't simpler pricing. It's better visibility.Real-time dashboards showing credit consumption, usage alerts before thresholds are hit, and forecasting tools that help buyers plan ahead all do the same thing: they convert uncertainty into legibility. A buyer who can see exactly what their credits are doing, and what they're delivering in return, is a buyer who renews.This is also where ROI reporting earns its keep. Showing customers not just what they spent, but what they got, closes the value loop that credits need to sustain themselves. "You consumed 80,000 credits this month. Here's what that delivered: 2,400 documents processed, 14 hours of analyst time saved, $18,000 in estimated cost avoidance." That's a renewal conversation, not a churn conversation. ## Use t-shirt sizing for complexity Not every credit interaction needs to be precisely priced. For buyers who find granular rate cards overwhelming, or for use cases where the range of possible outputs is wide, t-shirt sizing is a practical and effective alternative.Small, Medium, Large. Simple tasks, standard workflows, complex automations. Each tier maps to a credit range, and buyers can develop an intuition for roughly what things cost without needing to interrogate every line item.T-shirt sizing also scales naturally with the value of the work. Increased complexity generally means higher value, which means higher credit consumption, and customers tend to accept that logic intuitively. You're not hiding the pricing; you're making it navigable. ## Train the value before you discuss the price Pilots and proof-of-concept periods exist for a reason: they give customers the experience of value before they're asked to commit to paying for it.The right sequencing matters. Don't ask buyers to evaluate your credit pricing before they've felt what your credits deliver. Run the pilot. Generate the outcomes. Show the data. Then have the pricing conversation with a customer who already knows what a credit is worth; because they've experienced it.This changes the negotiation entirely. You're no longer selling an abstract unit at a price they have to take on faith. You're agreeing on a fair exchange for something they've already seen work. ## The one thing that undermines all of this Everything above falls apart if the internal teams closest to the customer (sales, customer success, solutions engineering), can't articulate credit value consistently and confidently.The most common failure mode we see isn't a bad credit model. It's a good credit model explained badly by a sales team that hasn't been given the tools to communicate it. Value narratives need to be codified, trained, and rehearsed until they're reflexive.Your credit pricing is only as strong as your weakest customer-facing conversation about it. ## In summary Explaining the value of a credit to a customer comes down to four things:**Lead with outcomes, not units.** Make the value felt before you explain the mechanism.**Anchor in human equivalents.** Give buyers a reference point they already know how to evaluate.**Make consumption visible.** Uncertainty is the enemy of credit adoption. Transparency is the cure.**Train before you price.** Let customers experience the value of credits before you ask them to commit to buying them.The companies that get this right won't just sell credits more effectively. They'll build the kind of customer trust that makes expansion natural, churn unlikely, and outcome-based pricing (the long-term destination for AI monetisation), a credible next step.*Paid makes it easy to launch and manage credit-based pricing; from configurable bundles and real-time consumption dashboards to automated top-ups and ROI reporting. **[Talk to us about adding credits to your product →](https://paid.ai/book-audit)* --- ### The Shift from Seats to Credits: What 30+ Founders Told Us Last Week - URL: https://paid.ai/blog/ai-monetization/the-shift-from-seats-to-credits-what-30-founders-told-us-last-week - Date: 2026-03-02 - Author: Calvin Field - Summary: The seat-based SaaS model is breaking. Discover why AI-native founders are shifting to credit-based monetization, featuring frameworks for outcome-based pricing and strategies to transition without losing customers. Over three days in San Francisco, we ran a series of monetisation workshops with founders and operators from some of the most interesting AI companies in the market. The sessions were led by [Madhavan Ramanujam](https://www.linkedin.com/in/madhavansf/) (ex Simon-Kucher) and [Manny Medina](https://www.linkedin.com/in/medinism/) (Paid.ai CEO & Founder), hosted in partnership with Lightspeed Ventures, Emergence Capital, and Sequoia Capital, and the same questions kept coming up in every room.That's a signal worth paying attention to.Here's what we took away. ## Everyone is asking about credits. Nobody feels confident about them yet. The move from seat-based SaaS pricing to credit-based AI monetisation is the defining commercial challenge for AI-native companies right now. Every founder in every session understood that the old model was breaking. But most were wrestling with the same underlying tension: credits feel right in principle, but they're hard to make concrete in practice.How do you explain the value of a credit to a customer who has never bought one before? How do you set the price? What happens when costs fall? Can you change the value of credits after launch? These questions came up so consistently across all three workshops that we're going to write a dedicated post addressing each one. Watch this space. ## The seat model is breaking down — and here's why Traditional SaaS pricing assumed that value scaled with human headcount. More users, more seats, more revenue. It was a clean model for a world where software augmented human work.AI changes the underlying assumption. Agents don't need seats. They consume compute. And the value they deliver — whether that's resolving a support ticket, generating a contract, or running a negotiation — is not linearly related to the number of humans involved.As AI agents reduce the need for human headcount in workflows, pricing per seat starts to misrepresent the value exchange entirely. Companies that hold onto seat-based models will increasingly find themselves in a race to justify pricing that no longer reflects what they actually deliver. ## Why credits work Credits solve the core problem of AI monetisation: how do you price something that does different things at different costs with different levels of value, across an ever-expanding set of workflows?The answer is a single fungible unit that abstracts away the complexity underneath.A credit doesn't price a specific task. It prices AI work — however that work is defined, whatever agent performs it, and regardless of the underlying compute cost. This gives you four critical advantages:**Flexibility.** New features and capabilities can be added without renegotiating contracts. Credits absorb the complexity of product expansion.**Predictability for customers.** Credit bundles give buyers something they can budget against. Paired with spend caps and real-time consumption visibility, they address the fear of unpredictable spend that causes decision paralysis.**Land-and-expand without friction.** Power users buy more credits. There's no seat ceiling to negotiate through, no renewal conversation to have. Expansion happens naturally.**A path to outcome-based pricing.** Credits are the bridge between where most companies are today — copilot mode, seat-adjacent — and where the market is heading: agents doing autonomous work priced on the outcomes they deliver.One of the attendees in our sessions had already made this transition. Their revenue increased 60-70% after switching to a credit-based model. The transition wasn't frictionless — there was some churn as customers adjusted — but the commercial trajectory was unambiguous. ## The Autonomy & Attribution Matrix One of the most useful frameworks we shared across all three sessions was a simple two-by-two that maps pricing models to product characteristics.The two axes are: how autonomous is the AI, and how attributable is its output to a specific outcome?- **Low autonomy + Low attribution** → Seats based model. 'Set it and forget it'- **Low autonomy + High attribution** → hybrid-based pricing. Usually seat based with credits provision.- **High autonomy + Low attribution** → Usage-based pricing. Consumption as a value proxy. The AI is doing a lot, but it's hard to tie to a specific outcome.- **High autonomy + High attribution** → Outcome-based pricing. The ideal end state.Companies like GitHub Copilot and Cursor started in the bottom left and have been evolving toward the right. Sierra is already operating in the top right with fully custom, outcome-based pricing per customer. The matrix is useful not just for understanding where you are today, but for charting a credible path to where you want to be. ## The hybrid transition: how to start without starting over The most common concern we heard across all three workshops wasn't whether to move to credits — it was how to do it without breaking what's already working.The answer is the 80/20 model. Keep your existing subscription pricing. Add an included credit allowance per seat. Design the allowance so that 80% of your customers stay comfortably within it. The remaining 20% — your power users — buy additional credits, creating a natural expansion motion without requiring a full pricing overhaul.The goal isn't to change your price. It's to start building the association between credits and value in your customers' minds. Once that association exists, the conversation about credits becomes much easier — and expanding credit usage becomes frictionless.Offering options also matters. Giving customers a choice between a fixed fee and a credit-based model — with a premium for the predictability of the fixed option — has been shown to increase ACV. Choice architecture creates anchoring effects that work in your favour. ## How you communicate value is as important as the model itself This came up in every session and it deserves its own emphasis. The best credit model in the world will fail if customers don't understand what a credit is worth.The benchmark for value communication is Superhuman: "$30 a month is a dollar a day to get four hours of productivity back." It takes an abstract subscription price and converts it into a felt, daily exchange of value. Every AI company should be able to articulate the equivalent for their product.The Human Value Equivalent (HVE) framework is a powerful tool for this in sales conversations. Framing credits in terms of human work displaced — and crucially, pitching against future hiring or BPO budgets rather than existing headcount — makes the value proposition concrete and separates the pricing conversation from cost entirely. ## The question nobody has fully answered yet: what happens when software uses your software? One of the most thought-provoking questions raised across the workshops was this: as agentic systems proliferate, what happens when it's not a human buying and consuming your product, but another AI system?The traditional assumptions of SaaS — a human user, a human buyer, a human decision-maker — start to break down in a world where agents are orchestrating other agents. The procurement model, the value communication model, the relationship model: all of it changes.The good news is that credits handle this transition gracefully. A credit doesn't care whether the consumer is a human or a system. It prices AI work, full stop. But the broader commercial and contractual implications of machine-to-machine consumption are something the industry needs to work through — and the companies that think about it now will be better positioned when it arrives at scale. ## Where this is all heading The long arc of AI monetisation ends at outcome-based pricing. Technology will increasingly function like an insurance carrier — underwriting the delivery of specific outcomes, and carrying some form of liability if those outcomes aren't delivered. We're not there yet, and most companies are right to focus on the hybrid transition rather than trying to leap to outcome-based models before they have the data or the trust to support it.But the direction is clear. Credits are the currency that makes the journey possible.The most important thing you can do right now is start. Add credits to your packages today — even modestly — so that the association between credits and value is already established by the time you're ready to have the bigger commercial conversation later this year.The companies that build those foundations now will set the commercial standard for the next decade of software.*Paid helps AI-native companies design and implement credit-based monetisation models. If you'd like to run one of these workshops with your team, get in touch.* --- ### From PLG and SLG to CLG: Monetization Playbook for the AI Era - URL: https://paid.ai/blog/ai-monetization/the-Monetization-Playbook-for-the-AI-Era - Date: 2026-02-12 - Author: Manny Medina - Summary: Move beyond seat-based SaaS pricing. Learn a new playbook—Credit-Led Growth—to monetize autonomous AI agents and capture real value. Co-authored by [Madhavan Ramanujam](https://www.linkedin.com/in/madhavansf/), [Joshua Bloom](https://www.linkedin.com/in/jbloom49/) & [Dimi Hiotis](https://www.linkedin.com/in/dimitrishiotis/)For two decades, the software industry has operated primarily under a single, dominant economic model: the seat-based subscription. This model was built for the SaaS era: a period defined by software as a tool designed to enhance human productivity. Legacy SaaS growth models like Product Led Growth (PLG) and Sales Led Growth (SLG) are inherently flawed for AI because they rely on human interaction - measured in seats - as a proxy for value. In the agentic era, software is no longer defined by worker access; it is increasingly a mechanism for autonomous productivity. When software stops assisting the worker and starts performing the work, the "seat" ceases to be an accurate proxy for value. ## **The Fundamental Shift: From Software as an engine to Software as a driver** The transition from traditional SaaS to AI-native applications represents a fundamental change in both the value proportion & cost of goods sold. In the legacy SaaS model, value was anchored in the user’s ability to take the wheel and drive the software to produce an output.  It was also a world where the marginal cost to the vendor for supplying another seat was near zero.In the Agentic AI paradigm, the software is an independent driver. Every "inference" or "agentic action" carries a real-time marginal cost in compute and tokens. Simultaneously, the value shifts from "enablement" to "completion". If an AI agent resolves a customer support ticket or refactors code, it is tapping into **labor budgets** rather than IT software budgets. Labor budgets are traditionally an order of magnitude larger, often **10x larger**, because humans represented the predominant driver of real business outcomes.Failing to recognize this shift leads to a structural misalignment. Founders applying traditional SaaS playbooks, PLG for bottom-up adoption and SLG for top-down selling, often find themselves trapped. They are forced to bundle expensive compute into flat-fee subscriptions, eroding margins while failing to capture the massive "alpha" created by their autonomous agents. To thrive, founders must move toward a value-based contract where the unit of monetization is the work itself. ## **Credit-Led Growth (CLG): Ensuring a repeatable sales processes** To capture this value, companies must adopt Credit-Led Growth (CLG), a model where the vendor sells a standardized bucket of credits with a simple schedule that assigns a fixed number of credits to every agent AI action or outcome.  Each activity draws from the same credit pool, creating a universal mechanism for monetizing autonomous agents and delivering strong sales-market fit without introducing pricing complexity.CLG solves the "expansion friction" inherent in SaaS. In a seat-based model, monetizing a new use case requires a new procurement cycle. In a credit-led model, the user can draw from their existing package (of credits allocated) for any new AI capability the vendor ships, bypassing renegotiation. This creates a scalable, frictionless sales process where teams purchase buckets of credits and only discuss expansion (more credits) once credits are exhausted.Product teams in a CLG world must develop entirely new capabilities centred on engineering for autonomy and attribution. This transformation begins with building agentic workflows where the AI operates independently to complete high-value outcomes rather than merely providing human-centric interfaces. A critical component of this new product motion is the "fair exchange" calibration, requiring an evaluation of agentic outputs to align them with a fair-market price based on the actual work completed. Furthermore, product teams must embed real-time telemetry to track consumption against provisioned credits, specifically focusing on the "Value-to-Burn" ratio. This ensures that every credit consumed is immediately justified by produced value units, such as resolved tickets or completed research briefs, displayed in real-time dashboards. The sales function must similarly evolve from selling tools to managing a strategic digital workforce. The new sales playbook starts with selling an initial standardized bucket of credits that serves as a universal currency for any AI capability the vendor ships. Sales teams must move away from a "set it and forget it" mentality and proactively keep in touch with customers regarding their credit consumption. By monitoring consumption velocity, sales representatives can turn potential overage conversations into value-discovery sessions where they showcase verifiable ROI, such as the number of automated reports created or hard cost savings. This enables the creation of a "CFO View" dashboard that uses historical data to provide an intelligent forecast of how many credits will be needed upon renegotiation. This data-driven approach turns the consumption model into a predictable growth engine, allowing sales to manage renewals as strategic expansions of the customer's synthetic labor force. ## **Operationalizing CLG with ****[Paid.ai](http://paid.ai)**:  Operationalization requires real-time **telemetry of agent tasks** to track consumption against provisioned credits. Crucially, the buy-side experience must prioritize transparency to avoid the bill shock often associated with usage-based models. Buyer-side dashboards are required to view produced value units (e.g., tokens consumed, number of accounts researched, resolutions made) and tools to manage, provision and forecast usage of credits. [Paid.ai](http://paid.ai) enables this by providing the necessary transparency, provisioning, and forecasting tools for buyers to feel in control. In addition [Paid.ai](http://Paid.ai) offers the ability to set targeted promotional credits for new AI workflows to spur engagement and adoption. **Case Study:** The transition to agentic models is rarely just a technical evolution; it is a fundamental commercial overhaul. We see this paradigm shift most clearly in the transition of legacy sectors, such as the move from traditional DevOps tools to "AgentOps." For one such provider, this evolution required a total departure from seat-based subscriptions toward a model of credit-enabled charges for high-value outcomes. Working alongside this business, Paid.ai helped define the foundational steps for this transition:- **Outcome Calibration:** A rigorous evaluation of the new agentic outputs to align them with a fair-market price based on the "work" completed.- **Commercial Re-architecting:** Moving beyond simple subscription selling to design an entirely new sales motion where representatives are trained to sell a proposition of labor rather than a suite of features.- **Operational Implementation:** Deploying a platform capable of handling the complexity of real-time outcome tracking and credit-based provisioning.This case study illustrates that realizing the full economic potential of AI requires founders to master both the big-picture strategy of labor replacement and the granular, detail-focused implementation of a modern commercial stack. ## **Conclusion: Building the Commercial Architecture of the Future** We have moved beyond providing tools for the workforce to providing a workforce that does the work. Monetizing work is not a mere pricing adjustment; it is a transformation of the organization's **Commercial Architecture**. It requires product engineering to incorporate telemetry, sales/RevOps to build ROI models, and value engineering teams to focus on driving --- ### Why AI Agent Deals Don’t Close Like SaaS (and Never Will) - URL: https://paid.ai/blog/ai-monetization/why-ai-agent-deals-don-t-close-like-saas-and-never-will- - Date: 2026-01-22 - Author: Manny Medina - Summary: AI agent deals fail when teams assume they should close like software. For the last decade, SaaS sales optimized around one motion: sell big upfront, lock in seats, defend the renewal. That worked because software value was mostly static.**AI agents break that model.****→ **They don’t behave like features.**→ **They don’t create value evenly.**→ **And they don’t prove themselves at contract signature.**→ **They prove themselves over time. ## **Big lands assume certainty. AI agents introduce uncertainty.** A big SaaS deal works when buyers already understand the value and can predict usage.With AI agents, neither is true.Buyers don’t know how much value an agent will deliver until it runs inside their workflows. CFOs know this too. That’s why large upfront commitments get slowed, scoped down, or killed.What gets approved instead is smaller.A pilot.Not because buyers lack conviction, but because conviction comes from evidence. ## **Pilots aren’t a compromise. They’re the starting point.** In SaaS, pilots were often a hurdle.In AI, pilots are how value is discovered.They answer the only question that matters:does this agent produce real outcomes in our environment?The mistake companies make is treating the pilot as the deal.It isn’t. ## **AI agent deals close after the pilot, not before it.** The winning motion is simple:**→ **Start small.**→ **Prove value quickly.**→ **Expand based on results.When agents work, usage grows.When usage grows, value becomes visible.When value is visible, expansion becomes obvious.Deals don’t “close once.”They compound. ## **Why usage replaces seats** Seats assume humans.Agents aren’t human.They run continuously.They spike unpredictably.They create value unevenly.Usage-based models align better with how agents actually behave, but only when customers can see what’s driving that usage.Without that visibility, usage gets capped and adoption stalls. ## **Natural Expansions vs Artificial Renewals** In SaaS, you have long term contracts that keep you safe from contractions but make expansions hardFor AI agents chargin per usage, expansion is natural. The agent just does more work and it gets paid more.If an agent delivers value, it spreads into new workflows, teams, and budgets long before renewal discussions begin.If it doesn’t, no contract structure will save the deal. ## **This is a reset, not a tactic** AI agents don’t fit the SaaS sales motion.They never will.Pilot replaces big land.Usage replaces seats.Expansion replaces renewals.The teams that accept this early will quietly outgrow the rest.The others will keep pushing for closes buyers aren’t ready to make — and wonder why their best pilots never turn into production. --- ### Predictability is overrated. Transparency is what people actually want. - URL: https://paid.ai/blog/ai-monetization/predictability-is-overrated-transparency-is-what-people-actually-want - Date: 2025-12-21 - Author: Manny Medina - Summary: “Usage-based pricing is unpredictable. Buyers don’t like it.” I hear this all the time. It sounds sensible. It’s also wrong. People don’t dislike usage-based pricing. They dislike moments where they open a product, see a number drop, and immediately think, *wait… what just happened?*The problem isn’t variability.It’s being surprised.The worst moment isn’t usage going up. It’s being asked to explain why it went up and realizing you don’t actually know. That’s when usage stops feeling flexible and starts feeling risky.→ Predictability is “tell me my bill to the dollar a year in advance.”→ Transparency is “show me what’s driving it, and let me control it.”Only one of those works in real systems. ## **We already accept usage when cause and effect are obvious** You don’t predict your electricity bill a year out.You just know that guests stayed over, the showers ran longer, and the bill went up. No panic. No spreadsheet. The story makes sense.AWS works the same way. So do Snowflake, OpenAI, Figma, Clay.All aggressively usage-based.What makes them workable isn’t fixed pricing. It’s that when something spikes, you can see what changed, when it changed, and whether it’s expected.Usage feels fine when cause and effect are obvious.It only feels scary when it isn’t. ## **Where credits quietly go wrong** Credits are meant to simplify usage-based pricing. One abstraction instead of tokens, calls, actions, or workflows. At first, they feel clean and easy to sell.→ Then customers log in.→ They see a balance.→ They use the product.→ The number goes down.And now they’re guessing.→ Was that expected?→ Did someone run the wrong thing?→ Did an agent behave differently overnight?When credits hide the story behind usage, every drop feels suspicious. That’s not a pricing problem. It’s anxiety. ## **Credits don’t fail. Blind credits do.** When credits feel unclear, teams fall into the same two patterns.Some become overly cautious and adoption stalls.Others ignore usage until the bill arrives, trust erodes, and pricing gets blamed.Either way, behavior shifts in the wrong direction.People stop asking “what value did this create?”They start asking “how do we burn fewer credits?”That’s when pricing starts working against growth. ## **AI removes the option to avoid this** Seats feel comforting. One number. One line item. One less thing to explain.But AI agents don’t behave politely. They run autonomously. They spike. They don’t create value evenly. Meanwhile, AI and inference costs don’t follow seat counts.Whether you price by outcomes, actions, workflows, or resources, you’re in a usage economy now.That economy only works if customers feel informed, in control, and confident as usage grows. ## **We’re arguing about the wrong thing** The industry keeps debating usage versus seats. Variable versus fixed. Predictable versus unpredictable.That’s the wrong debate.People don’t want a perfectly predictable bill.They want a bill that makes sense.**Usage isn’t the problem.****Not understanding what’s driving it is.**That’s what actually scales.If you’re building AI agents and thinking about how usage-based pricing and credits fit into that world, we’re working on **Paid Credits** - our take on using credits as a transparent usage layer that helps customers understand what’s driving usage, see what’s coming next, and stay in control as they scale. [Join the waitlist](https://paid.ai/product/credit) to get early access. --- ### The discovery trap: Why your best SaaS sellers are leaving money on the table with AI agents - URL: https://paid.ai/blog/ai-monetization/the-discovery-trap-why-your-best-saas-sellers-are-leaving-money-on-the-table-with-ai-agents - Date: 2025-12-05 - Author: Manny Medina - Summary: Your top performers - the ones who've been crushing SaaS quotas for years - are systematically undervaluing AI agents by 70-80%. Not because they're bad sellers. But because they're too good at what they've always done. There's a paradox killing AI agent deals right now.Your top performers - the ones who've been crushing SaaS quotas for years - are systematically undervaluing AI agents by 70-80%. Not because they're bad sellers. But because they're *too good* at what they've always done. ## **The comfortable prison of tools budget** Here's what I'm seeing: A seller demos an AI agent that can handle 80% of a support team's ticket volume. The buyer's eyes light up. The math is obvious - this could replace 8-10 FTEs or an entire BPO contract worth $800K/year.But then muscle memory kicks in.The seller pivots to ROI on Zendesk licenses. They position against the $50K tools budget. They close a $75K deal and hit their number. Everyone celebrates.Meanwhile, $725K in value just walked out the door. ## **Why great sellers stay small** Upton Sinclair nailed it: "It is difficult to get a man to understand something when his salary depends upon his not understanding it."Except here's the twist - their salary *does* depend on understanding it. They're just too successful to see it.Your President's Club winners got there by mastering a specific discovery playbook:- Find the pain in current systems- Quantify the cost of inefficient tools- Present your solution as a better mousetrap- Close against the IT/tools budget line itemThis playbook is so deeply ingrained that even when they're staring at a labor replacement opportunity, they reflexively steer the conversation back to software savings.It's comfortable. It's predictable. It works.And it's leaving millions on the table. ## **The new discovery map** AI agents don't just optimize workflows - they eliminate entire cost centers. Your discovery needs to match that reality.Stop asking: "What tools are you using?" Start asking: "What work are humans doing that shouldn't require human judgment?"Stop asking: "What's your software spend?" Start asking: "What's your fully-loaded cost per employee in this function?"Stop asking: "Who owns the tools budget?" Start asking: "Who owns the P&L for this entire operation?" ## **Three budgets you're not touching** **1. The shadow headcount budget** - Every department has approved headcount they can't fill or haven't filled yet. That's not in the tools budget, it's in the CFO's workforce planning model. An AI agent that eliminates the need for those hires is worth 10x the software it might replace.**2. The BPO/outsourcing spend** - Companies spending $2M/year on offshore customer service aren't thinking about that as "software spend." But an AI agent that handles those interactions better, faster, and cheaper? That's a direct substitution play.**3. The redeployment opportunity** - The highest-value positioning isn't even about cost reduction. It's about what happens when you free up 20 talented humans from repetitive work. What could they do instead? What new revenue could they drive? That value doesn't live in any tools budget. ## **Breaking the muscle memory** The hardest part isn't learning new discovery questions. It's *unlearning* the old ones when they've made you successful.Here's my advice: Run two discoveries in parallel for your next three deals.First, run your traditional SaaS discovery. Find the tools pain. Get to your usual number.Then, run it again with a labor lens. Map the human work. Count the fully-loaded costs. Find the workforce planning owner.Compare the two ACVs.I guarantee the difference will break your addiction to the tools budget. ## **The choice** You can keep closing $50K deals against Salesforce licenses while your competitors close $500K deals against headcount.You can keep having comfortable conversations with IT buyers while others are in the C-suite talking workforce transformation.You can keep being successful in a shrinking pool.Or you can evolve.The market has already chosen. AI agents aren't tools—they're digital workers. Price them accordingly, or watch someone else do it.Your commission check will thank you. --- ### The discovery framework for unlocking bigger budget pools - URL: https://paid.ai/blog/ai-monetization/discovery-framework-unlocking-bigger-budget-pools - Date: 2025-12-02 - Author: Manny Medina - Summary: Most SaaS companies are used to selling against the $1 software and tools budget pool. The new guard of AI natives are positioning themselves as labor replacement and fishing in the $10 labor pool. AI agents are unlocking new and bigger budget pools. As Matthew Scullion, CEO of [Matillion,](https://www.matillion.com/) put it on a recent episode of the [Get Paid podcast](https://youtube.com/playlist?list=PL5qAB_HAibeq94HkOVFLBhO2uN2nMUZJo&si=NoKpBmc-LUp1cWNH), for every $1 a company spends on software, they spend $8-10 on the labor around it. Think administration, training, implementation, management etc.Most SaaS companies are used to selling against the $1 software and tools budget pool - a pool that is already stretched thin to cover all of the dozens of tools customers are using to run their businesses. The new guard of AI natives are playing differently. They are positioning themselves as labor replacement and fishing in the $10 labor pool. As a result, they’re landing deals that are 2-5x higher ACV than their SaaS predecessors. If you’re a SaaS sales team selling in AI, you need to start fishing in the labor pool for much bigger wins. ## A new way of doing discovery To unlock bigger budget pools, you need to rethink how you do discovery. MEDDPICC taught us to find pain, implicate that pain,  and position our product as the solution. That worked when we were selling software.But accessing labor budgets requires a different approach. You need to uncover not just the pain, but **how they're solving it right now, **or how they're planning to solve it in the near future. In most cases, the answer is hire someone, or outsource the problem to a BPO.I caught myself doing this recently. While discussing business gaps with my co-founder Raj, his instinct was to buy or build a tool. Mine? Hire someone to fix it."Why are you always hiring people?" he asked.Good question. My cohort of SaaS founders has been conditioned to solve problems by finding smart people. I was ready to open a $200K role. He was thinking of a $10K AI tool that does the same thing. Same value, wildly different price tag.**If you're not uncovering this default behavior in discovery, you're leaving money on the table.** ## Three ways to unlock the labor pool If you're a founder, CEO, CRO or sales leader selling AI capabilities, here's what needs to change:**1. Expand your discovery questions**Find the pain AND how they're currently solving it. Where are those budgets coming from? How large are they? Map the full cost of their current approach - salaries, benefits, training, turnover, BPOs, management overhead etc.**2. Implicate both the pain AND the solution**If they open that headcount or outsource to a BPO, what's the budget impact this year? Next year? What about time-to-productivity? Scaling limitations? Compete with the hiring plan, not your SaaS peers.**3. Leverage the agentic moment**Boards are asking CEOs for agentic strategies. Position yourself as the agentic solution and give your champions something compelling to take back to the board. Your competitors are solving problems, but you’re delivering on a strategic mandate. ## More than just a sales problem If you're serious about capturing labor budgets, your entire company needs to rethink how you adapt strategy to compete with hiring plans. - **What you're building** – Are your AI features positioned as "nice-to-haves" or labor replacement?- **How you're positioning it** – Does your messaging speak to software buyers or budget owners?- **How you're marketing it** – Are you creating content that resonates with the labor replacement narrative?The companies that figure this out will command premium pricing and win strategic, high-ACV deals. The ones that don't will continue splitting that $1 software budget with everyone else. ## Which pool are you fishing in? If you're still competing on features and fighting for software budgets, you're playing the old game. The new game is about replacing $200K hires, and capturing a fraction of that labor budget.The labor pool is 10x bigger. The deals are 2-5x larger. And the buyers are ready.Are you? ## Get back to growth At Paid, we’ve built the world’s first growth engine for AI, helping companies navigate the transition from SaaS to AI agents. [Book a demo today](https://paid.ai/get-started) to see how Paid can help you get back to growth. --- ### The 5 rules for winning AI proofs of value - URL: https://paid.ai/blog/ai-monetization/5-rules-for-winning-ai-proofs-of-value - Date: 2025-11-28 - Author: Manny Medina - Summary: Winning companies will be the ones proving value in the shortest possible time, locking out competitors and creating customers who succeed - and who stick. The proof of concept is dead. Long live the proof of value. SaaS companies continue to make the mistake of using proofs of concept to prove that their technology works. Does the integration work? Can it handle the data volume? Does it meet the security requirements?These are important questions to ask, but they're not the questions that close deals.What buyers really want to know is, what does this mean for me? Both personally and professionally. Your buyer isn’t going to get promoted because they successfully validated a vendor's API. They get promoted because they delivered measurable business value. They boosted revenue. They improved customer loyalty. They solved a problem that was keeping executives up at night.The companies that successfully turn proofs of concept into repeatable revenue are those who use this period to prove the value, not the concept. ## Trading short sales cycles for higher conversion rates Sales leaders often resist POCs because they worry about lengthening sales cycles. And they’re right. POCs or POVs will lengthen the sales cycle by the length of the POV period. But what most sales leaders overlook is the fact that POVs drastically improve conversion rates. A typical opportunity moving from S1 to closed-won converts at 10-20%. But an opportunity that progresses to a POV? In my experience, conversion rate jumps to over 80%. Yes, your sales cycle gets longer. But if you get in first, you lock competitors out. Most companies won't run multiple POVs simultaneously. Get in first, prove value fast, and you've effectively closed the door behind you. ## What if your POV fails? The simple answer is, don't let it.This is about going all-in on customer success in a way most vendors won't. Throw your forward-deployed team at it. Embed your people with theirs. Make their win your only job for the duration of the POV. Because if you don't, someone else will. Now, sometimes things still don't work out. Maybe there's team misalignment. Maybe there's a tech issue you didn't anticipate. Fine. Figure out what happened, tell them directly, and reset the POV. Don’t mark it as closed-lost and give up. This is your chance to prove that your product delivers the value you say it does. Try again. ## The 5-step POV playbook Here's how to win consistently with proofs of value:**1. Get there first - **When it comes to POVs, speed matters. Be the first vendor to propose a POV, and you'll close the door on competitors who are still scheduling demos.**2. Study their POV history - **Do your homework. What did they pay for similar pilots in the past? What worked? What failed? Understanding their POV track record gives you a massive advantage in structuring yours for success.**3. Find their approval threshold - **Know exactly how much they can spend without requiring CEO or CFO sign-off. This number determines your pricing strategy and timeline.When in doubt, charge double the last successful pilot price.** **Most companies leave significant money on the table during POVs. **4. Win fast - **Once you’re in, you need all hands on deck. Embed your team with theirs. Prove value as quickly as possible. The faster you demonstrate results, the faster you close, and the less opportunity competitors have to interfere.**5. Get the testimonial immediately - **The moment you win the POV and close the deal, capture a testimonial. Use it to win your next POV. Then repeat the process.POV customers are way less likely to churn than those who buy without a POV. They've already experienced your value firsthand. They've invested time and political capital in bringing you in. They've built internal momentum around your solution.  ## The POC is dead. Long live the POV. If you’re not running POVs, you can be sure somebody else will. If your competitors get in first, they’re going to close the door on you. Don’t let them. Winning companies will be the ones proving value in the shortest possible time, locking out competitors and creating customers who succeed - and who stick. So stop proving your technology works and start proving what it means to the people who matter. ## Get back to growth At Paid, we’ve built the world’s first growth engine for AI, helping companies navigate the transition from SaaS to AI agents. [Book a demo today](https://paid.ai/get-started) to see how Paid can help you get back to growth. --- ### The pricing model that kills growth (and two alternatives for high growth SaaS) - URL: https://paid.ai/blog/ai-monetization/the-pricing-model-that-kills-growth - Date: 2025-11-24 - Author: Manny Medina - Summary: More and more SaaS companies are launching AI features, or even entirely new AI products. But many of them are approaching AI pricing in the same way they priced features in 2015. And that's where billions in potential value are being left on the table. SaaS is going through a crisis. According to [Aventis Advisors’ SaaS valuation report](https://aventis-advisors.com/saas-valuation-multiples/#growth), growth rates have dropped from 33% pre-COVID to just 13% today.  [Altimeter’s Jamin Ball](https://www.saastr.com/the-great-saas-slowdown-what-q1-2025-numbers-reveal-about-the-cloud-software-market/) says SaaS companies added $1.65bn net new ARR in Q1 2025, down from $2.33 billion in Q1 2024 - a year on year decline of 29%.For an industry built on predictable, compounding growth, these figures signal a fundamental shift in how the SaaS market operates.Billions of dollars in enterprise value are now trapped in what can only be described as SaaS no man's land. It looks bleak, but some of the more innovative SaaS companies have realised that the solution to this crisis is already sitting in their data.  ## The hidden asset in every SaaS company Your SaaS company possesses something that AI-native startups would pay millions for. You have years of workflow data, documented operating procedures, and institutional knowledge encoded. Customers have been teaching you how they work for years. You know their processes, their pain points, and their desired outcomes.The companies pulling ahead right now are transforming this accumulated knowledge into AI agents that can execute work, not just assist with it.More and more SaaS companies are launching AI features, or even entirely new AI products. But many of them are approaching AI pricing in the same way they priced features in 2015. And that's where billions in potential value are being left on the table. ## The pricing strategy that kills growth: The mega seat When SaaS companies decide to "add AI," the default move is to bundle it into existing seat licenses at a premium price. It feels safe because it doesn't require rethinking your sales or customer success motions.This is exactly the wrong approach.Procurement teams are already cutting 10-20% of low-usage seats. Workforce reductions mean fewer potential buyers. It doesn’t make sense to respond to seat compression with bigger seats. Your champion might buy in, but when they go to procurement, they’re going to be told to push for the premium tier at the same price they're paying today. You don’t agree? No problem, they’ll go elsewhere. It might work in the short term, but in the long term, you end up with flat renewals or customer churn. ## Today’s winner: Seat + credits This is where many companies should start today.The seat + credits model attaches AI consumption or outcome credits to your existing seat licenses. Credits allow you to start demonstrating value without shifting to a complete outcome-based model. Done right, a seat + credits model will teach your customers how to buy differently, gradually shifting their mental model from productivity upticks per seat to value created at large. The gotcha here is usage. If your customers aren’t spending their credits, they’ll downgrade, or even churn. To get past this, you need to retrain customers to use your AI features - no small feat when they’re comfortable with the old way. If you don’t invest in driving adoption, you can bet that a new AI-native company will convince your champions that your 10-year old architecture is obsolete. They'll position themselves as the future while you're still the past with AI sprinkled on top.The SaaS winners will see this as an opportunity, using seat + credit models to move closer to a pure outcome-based world.  ## The future: Launch an AI-native product and charge for outcomes For most companies, the ultimate destination is a standalone AI product with outcome-based pricing. Adoption will be slower in some spaces than others. Customer service agents like Fin and Sierra are using outcome-based pricing, charging per resolution of customer service inquiry. Some systems of record type products like CRM and ERP might have a harder time moving to outcomes. For these companies, it’s harder to define what outcomes might look like, and it’ll be harder still to educate customers to shift their mindsets to outcome-based pricing.That said, companies that *can* launch new AI-native products and charge for outcomes will see massive benefits. **1. You own the category before competitors define it.** When Intercom launched Fin as a distinct product, they didn't just add a feature, they claimed ownership of "AI customer service agent." That's a category they can defend for years.**2. Pricing aligns with value creation.** When customers pay for outcomes (tickets resolved, reports generated, workflows completed) rather than seats, there's no argument about ROI. The product literally pays for itself in measurable terms.**3. Sales cycles collapse.** Instead of navigating seat reallocation and budget shuffles, you're selling net new value. Finance teams don't need to approve a replacement, they're approving an investment that reduces costs elsewhere.**4. You escape the seat compression spiral.** Your revenue is no longer tied to headcount. In fact, your product becomes more valuable when it helps customers do more with less, which is exactly what their boards are demanding.Salesforce didn't just add AI features to Sales Cloud. They launched Agentforce as a distinct product because they understood that the buying motion, pricing model, and value proposition were fundamentally different. ## Moving towards outcomes, even if you’re not there yet The companies already making this shift are pulling ahead. The rest are still debating seat pricing while their ARR bleeds out.Purely agentic products and outcome-based pricing aren’t for everyone, but the pendulum is swinging in that direction. Those who move fast and figure out how to package their data into AI agents, and price for value accordingly, will emerge as the leaders of the next wave of SaaS.  ## Get back to growth At Paid, we’ve built the world’s first growth engine for AI, helping companies navigate the transition from SaaS to AI agents. [Book a demo today](https://paid.ai/get-started) to see how Paid can help you get back to growth. --- ### The rule of 40 breaks for AI agents (and how to fix it) - URL: https://paid.ai/blog/ai-monetization/rule-of-40-ai-agents - Date: 2025-11-21 - Author: Manny Medina - Summary: The traditional metrics that guided software companies for decades are breaking down, creating a profitability crisis across the AI industry. Here's what's happening and how the winning SaaS companies are adapting their business models to survive the shift. If you're a SaaS company transitioning to AI agents, you've likely noticed something alarming. Your unit economics don't make sense anymore. The traditional metrics that guided software companies for decades are breaking down, creating a profitability crisis across the AI industry.Here's what's happening and how the winning SaaS companies are adapting their business models to survive the shift. ## The rule of 40 For those unfamiliar, the Rule of 40 has been the gold standard for measuring SaaS company health. The formula is simple: **growth rate + profit margin = rule of 40 score.**A healthy company will score 40+.A score of under 40 indicates that your company is not balancing growth and profitability effectively.A healthy SaaS company typically operates with:- 20% net profit margin- 20% year-over-year growth- Rule of 40 score: 40% (meeting the benchmark)This worked because traditional SaaS had predictable cost structures that looked something like this:- 20% Cost of Goods Sold (COGS)- 25% sales & marketing- 25% research & development- 10% general & administrative- 20% net profitWith 20% profit, 20% growth was sufficient to keep your company in the black. ## AI agent economics are different The rule of 40 looks very different for AI-native companies, or SaaS companies with strong AI offerings. The biggest difference is COGS. You still have all your original costs, but now you have a ton of AI costs associated with LLM and tool calls. Instead of 20%, COGS is more like 40%,But **none of your other costs go down**.That’s probably why [Bessemer’s State of AI 2025 report](https://www.bvp.com/atlas/the-state-of-ai-2025) found that negative margins are not unusual for the fastest growing AI companies. Let’s look at what happens to your profit when COGS goes up.- 40% COGS- 25% sales & marketing- 25% research & development- 10% general & administrative- **-20% net profit**With -20% profit margins, AI agent companies need to grow at **60% year-over-year just to meet the rule of 40 threshold**.The hottest AI companies are hitting these growth rates. But this pace isn’t sustainable long-term. Something has to give. ## Where are you going to cut? Okay, so it’s not looking good. But surely you can make some cuts and get those margins back to a healthy level?This is a reasonable reaction, but to make the numbers make sense, these cuts are going to have to come in the form of layoffs. And layoffs are going to impact growth. If you cut your R&D costs, you’ll be losing the engineering talent that gives your product its competitive edge. In fact, as you build more agentic capabilities into your products, you’re going to need to spend *more* on great AI and ML talent, not less. Maybe you can tighten up G&A, but making significant cuts to your perks, rent, administrative staff, travel costs and/or software licenses is going to drive your top talent into the hands of your competitors.  It’s tempting to think you can cut back on sales and marketing. Recouping those 12% commissions you’re paying out to salespeople would go some way towards restoring your margins. But during this shift from traditional SaaS to agentic AI, sales and marketing is more important than ever. You’ll need to make significant investments in marketing to retrain your ICP on your new way of thinking. Salespeople will need to be strategic advisors, hand holding customers through the most disruptive software shift since SaaS. There might be some wiggle room here. And it might look a bit different to what you’re used to… ## Aligning sales compensation with margins Hear me out. What if more companies started moving sales commissions from percentage of revenue to percentage of margins?It’s not that far out. Door-to-door salespeople selling encyclopedias and cookware were selling this way long before SaaS existed. If they offered a discount, it came out of their own pockets. Margin-based compensation aligns sellers with the same metrics that matter to your business: more deals, bigger deals, and bigger margins.When sellers are compensated on margins, they're incentivized to:- Negotiate pricing that reflects value- Upsell higher-margin features and capabilities- Qualify prospects who will be profitable customers- Minimize discounting that erodes already-thin margins ## The broader implications for AI SaaS This idea of shifting to margin-based compensation is just one way companies might rethink their business models in the journey from traditional SaaS to agentic AI. Other moves will include:- **A shift toward outcome-based pricing**: Quantifying the value a product delivers will enable companies to increase ACVs while remaining highly competitive. - **New product lines**: More companies will break out AI functionality into entirely new product lines where they can retrain customers on how to think about value. - **More quantifiable proofs of value**: The proof of concept is here to stay, but companies will need to quantifiably demonstrate value to turn POCs into ARR.The companies that adapt quickly will have a significant competitive advantage. Those that try to operate AI agent businesses with traditional SaaS assumptions will struggle to achieve sustainable profitability. ## **Get back to growth** At Paid, we’ve built the world’s first growth engine for AI, helping companies navigate the transition from SaaS to AI agents. [Book a demo today](https://paid.ai/get-started) to see how Paid can help you get back to growth. --- ### How SaaS companies can transition from subscriptions to outcome-based pricing - URL: https://paid.ai/blog/ai-monetization/subscriptions-to-outcome-based-pricing - Date: 2025-11-18 - Author: Sidin Vadukut - Summary: A framework to help SaaS companies find the right pricing model and transition from seat-based to outcome-based pricing. SaaS companies face an uncomfortable reality. You've built a business on seat-based pricing, but AI agents are making that model obsolete. Your agents are starting to displace the seats you bill for, while your compute costs climb.As revenue shrinks, and your margins compress, you're catching what the author of *Monetizing Innovation* and *Scaling Innovation*, Madhavan Ramanujam calls "a falling knife."So how do SaaS companies respond to this reality?When we spoke with [Madhavan on the Get Paid podcast](https://agenttalk.substack.com/p/price-before-product-madhavan), he described two distinct approaches: the incremental path where you move from seat-based pricing to a hybrid model, and the revolutionary path where you build net-new agent products with outcome-based pricing from day one. Madhavan's latest book *[Scaling Innovation](https://www.simon-kucher.com/en/insights/scaling-innovation-how-smart-companies-architect-profitable-growth-0)**,* co-authored with Eddie Hartman, provides a framework that clarifies which path fits your current situation. Built on two dimensions, the framework determines not just which pricing model works, but which transition path you should take. ## **The attribution-autonomy framework** Madhavan's framework measures two things:- Autonomy asks whether your agents can work independently or need humans in the loop. - Attribution asks whether you can prove the agent created a specific outcome. Plot these on a grid and you get four quadrants, each representing where companies sit today and where they need to move.Most SaaS companies discover they're in the bottom left quadrant with low autonomy and low attribution. Your products assist humans but don't complete work independently, and you can't clearly prove their impact on business outcomes. This is where a product like Slack lives today. Everyone knows it boosts productivity, but you can't measure it or charge based on it. Workday, Figma, and Grammarly are here too. The value exists, but the attribution doesn't.If you're deploying agents from this position, you're in trouble. Every agent you ship displaces seats you're billing for.  ## **The incremental path: Move to a hybrid pricing model** The incremental path means moving from a seat-based model to a hybrid model, typically by offering your AI capabilities in bundles of credits on top of your monthly subscription. This is what Clay did. Users pay a regular seat-based subscription with additional credits to spend on different data points and actions.This hybrid model hedges both ways where seats maintain predictable revenue, while consumption creates expansion as agent usage grows.Madhavan's advice for this path is to bundle a fixed monthly price with included usage. Convert your existing tiers to hybrid bundles where, for instance, Professional becomes $200 per seat with 3,000 tokens included, and Enterprise becomes $500 per seat with 10,000 tokens. When customers exceed their allocation, "you have an automatic conversation. You can just go to the next plan. You invented a land and expand model."The key is defining what a token represents. Don't tie it to compute costs because those can vary 100x based on task complexity. Tie it to value delivered where one token equals one completed unit of work, like a ticket resolved or a meeting scheduled. This requires infrastructure that tracks outcomes and calculates what that work would have cost if done manually, not just API calls and compute time.Most companies should take this incremental path because it requires less technical sophistication, less customer education, and less risk. You're meeting customers where they are with seat contracts, and layering on consumption models they've already seen with AWS or Snowflake. But understand what "incremental" actually requires: you need infrastructure that tracks outcomes, calculates value, and ensures expansion improves agent profitability instead of eroding it. ## **The revolutionary path: Build for outcomes from day one** The revolutionary path means jumping from the bottom left quadrant directly to the top right where you have both high autonomy and high attribution. You build a net-new agent product that works independently and prices on outcomes from the start.This is what Intercom did with Fin. They didn't retrofit their existing support platform or add consumption charges to existing plans. They built Fin as a completely separate product with its own pricing model where users pay $0.99 per resolution. ChargeFlow took the same approach by autonomously recovering disputed chargebacks and charging 25% of recovered revenue. The outcome is measurable, the attribution is clear, and the value split is explicit. No outcome means no charge. Perfect incentive alignment.Madhavan notes that companies in this quadrant can capture 25-50% of the value they create, compared to 10% for traditional SaaS. The difference comes from increased autonomy and provable attribution. When you can demonstrate causation and quantify impact, customers will pay for results at rates that seem impossible under seat-based models.To really enable this path Madhavan recommends treating early customer conversations as business case building exercises rather than tech validation. Co-create the case with customers to understand what metrics matter to them and how they measure success. This sets up natural outcome-based pricing conversations because you've quantified the value together.Revolution, to be sure, comes with risks. This path works if your core business is stable enough to fund experimentation and you have technical sophistication to build closed-loop attribution. You need to identify workflows where you can achieve both high autonomy and high attribution. Support ticket resolution is the canonical example with its clear start point, clear end point, and measurable outcome. But look for others like sales qualification, data processing, or compliance checking. ## **Choosing the right pricing model for you** Where are you today on autonomy and attribution? Which dimension can you improve faster? That determines your path.If you can build better measurement systems to prove agent value but can't yet build fully autonomous agents, take the incremental path. Move to hybrid pricing where you keep agents in the copilot role but introduce consumption charges based on demonstrated value. This is the safer path that most companies should follow.If you can build both high autonomy and closed-loop attribution, consider the revolutionary path. Build a net-new agent product that works independently and prices on outcomes from day one. Run it as a separate initiative with its own economics, like Intercom did with Fin. This path captures more value but requires more sophistication.Some companies will do both. Hybrid pricing for the core product protects existing revenue while you build outcome-based pricing for new agent products to capture the future. The hybrid model funds the transition while the outcome model proves the economics work.But staying in the bottom left quadrant isn't an option. You need to transition to agent pricing, and these are the two paths that work.The companies that choose a path deliberately and build the infrastructure to support it will thrive. The ones that wait will watch their competitive position erode. Until one day the transition becomes a crisis instead of a choice.Are you exploring pricing models for your AI offering? [Book a Paid demo today](https://paid.ai/get-started) and see how the world’s only growth engine for AI can help you. --- ### Why tracking and billing can't be separate systems anymore - URL: https://paid.ai/blog/ai-monetization/why-tracking-and-billing-cant-be-separate-systems-anymore - Date: 2025-11-14 - Author: Sidin Vadukut - Summary: When tracking and billing are unified around human equivalent value, everything changes. You can see what agents cost to run versus what value they delivered. When you can see both numbers, you can find the pricing that captures value while protecting margins. The agent economy has arrived faster than anyone expected.Companies that spent years perfecting their SaaS businesses suddenly find themselves deploying AI agents: autonomous systems working 24/7, handling tasks that used to require people. Sales agents now book meetings, support agents resolve tickets, and operations agents process workflows.There’s just one problem: we're trying to monetize these agents using infrastructure built for a completely different business model. ## The SaaS inheritance problem Most companies transitioning to agents today inherited their billing approach from traditional SaaS: an approach that ran two systems in parallel. One system tracks what agents do: tasks completed, minutes run, tokens consumed. Another system sends invoices, built on solutions like Stripe or Chargebee.This worked fine when you charged per seat. Your billing system counted active users and multiplied by $99. You didn't need to know what those users actually *did* because usage didn't affect pricing. A seat was a seat.Agents don't work that way. They run continuously, doing different work for different customers with volumes that vary wildly. Now your legacy tracking system knows what happened: Agent A processed 2,847 support tickets, Agent B booked 47 sales meetings, and Agent C handled 1,203 customer queries. Your legacy billing system knows what you charged: Customer X paid $5,000. Customer Y paid $12,000.Problem is, now nobody knows whether those numbers match up. Whether you charged the right amount to the right customer, and whether you're even making money. As a SaaS company transitioning into AI, what you really need are tracking and billing systems that talk to each other and share the same data.Better yet: a unified system that builds both on top of a single source of truth. ## Data silos stack up fast Jonathan Sanders, CEO and founder of Light, saw this SaaS inheritance problem everywhere during his years at Cleo and Juni. Finance systems that don't talk to operational systems. Data trapped in silos. All culminating in teams fighting over whose numbers are correct.When [we spoke with him on the Get Paid podcast](https://agenttalk.substack.com/p/netsuite-is-where-erps-go-to-die), Jonathan described the real costs: "The downside to this siloed system is that you can't experiment, you can't prove value to customers, and you can't actually know if you're making money. Everything becomes manual reconciliation and guesswork."The practical implications of these data silos stack up fast.Your tracking system knows an agent resolved thousands of support tickets. But what's that worth? How many hours of human work did that represent? What would it have cost to do manually? Your billing system has no way to calculate this, so you're either guessing at pricing or defaulting to cost-plus models that miss the value you're creating.The two-system approach also hampers your ability to experiment. Want to switch from per-minute billing to outcome-based pricing? You need to rebuild both systems and manually map data between them. That project takes months. By the time you've solved this problem at human-speed, the market has moved at agent-speed.The stakes are existential. Agents work invisibly. Which means customers can't see what's happening, so they don't feel the value. In our experience 80% of agent companies today hit a renewal cliff between months 6 and 9. Without proof, you're asking customers to renew on faith. And faith doesn't survive budget reviews.Meanwhile, as agents get more versatile, numerous, and performant, legacy systems break under the weight of reconciliation. Every month, someone pulls usage data from one system, billing data from another, and tries to reconcile them manually in Excel. ## Pricing on value, not costs When we started building Paid, we asked ourselves: what does agent billing actually need to do?The answer became clear once we stopped thinking about compute costs and started thinking about value.No doubt compute costs matter. The fastest-growing agent companies average just 25% margins compared to 80% for SaaS, with some top performers running negative. [That's almost always a good thing.](https://paid.ai/blog/ai-monetization/80-margins-mean-your-agents-aren-t-working)But you can't price on costs alone when a single agent operation might cost $0.10 or $10 depending on complexity. Cost-plus pricing breaks when costs can vary 100x or more.What you can price on is value delivered. One easy way to do this is to measure value in labor terms: Human Equivalent Value (HEV).A support agent that resolves 2,847 tickets in a month costs you $340 to run. But it delivers 712 hours of human work. At a typical support analyst rate of $45 per hour, that's $32,040 in value. You need to know both numbers: what it cost you *and* what it's worth to the customer.Calculating HEV means understanding what agents actually do, then billing for that value. And ideally you need one system that captures agent signals, translates them into human equivalent work, and prices them based on value delivered.That's how we built Paid. We capture every action your agent takes and calculate the human equivalent work: this would have taken 15 minutes of analyst time, or this replaced 2 hours of sales rep work. And we enable you to price the task accordingly.With our platform, customers configure pricing rules based on value, not cost. Charge $50 per meeting booked because a meeting booked is worth far more than the $0.30 it cost in compute. Charge $15 per support ticket resolved because you're saving the customer an hour of manual work.The unified Paid system tracks the signals, calculates human equivalent value, applies your pricing rules, and generates invoices. One data set. One source of truth. ## Aligning around value When tracking and billing are unified around human equivalent value, everything changes.You can see what agents cost to run versus what value they delivered. When you can see both numbers, you can find the pricing that captures value while protecting margins.The transparency that emerges from this completely shifts customer conversations. Instead of defending costs, you're now demonstrating value: "Your agents booked 47 sales meetings this month. That's 94 hours of SDR work, equivalent to $8,460 in fully loaded labor cost. We charged you $2,350."In fact, you don't need to limit your thinking to human work. While it might have cost $8,640 to have reps book those 47 meetings, the value of those meetings will be much higher. Let's say every meeting booked drives $10k in sales. Now your agent has created $470,000 in value. That $2,350 you charged them suddenly looks pretty low.Pricing experiments that used to take months now happen in hours. Want to change how you define outcomes by use case or customer? Change the rule. Want to tier pricing by complexity? Add the calculation. Want to offer volume discounts on high-value work? Just input the numbers and generate an invoice. With a unified agent growth platform, all of this is possible without rebuilding your systems.And the manual reconciliation work just disappears. Usage, value, and revenue live in one place. Reports generate automatically.This is the sustainable and scalable infrastructure that makes durable agent economics possible. ## Focus on growth, not just billing If you’re growing your AI offering right now, look at your billing infrastructure. Can you measure human equivalent value for the work your agents do?Instead of asking "how many tokens did this consume?" or "how many minutes did this run?" you should be asking "how much human work did this replace, and what would that work have cost?"If you can't answer that question with data, you're working with a fundamental handicap.Traditional billing platforms weren't built for this. Stripe and Chargebee are excellent at processing payments based on fixed contracts and seat counts. But they can't understand agent signals, the discrete work events that need to be translated into value and priced accordingly.That's why companies using these platforms for agent billing end up building custom tracking systems. And even then, they're usually just tracking usage metrics, not calculating the value of work delivered.You need a unified system that captures agent signals, calculates value, and bills your customers based on value delivered. You need one source of truth for what agents do, what that work is worth, and what you charge. --- ### Why 80% margins mean your agents aren't working - URL: https://paid.ai/blog/ai-monetization/80-margins-mean-your-agents-aren-t-working - Date: 2025-11-07 - Author: Manny Medina - Summary: The companies actually winning with AI are seeing margins in the 40-60% range, and that's exactly how they know their AI strategy is working. As SaaS companies race to introduce AI agents, most are adopting a hybrid pricing model - seat-based fees plus credit bundles for AI features.I recently spoke to a company that's seeing upwards of 80% margins on this model and that to me is a major red flag.Why? Because 80% margins tell me that your AI agents aren’t doing much of anything.Most companies actually winning with AI are seeing margins in the 40-60% range. According to [Bessemer's State of AI 2025 report](https://www.bvp.com/atlas/the-state-of-ai-2025), it's more like 25% for the fastest growing AI companies, with many actually seeing negative gross margins. ## **80% margins reveal low agentic use** The seat + credit pricing model has become the default for AI features in SaaS, and for good reason. It's familiar to customers, easy to understand, and provides a clear path to land and expand. But an 80% margin on credits means your users aren’t getting value from your AI features. It means:- Customers aren't consuming enough AI to impact your costs meaningfully- Your AI features aren't central enough to user workflows to drive real usage- You're charging premium prices for functionality that isn't delivering proportional value ## **The 40-60% sweet spot** So why do margins compress to 40-60% specifically when AI is working?This range represents the equilibrium point where AI is delivering genuine value to users, but you're investing enough to do it right. You've moved beyond trivial automation into meaningful product transformation.Here's what companies in this margin range typically look like:- **High credit consumption.** Users aren't just trying the AI feature occasionally, they're burning through credits because the AI functionality has become essential to their workflow. This drives up your compute costs significantly.- **Deep product integration.** The AI isn't a separate feature users navigate to; it's woven throughout the core product experience. This requires substantial engineering investment to build the connective tissue between your existing product and AI capabilities.- **Proactive agent behavior.** Instead of users manually triggering AI tasks, agents are working in the background, analyzing data, generating insights, and automating workflows without explicit prompts. This "always-on" approach drives usage (and costs) much higher.- **Sophisticated context management.** Your agents have access to user-specific data, historical context, and relevant knowledge bases. Building and maintaining these retrieval systems represents a decent chunk of your AI infrastructure costs.Companies operating at these margins are able to outgrow competitors by serving more use cases, delivering faster results and handling complexity that would be impossible with manual workflows or simple AI features. ## **What your margins are telling you** This doesn't mean you should artificially inflate your costs or give away AI for free. The point isn't that lower margins are inherently better. The point is that meaningful AI adoption requires investment, and that investment naturally compresses margins in the short to medium term.If you're sitting at 80% margins on your AI offering, here are the questions you need to ask:- **What's your credit consumption rate?** If users are only burning through 30-40% of their allocated credits each month, your AI isn't integrated deeply enough into their workflows- **Could a competitor replicate your AI in two weeks?** If you're just a thin wrapper around OpenAI's API with basic prompts, you don't have an AI strategy, you have an API subscription with markup.- **What percentage of your engineering team is focused on AI?** If it's less than 20%, you're not building AI-native, you're building traditional software with AI sprinkled on top.- **Are you investing in proprietary AI capabilities?** Custom fine-tuning, specialized evaluation datasets, unique agent orchestration, these create defensibility. Generic API calls don't.The uncomfortable truth is that the companies playing it safe with AI, maintaining high margins and minimal investment, are setting themselves up to be disrupted by competitors willing to compress margins today to build unassailable advantages tomorrow. ## **Your margins will tell the story** 80% margins say you're playing it safe, adding AI as a feature rather than rebuilding your product around it. 40-60% margins say you're making the hard investments that compound into defensibility.The question every SaaS founder needs to answer right now isn't "How do we add AI to our product?" It's "Are we willing to sacrifice short-term margins to build an AI-native product that competitors can't replicate?"The window for making this decision is shorter than you think. In two years, the AI-native companies will have pulled so far ahead that catching up will be nearly impossible. --- ### [2025-10-29] The $2.7 billion agent tax crisis: First-ever study reveals how AI companies are legally avoiding tax - URL: https://paid.ai/blog/ai-agents/the-2-7-billion-agent-tax-crisis-first-ever-study - Date: 2025-10-29 - Author: Arnon Shimoni - Summary: The US faces $2.7B revenue crisis as AI agents legally dodge taxes. First study shows identical AI, different structure across states, with the window of opportunity to legislate closing quickly. Some of you may remember when the then new e-commerce broke state tax systems in the early 2000s - and that was a multi-billion problem that took two decades to fix [as states couldn't figure out a way to tax online sales](https://www.minneapolisfed.org/article/2003/the-taxing-issue-of-ecommerce).I'm about to reveal makes that look like a rounding error - as we just completed the first comprehensive study on AI agent taxation with [Commenda Technologies](https://www.commenda.io/).We analyzed AI agent companies, submitted Public Letter Ruling (PLR) requests to almost all US states, and discovered something that should have state treasurers scheduling followup meetings.**Two AI companies doing identical work. One pays tax in 22 states. The other pays in just 4 states.**The big difference is in how they structure their offering. ## A completely legal arbitrage that most don't know about Here's what our study uncovered: The US tax system was built for a world where humans do work and companies sell products. AI agents shatter every assumption.When companies can choose between being classified as taxable SaaS (affecting millions in tax liability) or non-taxable professional services (paying nothing), rational actors will always choose the latter.- One mortgage automation platform in our study explicitly positions itself as a "fully managed service, not licensed software" with outcome-based pricing. Result: **Tax-exempt in 46 states.**- Their competitor, offering the same capability but structured as software? **Taxable in 22 states.**To be clear - this isn't tax evasion - as it's completely legal and works as intended. It's tax optimization through structural choices that are entirely legal under current law. ## Why this should scare states When an AI agent replaces a $60,000 employee, states don't just lose income tax. They lose everything:- No income tax: $6,000- No payroll tax: $4,500- No state disability/unemployment: $1,500- No local taxes: $6,000- **Total loss per replaced job: $18,000**The US has an estimated 2.5 million administrative workers. If just 10% get replaced by AI agents, that's** $2.7 billion in lost annual revenue.** ## Our key findings Our analysis revealed two critical findings that every AI company needs to understand:**Finding 1: Four questions determine if you're tax liable**The study provides a clear decision framework. Answer these four questions wrong, and you could face millions in unnecessary tax liability. Answer them right, and you could operate tax-free in most states.**Finding 2: The Window Is Closing**Based on state response patterns and legislative timelines, companies have 3-4 years to lock in advantageous tax structures before regulations catch up. ## Why it matters Just like our [partnership with GitLaw on the Agentic MSA](/blog/ai-agents/paid-gitlaw-introducing-legal-contracts-built-for-ai-agents), this report addresses infrastructure the agent economy desperately needs. While GitLaw handles the legal framework for what happens when agents act autonomously, this study reveals the financial framework for how those agents get taxed.Both are about the same thing: **The traditional structures built for software don't work for agents.**You can't use SaaS contracts for autonomous systems (that's why we built the MSA with GitLaw). You can't use SaaS tax treatment for outcome-based services (that's what this study proves). ## What's in the Full Report The complete report includes:- **Detailed case studies** of the companies analyzed- **State-by-state breakdown** of tax treatment and exemptions- **The complete decision tree** for determining your tax liability- **Specific structural recommendations** for each pricing model- **Action items** for companies building todayAdditionally, how outcome-based pricing (Paid's core focus) consistently achieves better tax treatment than any other model. ## The clock is ticking States *are* starting to notice.[Ohio just introduced legislation about AI agents](https://futurism.com/artificial-intelligence/ohio-ban-ai-marriage) - though hilariously, they're worried about robots getting married while missing the actual revenue crisis.By the time legislators stop debating AI personhood and start asking where the tax revenue went, it'll be too late. The companies that act now will have grandfathered structures. Everyone else will face whatever desperate measures cash-strapped states implement.**[Download the Full Report: The $2.7 Billion Agent Tax Crisis, written by Commenda and Paid](https://get.paid.ai/hubfs/The%20$2.7%20Billion%20Agent%20Tax%20Crisis.pdf)**> *"This report shows the tax crisis is already here. When an AI agent replaces a $100,000 junior lawyer, the state loses thousands income tax overnight. The agent doing that same work pays nothing in some cases. Multiply that across administrative roles, paralegals, analysts, and you're looking at billions in lost revenue before anyone figures out how to replace it.**We have a short window before this becomes too big to fix. Right now, companies can structure their agents as taxable software or non-taxable services, and that choice determines whether they pay in dozens of states or just a handful - that's that's arbitrage. We need to establish clear frameworks now, or states will panic and create a patchwork of desperate policies that kill innovation."****Manny Medina, CEO of Paid****This study was conducted in partnership with Commenda Technologies Inc., led by Sachhin Kunjalwar, Head of Tax Content and Arnon Shimoni from Paid.**For specific tax guidance and Private Letter Ruling assistance, **[head over to Commenda](https://www.commenda.io/)**.**For billing infrastructure designed for the agent economy's unique tax requirements, visit **[Paid.ai](https://paid.ai)**.* --- ### [2025-10-26] Notes on where seat-based pricing is going - URL: https://paid.ai/blog/ai-monetization/notes-on-where-seat-based-pricing-is-going - Date: 2025-10-25 - Author: Arnon Shimoni - Summary: Seat pricing isn’t evolving, it’s being outgrown. As AI replaces users with agents, software must price by work done, not by humans doing it. Everyone (mostly us) keeps saying seat pricing is dead, and some people are arguing with us online.I think that's cool, but it's because they don’t understand *why* it’s dying.For twenty years, SaaS revenue scaled with headcount.*More humans = more seats.*That logic held so long that we stopped questioning it.Then the humans disappeared.**So seat pricing didn’t just die, its margins migrated. The value once captured in human licenses now lives in compute bills.**But let's understand why: ## The world seat pricing was built for Seat-based pricing made perfect sense in the era when software automated *people*. A CRM helped a sales rep. A design tool helped a designer. A support platform helped an agent.You could literally count your customers by the number of logins.Here, I'll put it in a table:EraUnit of workUnit of valuePricing proxySaaS 2005–2020Human performs workflowProductivity per humanPer-seat licenseThat means growth looked like this:more employees → more seats → more ARR.I definitely experienced this first-hand at the SaaS companies I worked at, and it was beautifully linear and predictable.Dropbox's public data shows it pretty well too.And for a long time, that linearity made everyone lazy. ## The first cracks Seat pricing broke quietly, and then all at once.Three structural shifts made it inevitable:ShiftWhat changedWhy seats fail**Agents replace users**Work is done through APIs and automationAgents don’t log in and don't use a seat**Teams get smaller**Ten people with agents can do the work of a hundredHeadcount stops scaling with output, and not everyone needs these tools**AI features eat their own revenue**Every “productivity” feature removes usage minutesSeat expansion becomes self-defeatingSeat pricing depends on more humans doing more work. AI depends on *fewer* humans doing *less* work.That’s not a market correction. It’s a phase change. ## Hybrid pricing is just a pause button Hybrid pricing (seats + usage + minimums) is a holding pattern. Companies add "usage components" not because it’s the future, but because it helps them stay afloat while their core metric stops working.Hybrid pricing is what disruption looks like from the inside - and you can see it in product telemetry across the industry:- Seat count per customer is flat or shrinking.- Compute cost per customer is rising.- AI adoption cuts user logins but increases workload volume. ## How we got trapped by our own success Seat pricing worked so well it trained an entire generation of SaaS companies to chase adoption metrics that don’t matter anymore.User logins.Seats activated.Seat expansion revenue.Those signals once indicated health.Now they measure inertia.When customers replace 50 support agents with one orchestrator running 50 AI assistants, seat metrics collapse — even though the *system’s output* increases 10×.The problem isn’t that usage-based pricing is new.The problem is that SaaS never had to think about *what it’s actually selling*. ## The new unit of value The next pricing substrate is not per-seat, but per-work.We’re already seeing three archetypes emerge:ModelExampleUnit of valueUsage-basedAPI calls, tokens, compute minutesWorkload volumeOutcome-basedLeads verified, tickets resolvedBusiness resultAgent-basedCost per autonomous agent / monthSynthetic laborThe line running through it all is clear (at least to some of our partners): value tracks work done, not humans doing it.Unfortunately, this *will* feel messy for a few yearsJust like AWS billing did in 2008: complex, unpredictable, but a lot more aligned with the actual source of value. ## 2025 or Q1 2026 is the time to act The SaaS economy grew up during an era of abundance with cheap capital, expanding headcount, endless GTM motion.Seat pricing thrived because companies were hiring faster than they were automating, but 2021-2022 is over.AI agents don’t take vacations, and they don’t show up in HRIS exports (not yet, anyway).They don’t need training seats, onboarding credits, or license renewals.They just execute.Seat pricing is a tax on humans. And in the next decade, humans stop being the primary unit of work. ## What replaces the seat We’re heading toward *agentic billing models* that look a lot like infrastructure pricing:- Pay for outcomes, not headcount.- Bill for concurrency, not users.- Measure success in work completed per dollar spent.Yes, it'll start rather messy with the tokens/credits, hybrid models and caps - but this will converge toward *work-per-unit-time and eventually outcomes*.When that happens, pricing becomes both simpler and truer:Software bills for what it delivers, not who touches it or how much inputs it used. ## What this means for SaaS founders (and investors) - If your growth depends on seat expansion, you’re already in decline.Expansion revenue now fights automation, not benefits from it.- If your pricing isn’t margin-aware under AI load, it will implode.Compute costs eat seats for breakfast.- If your product’s AI layer doesn’t come with a billing and monetization rethink, it’s just decorationYou're not disrupting your own business enough, and someone else will. ## The new denominator Every few decades, software changes its unit of measurement.EraUnit of measureDesktopLicenseCloudSeatAI / AgenticWork (and later value)The tl;dr is seat pricing isn’t dying from lack of innovation - it’s dying from irrelevance.You can fight it, but when the average company has more agents than employees,the only question left is simple:*What’s your new denominator for software value?* --- ### [2025-10-24] Usage-based pricing for SaaS: What it is and how AI agents are breaking it - URL: https://paid.ai/blog/ai-monetization/usage-based-pricing-for-saas-what-it-is-and-how-ai-agents-are-breaking-it - Date: 2025-10-24 - Author: Arnon Shimoni - Summary: Usage-based pricing built AWS and Twilio. For AI agents, it creates 100x cost variance and negative margins. Here's why the SaaS playbook fails for AI. By now everyone knows usage-based pricing charges customers based on product consumption instead of flat subscription fees. In traditional SaaS, you paid per API call, per gigabyte stored, or per message sent.For AI agents, this model is creating a margin crisis, and that's why usage-based is wrong for AI and specifically AI agents. ## What is usage-based pricing? **Definition:** A pricing model where costs scale directly with measurable product usage like API calls, compute hours, or data processed.You pay for what you use. Nothing more, nothing less. ## Usage-based pricing examples in SaaS The model powered some of the biggest SaaS companies:- **[Twilio](https://www.twilio.com/en-us/pricing)****:** $0.0083 per SMS, $0.0085 per voice minute- **AWS:** $0.023 per GB of S3 storage- **SendGrid:** changes, but around $0.0006 per email- **[Stripe](https://stripe.com/pricing)****:** 2.9% + $0.30 per successful transactionThese work because costs stay relatively predictable. An API call costs roughly the same whether it's your first or your millionth (assuming you didn't pre-buy a package with overage fees). ## Usage-based vs subscription pricing FactorSubscriptionUsage-BasedBill predictabilityAs predictable as you can getVariableCustomer riskYou pay regardless of useUsage can exceed what is plannedRevenue predictabilityGuaranteedVariableExpansion potentialLimitedUnlimitedCFO preferencePreferred (makes forecasting easy)Cautious (harder to forecast on)- **Subscription:** $50/month whether you use it once or 1,000 times- **Usage-based:** $0.01 per API call (100 calls = $1, 10,000 calls = $100)- **Hybrid:** $20/month base + $0.01 per API call over 1,000 ## Why AI agents break usage-based pricing ## 1. Cost variance destroys margins **Traditional SaaS:** If an API call costs $0.001 and you charge $0.005, your margin holds at 80%.compare with **AI agents:** A "resolved ticket" costs anywhere from $0.99 to $3.00.One customer service platform charges $0.99 per resolved ticket. Simple questions cost them $0.04 in resources. Complex issues cost $2.80. Average margin is ~60%, but engaged customers generate losses. This variance isn't a bug. Agents assess complexity and adapt. While simple queries need one LLM call, complex ones trigger research, use memory, multiple reasoning steps, and generate more detailed responses. It's quite hard to price for average when the range is so big.Manny Medina (our CEO who built Outreach to a $4.4B valuation) sees this playing out across the industry: ## 2. Autonomous decisions break attribution Traditional software waited for instructions. Click send, email goes out, charge applies.AI agents make independent decisions. An AI SDR might send five touchpoints instead of two, adjusting based on engagement. Do you charge for two emails or five? For the outcome or the process? For successful actions or every attempt?[Cursor](https://www.cursor.com/pricing) faces this daily. When their AI generates code, refactors it, then fixes a bug, how many "completions" was that? Developers wanted working auth. They don't know if that costs 1,000 tokens or 10,000.The whole metering approach of "count the things, charge per thing" breaks down here, as one outcome could have many usage events. ## 3. Token pricing is a proxy for something else, obfuscates the value Charging for tokens maps directly to infrastructure costs and obfuscates the value. Anthropic does it. OpenAI does it, but customers think in outcomes, not tokens - which is why so many people complain about their bills."Implement user authentication" is clear. Estimating 8,347 input tokens and 2,156 output tokens requires expertise most don't have.This creates two BIG problems:- **Customers can't forecast.** Finance needs predictability. Token consumption varies in non-intuitive ways.- **Customers self-police (optimize for saving, not using).** Unpredictable bills kill usage. This destroys expansion and prevents full value realization. An AI SDR that books fewer meetings but uses fewer tokens isn't more valuable. ## 4. The race to zero accelerates AI model costs drop 90% annually while better models launch constantly.When you price based on tokens or compute, your pricing tracks the collapsing cost structure. As GPT-4 gets cheaper, customers expect matching price drops.One document processing company started at $0.50 per page. When costs dropped and competitors entered at $0.30, then $0.15, they had no defense.Companies charging for outcomes maintained pricing power. A reviewed contract is worth the same whether processing costs $2 or $0.20. ## The illusion of simplicity Usage-based pricing feels simple in theory - you count, charge, and you're done.But a customer service agent doesn't just "resolve tickets." It searches your knowledge base, generates responses, checks policies, formats output, and learns from interactions.Five usage events for one outcome. Which do you charge for? All of them? How do you explain that? ## The measurement gap Usage-based pricing assumes you can measure usage accurately and cheaply.For traditional SaaS, counting API calls is trivial. For AI agents, you need to track LLM calls across providers, attribute costs to workflows, monitor tokens in real time, calculate per-interaction margins, and catch cost spikes before they kill profitability.Stripe counts transactions but doesn't know workflow costs. Chargebee generates invoices but can't tell you which customers are profitable.The infrastructure gap means flying blind while thinking you're data-driven because you're "usage-based." ## Why the SaaS playbook fails Traditional SaaS had simple economics:- Development costs upfront- Marginal costs approach zero- More users = more profit- Predictable usage patternsAI agents invert everything:- Development costs ongoing- Marginal costs scale with complexity- More usage *might* mean less profit- Usage patterns are autonomousAt this point, you might be thinking: "Fine, usage-based is broken. Seats are dying. What's actually supposed to work?"...We have a surprisingly simple answer:> "You don't need seats. You don't need a ton of thinking. The moment you establish what the exchange value is going to be, then you're in good shape to price."This sounds almost too simple. But it cuts through the complexity: figure out the value you're creating, take 20-50% of that, and build a system that lets you price differently for each customer. Not a perfect formula. Not a one-size-fits-all model. A framework for thinking about value exchange.[Martin Casado and Kyle Poyar recently wrote about why all pricing models have flaws](https://www.growthunhinged.com/p/your-pricing-is-broken). While some of the criticism is fair - not everything requires a PhD or is a pipe-dream. Where they are right is that usage-based pricing for AI is misaligned with the economics.Usage-based pricing worked in SaaS because it aligned incentives. Customers paid for what was performed, and vendors captured it.For AI agents, it creates misalignment:- Customers want outcomes but pay for consumption- Vendors need margin predictability but face 100x variance- Both unfortunately optimize for wrong metricsOur [SaaS to AI Agent transition report](/blog/ai-monetization/the-saas-agent-transition-report-from-vibe-pricing-to-value-capture) analyzed 250+ companies - from pure usage-based pricing correlates with 70% churn and negative margin to companies evolving to workflow or outcome models that maintain 94% margins.At this point, you have to recognize that usage-based pricing optimizes for infrastructure efficiency when the world demands outcome delivery. --- ### [2025-10-16] The SaaS Agent Transition Report: From Vibe Pricing to Value Capture - URL: https://paid.ai/blog/ai-monetization/the-saas-agent-transition-report-from-vibe-pricing-to-value-capture - Date: 2025-10-22 - Author: Arnon Shimoni - Summary: 75% of SaaS companies lack systematic agent pricing. Those moving from usage-based to outcome-based models achieve 94% margins vs. negative margins for competitors. *This report is based on analysis of 250+ conversations with SaaS companies and agent builders navigating the AI transition* ## **tl;dr:** SaaS companies adding agent features face a critical monetization challenge. After analyzing 250+ conversations with companies ranging from early-stage startups to enterprises, a clear pattern emerges: 75% of companies building agents have no systematic approach to pricing them.This creates what we call "vibe revenue". Strong initial adoption followed by 70% churn rates.There is a small subset of companies are thriving, who have moved beyond usage-based pricing to value-aligned models that defend against commoditization and build sustainable revenue streams.This report outlines their playbook. ## Key Findings **State of the market**- 75% of agent-building companies admit uncertainty about pricing strategy- 70% churn rates observed in certain agent segments- 100x cost variance between simple and complex agent workflows- Pricing compression of 90% in competitive categories within 12 months**Distribution of pricing models**- 45% usage-based pricing (most vulnerable to commoditization)- 20% per-agent pricing (FTE replacement positioning)- 15% workflow-based pricing (middle ground)- 20% outcome-based pricing (highest margins, lowest churn)**Success indicators** Companies with outcome-based elements achieve:- 94% gross margins vs. sometimes negative margins for pure usage-based- 8.3x value delivery vs. price charged- Significantly lower churn through value alignment ## 1. The Vibe Pricing Problem ## Symptom: Guesswork driving strategy Representative quotes from our interviews:> *"We are closing our first enterprise client but we just kind of made a ballpark estimate on price."* — Sustainability compliance platform> *"To be perfectly honest, pricing is a mess. It's all over the place."* — AI SDR platform> *"We entered the market at $1,500 as a base price. We were having lots of positive conversations until we got to price. So we went to $750 and were still getting eyebrows raised. So we went down to $500."* — AI SDR platformThis informal, intuition-driven approach creates three problems:- **No grounded assessment of value**- **No understanding of unit costs**- **No differentiation strategy** ## The Commoditization Spiral When supply explodes, vibe pricing accelerates race-to-bottom dynamics:- Customer service agents: 30+ competitors per deal- AI SDRs: 90% pricing compression in 12 months- Document processing: Commodity pricing within 6 months of launchThis means that your initial explosive ARR has now stalled. ## The Margin Blindspot Traditional SaaS had predictable costs, but agent features introduce radical variability.ScenarioCost per InteractionSimple task$0.02Complex reasoning$2.00**Variance****100x**One company discovered their AI SDR sent emails at negative 22% margin. Every email lost money.> *"One of the biggest challenges I'm seeing is not only figuring out how to monetize properly, but also how to manage the resources they're using."* — Small business operations platform ## 2. The emerging AI pricing archetypes Based on our analysis, the market has converged on four approaches:ModelMarket ShareDefensibilityMargin ProfileChurn RiskUsage-based45%LowVariableHighPer-agent20%MediumMediumMediumWorkflow-based15%Medium-HighGoodLow-MediumOutcome-based20%HighExcellentLow ## Model Breakdown ## 1. Usage-Based: The Commodity Trap **Structure:** $0.10 per email, $0.20 per call, $0.05 per data point**Advantages:**- Simple to implement- Easy customer understanding- Low barrier to entry**Vulnerabilities:**- When frontier model costs drop 90% annually, pricing must follow- No differentiation from competitors- Constant price pressure- Customers shop on cost alone ## 2. Per-Agent: The FTE Replacement Play **Structure:** $2,000/month for an AI SDR vs. $5,000 for human equivalent**Advantages:**- Taps into larger labor budgets- Clear value comparison- Predictable revenue**Vulnerabilities:**- As agent creation becomes effortless, why pay per agent?- Replicable by competitors- Price anchored to declining model costs ## 3. Workflow-Based: The Middle Ground **Structure:** $10 for completed multi-step process (research, qualify, compose, send, follow-up)**Advantages:**- Higher switching costs than pure usage- Captures more value than individual actions- More defensible positioning**Considerations:**- Requires clear workflow definition- Attribution complexity increases ## 4. Outcome-Based: Maximum Value Alignment **Structure:** $200 per qualified meeting, $0.99 per ticket resolved, $500 per contract processed**Advantages:**- Perfect value alignment- Highest margins observed- Highly defensible- Customers compare results, not prices**Challenges:**- Requires attribution methodology- Quality definition complexity- Higher implementation barrier ## 3. The DRIVE Framework: Evolution path to value-aligned pricing Successful companies don't jump to outcomes.We've identified a path through five phases: ## D - Demonstrate proof Start with low-risk, high-visibility tasks:- Pick easy wins- Deliver 100% completion- Show every action transparently- Earn permission for complex work*"We started with simple lead enrichment before moving to full sales automation."* — CRM platform that grew to 2,000 seats in 90 days ## R - Record everything Track obsessively:- Every agent action and outcome- All compute, API, and token costs- Margin per customer, per outcome- Value delivered vs. cost incurred> *"Our AI SDR had negative 22% margin per email until we implemented proper tracking. Now we maintain 67% margins through dynamic optimization."* ## I - Integrate deeply Create stickiness through:- Embedding into existing workflows- Connecting with tools customers use daily- Making switching painful- Becoming mission-critical infrastructure ## V - Value in human terms Every invoice should communicate something like this:- Tasks completed: 5,000- Human equivalent hours: 625- Human equivalent value: $25,000- Your price: $3,000- Value delivered: 8.3xThis anchors discussions to $60,000 salaries, not $49 software subscriptions.> *"Showing the work gets people really excited because they feel like they're getting more than what they actually pay for."* — CEO of a Cybersecurity platformPaid's value receipts are a great example of this. ## E - Experiment toward outcomes You should move towards outcomes as much as possible. Your evolution timeline:- Month 1: Usage-based (build trust)- Month 2: Add success bonuses (share upside)- Month 3-5: Hybrid model (base plus outcomes)- Month 6+: Pure outcomes (maximum value capture)> *"The pricing we launch is going to be something basic just credits because we need to get this thing out and test it. In later workstreams we'll evaluate pricing in a more nuanced way."* — Founder of a Sales enablement platform ## 4. Industry-Specific Patterns ## RevTech (AI SDRs and AI AEs / Marketing Agents) **Current State:**- 60% consumption-based- 20% outcome-based- 20% workflow-based**Winner Profile:** Companies charging per qualified meeting or per opportunity created. One AI SDR company charges $200 per attended meeting with 94% gross margins. ## Customer service **Current State:** Moving fastest to outcomes**Examples:**- Intercom: $0.99 per resolved ticket- Zendesk: Following with similar model**Winner Profile:** Resolution-based pricing with quality guarantees ## Developer tools **Current State:** Most stuck in seat-based models, some moving to usage-based "copilot" pricing**Vulnerability:** Highly susceptible to disruption due to low switching costs**Winner Profile (Emerging):** Companies charging per successful deployment or per bug fixed. Still rare. ## Common success factors Winners across verticals share three characteristics:- Track unit economics obsessively- Communicate value in human terms- Evolve pricing every 3-4 months based on data ## 5. The attribution challenge Moving to outcome-based pricing introduces four common objections: ## Challenge 1: Causation arguments > *"The AI didn't cause that meeting."***Paid recommends:** Define attribution windows and methodology upfront. Last-touch, 30-day windows work for most B2B scenarios. ## Challenge 2: Quality disputes > *"That wasn't a qualified lead."***Paid recommends:** Establish mutual quality criteria before deployment. Document in contract. ## Challenge 3: Value variance > *"An SDR meeting isn't worth the same as a C-suite meeting."***Paid recommends:** Create tiered outcome pricing based on lead quality or seniority. ## Challenge 4: Delivery risk > *"Too much responsibility shifted to vendor."***Paid recommends:** Hybrid models that balance risk. ## The hybrid approach that works One AI SDR company's working model that we helped them develop included:- Base fee: $500/month (covers vendor costs)- Success fee: $50 per qualified meeting (aligns value)- Quality criteria: Mutually defined upfront- Attribution window: Last-touch, 30 days> *"We moved back from pure outcomes to hybrid. It balances risk while maintaining alignment."* ## Alternative approach: outcome bundles Instead of pure variable pricing:- 1,000 resolutions for $900 (vs. $0.99 each)- Unused credits roll over- Annual commits for deeper discountThis gives enterprise customers predictability while maintaining value alignment. ## 6. Strategic recommendations These are specifically for SaaS companies adding agentic features**Immediate actions (Month 1 - do this now):**- Implement comprehensive tracking of agent actions and costs- Calculate true unit economics per agent workflow- Document value delivered in human-equivalent terms- Establish baseline pricing using current model**Short-term evolution (Months 2-3):**- Add success-based bonuses to existing pricing- Test value communication in customer conversations- Identify 2-3 pilot customers for outcome-based trials- Build attribution methodology for key outcomes**Medium-term transition (Months 4-6):**- Launch hybrid pricing for new customers- Migrate existing customers to new model- Document and share success stories- Refine attribution based on customer feedback**Long-term position (Months 6+):**- Move qualified segments to pure outcome pricing- Maintain hybrid option for risk-averse buyers- Iterate pricing every quarter based on data- Build defensible moats through integration depth ## ## Conclusion The underlying AI models will commoditize. OpenAI, Anthropic, and others will drive costs down 90% or more, but new models will come by - your differentiation cannot come from the models themselves.The SaaS companies that capture value will be those that master the journey from usage to outcomes faster than competitors.The path is clear:- Start where you are- Track everything- Show value in human terms- Experiment toward outcomes- Iterate faster than competitorsWe believe this transition is inevitable - but you control the pace in which you adopt it. ## About This Research This report synthesizes findings from 250+ conversations with SaaS companies and agent builders conducted between January and October 2025. Participants ranged from early-stage startups to public companies with 60,000+ employees, across verticals including RevTech, customer service, developer tools, compliance, and business operations.Paid helps SaaS companies understand agent economics, track value delivery, and implement outcome-aligned monetization models. --- ### [2025-10-17] AI Billing for Sales Development: How to Price and Monetize AI SDR Platforms - URL: https://paid.ai/blog/ai-monetization/ai-billing-for-sales-development-how-to-price-and-monetize-ai-sdr-platforms - Date: 2025-10-21 - Author: Arnon Shimoni - Summary: AI SDR billing guide: How sales development platforms track agent outcomes, measure ROI per meeting booked, and move from seat-based to performance pricing. Sales teams are replacing BDRs with AI agents - specifically AI SDRs, which have cropped up like mushrooms in fall.These AI SDR agents prospect leads, write outreach sequences, book meetings, and manage entire outbound pipelines. The technology works.The challenge isn't so much in building the AI SDR, but rather charging for work that never stops.When you're selling AI-powered sales development tools, "our agent sends emails" doesn't justify enterprise pricing. Sales leaders want pipeline metrics. CFOs need cost-per-meeting numbers. Revenue teams want to know if the AI outperforms human BDRs. You need to prove value in terms sales organizations already measure.This creates a billing problem traditional sales software never faced. ## Why standard pricing fails for AI SDRs Most sales development platforms use seat-based pricing inherited from CRM and email automation tools. You charge per user or per sales rep accessing the platform. This made sense when software waited for humans to click send.AI SDRs don't work that way.Your agent might research 500 prospects per day. It writes personalized outreach, manages follow-up sequences, responds to replies, and books meetings automatically. The work happens continuously. A single "seat" can handle the workload of an entire BDR team.But your billing system treats all of this like a $100/month user license.This disconnect creates three problems:**Massive underpricing**: When your AI SDR books 50 qualified meetings per month, charging $2,000 feels absurd compared to the $60,000 annual cost of hiring a human BDR who books 20 meetings.**Unclear value**: Your customers can't see what the agent accomplishes. Without measurement, they compare your pricing to traditional email automation tools. "Why pay $10,000 when Outreach costs $1,000?" becomes a deal killer.**Broken unit economics**: Running AI SDRs costs real money. LLM calls for personalization, data enrichment APIs, email infrastructure. If you don't track costs per customer, your best customers subsidize your worst. You scale revenue while margins collapse.Standard billing platforms can't solve these problems because they weren't built for products that replace entire job functions. ## What AI SDR platforms actually need Sales development companies moving to AI-powered prospecting need three capabilities their current billing stack doesn't provide:**Outcome tracking**: Measure business results, not email volume. Track meetings booked, qualified conversations started, pipeline generated. The metrics sales leaders report to their board.**Value attribution**: Connect agent actions to revenue impact. When your AI books a meeting that turns into a $50,000 deal, you can calculate precise ROI per interaction. When it sends 1,000 emails that generate zero pipeline, you know which campaigns need optimization.**Flexible pricing models**: Move beyond seats to performance-based models. Charge per meeting booked. Take a percentage of influenced pipeline. Combine base fees with success bonuses. Models that align your pricing with customer success. ## How modern AI SDR platforms measure and monetize value Companies building AI-powered sales development tools are solving this with outcome-based infrastructure instead of usage-based billing.Here's how it works:**Track what matters to sales leaders**: Instead of counting emails sent or API calls made, track completion of sales workflows. Meetings booked with qualified prospects. Opportunities created. Pipeline influenced. Response rates achieved. The metrics sales teams already report to leadership.**Connect to existing CRM data**: The best measurement strategies don't require new systems. They plug into data you already have. If prospects are in Salesforce and meetings sync to calendars, you're already tracking outcomes. Connect that to agent activity and ROI calculation becomes straightforward.**Show value continuously**: Don't wait for monthly business reviews to prove impact. Embed live dashboards showing agent performance, meetings booked, response rates. Make the invisible prospecting work visible in terms sales teams understand. ## A real implementation: AI SDR outcome-based pricing One well-known AI SDR built an agent that handles the entire outbound function. The agent finds leads, enriches contact data, writes personalized sequences, and books meetings with qualified prospects.While they did use seat-based pricing initially, their agent didn't occupy a seat for real - it works continuously and handles workload equivalent to multiple human BDRs.Eventually they realized they needed to:- Track every action the agent performs for each customer- Bill based on outcomes delivered, not seats purchased- Monitor margins in real time as agent performance improves- Automate the entire billing process without custom code> "Customers see exactly what the agent delivers. Transparent value tracking makes renewals straightforward. When you can show 200 qualified meetings booked, the ROI justifies renewal."*- AI SDR customer*Legacy billing platforms couldn't do this. In particular, they were built for software subscriptions, not autonomous agents.The AI SDR vendor integrated Paid's AI-native billing platform to handle four critical pieces:- **Automated billing**: No manual invoicing. No spreadsheets. Invoices generate automatically based on agent activity and outcomes achieved.- **Outcome-based pricing**: They charge for results, not access. The system tracks meetings booked, leads qualified, and pipeline influenced. Pricing scales with performance.- **Agent action tracking**: Every action the agent takes gets logged. They know exactly what value each customer receives. Which sequences work. Which prospects engage. Which meetings convert.- **Customer-level margins**: They see profitability per customer in real time. They optimize pricing and resource allocation on the fly. Customers with high meeting conversion rates get different pricing than those with poor targeting.The results changed their business:**Retention improved**: Customers see exactly what the agent delivers. Transparent value tracking makes renewals straightforward. When you can show 200 qualified meetings booked, the ROI justifies renewal.**Pricing became defensible**: When a single agent books 50 meetings per month at $250 per meeting, the $12,500 monthly cost makes sense. Compare that to a $5,000/month human BDR who books 20 meetings.**Operations simplified**: As agent performance improves, they capture more value automatically. Better outcomes mean higher revenue without changing the pricing model. No contract renegotiations needed. ## Three pricing models that work for AI SDRs Traditional seat-based pricing doesn't match how sales development AI actually delivers value. Three models work better:ModelStructureBest ForCustomer PredictabilityRevenue CapturePer-outcome$200 per meeting bookedHigh-volume, proven targetingLow (varies with performance)Highest (scales with value)Hybrid$3k base + $200/meetingMid-market, steady pipelineMedium (known minimum)BalancedPipeline sharing3-5% of influenced pipelineEnterprise, long sales cyclesLow (depends on close rates)Variable (high upside)**With per-outcome pricing**, you charge per meeting booked ($100-500 depending on industry and deal size), per qualified conversation started, or per opportunity created. Customers only pay for results.**With the hybrid model**, you charge a ase platform fee ($3,000/month) plus performance fees ($200 per meeting booked). Provides predictable minimum revenue while scaling with customer success.**With pipeline sharing**, you agree with your customer to charge a percentage of influenced pipeline (3-5%) with monthly minimums and caps. Aligns incentives completely but requires deeper CRM integration and attribution tracking across the flow.The best pricing strategy depends on what you can measure reliably and what your customers already track. If your platform integrates with their CRM, pipeline-based pricing becomes feasible. If not, start with per-meeting pricing and evolve as your measurement improves. ## Why AI SDRs break traditional billing Here's what makes AI sales development different from traditional sales software:FactorHuman BDRAI SDRBilling ImplicationWork schedule40 hrs/week, PTO, sick days24/7/365 continuousSeat pricing undervalues by 4xCost structureSalary, benefits, equipmentAPI calls, compute, data enrichmentVariable costs require margins trackingDecision makingTakes direction, needs managementAutonomous targeting and sequencingPricing should reflect job replacementPerformance curveFlat after trainingImproves continuously with dataPricing should capture improvement- **Continuous operation**: Human BDRs work 40 hours per week. AI SDRs work 168 hours per week. Seat-based pricing treats them the same.- **Variable costs**: Every personalized email costs money. Every data enrichment API call costs money. Every LLM interaction costs money. These costs vary wildly per customer based on targeting, response rates, and sequence complexity. While the LLMs do drop in cost, you want to stay on the frontier models which tends to actually get more expensive.- **Autonomous decisions**: The agent decides which prospects to target, when to send follow-ups, how to respond to replies. It's not executing human instructions. It's performing a job function.- **Performance improvement**: As the AI learns, it books more meetings with the same inputs. Traditional software doesn't get better at its job over time (barring extreme updates). AI agents do, regularly. Your pricing should capture that improvement as it happens and not years later when you renew the contract.Most billing platforms used by sales technology companies were built for traditional SaaS. They handle seats, usage tiers, and subscription management well.They can't track what an AI SDR does. They can't measure the value a prospecting workflow delivers. They can't price based on outcomes because they weren't designed for products that replace entire teams.Companies building AI-powered sales tools need different, AI-native infrastructure like Paid, with:- **Telemetry-based tracking**: Capture business events (meetings booked, opportunities created) not just technical events (emails sent, API calls made)- **Cost attribution**: Understand what each customer interaction actually costs to run, so you can price with healthy margins even as agent usage scales- **Value visualization**: Show customers what their agent accomplished in metrics they already report to their sales leadership- **Flexible billing**: Support outcome-based pricing, performance fees, and revenue sharing that traditional platforms can't handlePaid's agentic approach adapts to agent's changes over time, and supports showing value, attributing all costs, and flexible billing that can change as fast as your agent does. ## What to do next If you're building AI-powered sales development technology, you face a measurement challenge that becomes a pricing challenge that becomes a growth challenge.Your agent delivers real value. You need to prove it, price for it, and show it continuously to customers.These are the steps you must take if you're building an AI SDR:- **Start measuring outcomes now**: Don't wait until you have perfect attribution. Begin tracking the signals that matter to sales leaders. Meetings booked, response rates, opportunities created. Directional data beats no data.- **Test outcome-based pricing**: Pick your best customer and try performance-based fees. Charge per meeting booked or per qualified opportunity. See how they respond. Refine based on feedback. You don't need to change all contracts at once.- **Connect to existing systems**: The best ROI calculations use data customers already trust. If meetings sync to their CRM, track attribution there. If they measure pipeline in Salesforce, connect your agent's activity to their opportunity data.The AI SDR companies that figure out how to measure and monetize outcomes will dominate their categories. The ones that stick with seat-based pricing will struggle to justify their value as costs scale with usage.Your agent works. Now it's your job to make sure you can prove it and charge for it (and Paid is here to help make that happen)**Building AI for sales development?** Learn how AI-native billing infrastructure helps you track costs, prove ROI, and move to outcome-based pricing. Start with free cost tracking or book a demo to discuss your specific needs. --- ### [2025-10-17] AI Billing for HR Tech: How to Price and Measure Compensation Intelligence Platforms - URL: https://paid.ai/blog/ai-monetization/ai-billing-for-hr-tech-how-to-price - Date: 2025-10-17 - Author: Arnon Shimoni - Summary: How HR tech companies price AI agents that analyze compensation, ensure pay equity, and automate hiring decisions. Move beyond seat-based billing. Modern AI-powered HR technology companies are embedding more than just AI - but rather AI agents into compensation workflows. These agents analyze pay equity, generate offer recommendations, and help companies make smarter hiring decisions. The technology works.The challenge isn't just in the tech, but in how it delivers the value and how that's proven to customers.When you're selling AI-powered compensation tools to enterprises, "our platform saves time" doesn't close deals anymore.Our experience with CFOs is that they want ROI numbers. HR leaders need budget justification, and of course Legal and Compliance teams want compliance accountability. You need to show value in terms your customers already measure.This creates a billing problem traditional HR software never faced. ## Why standard billing fails for HR AI agents Most HR tech companies still use seat-based pricing models inherited from legacy HRIS systems. You charge per user or per employee in the system. This made sense when software sat idle until someone clicked a button.AI compensation agents don't work that way.Your agent might analyze 50 candidates per day for a single recruiter. It checks market data, reviews internal pay bands, flags equity issues, and generates recommendations continuously. The work happens automatically. The value compounds.But your billing system treats all of this like a single seat license.This disconnect creates three problems:- **Underpricing**: When your agent helps a company avoid $500,000 in pay equity adjustments, charging $10,000 per year feels misaligned with the value delivered.- **Poor visibility**: Your customers can't see what the agent does. Without measurement, they question renewal. "What did we get for this?" becomes a deal killer.- **Margin risk**: Running AI agents costs real money. LLM calls, data processing, API usage. If you don't track costs per customer, profitable accounts subsidize unprofitable ones. You scale revenue while eroding margins.Standard billing platforms can't solve these problems because they weren't built for products that deliver autonomous outcomes. ## What AI-powered compensation platforms actually need HR tech companies moving to AI-based compensation intelligence need three capabilities their current billing stack doesn't provide:- **Outcome tracking**: Measure business results, not API calls. Track offers generated, equity gaps prevented, time saved per hiring decision. The metrics HR leaders actually care about.- **Value attribution**: Connect agent actions to measurable impact. When your pay equity analysis shows a customer saved $300,000, and they used your AI for 1,000 compensation decisions, you can calculate precise ROI per interaction.- **Flexible pricing models**: Move beyond seats to hybrid models. Base platform fee plus outcome fees. Subscriptions combined with success-based pricing. Revenue sharing on documented cost savings. Models that align your pricing with customer value. ## How modern HR AI platforms measure and monetize value Companies building AI-powered compensation tools are solving this with signal-based infrastructure instead of usage-based billing.Here's how it works:- **Track what matters to HR leaders**: Instead of counting tokens or API calls, track completion of compensation workflows. Offers generated. Equity analyses completed. Compliance checks passed. Time saved per decision. The metrics HR teams already report to leadership.- **Connect to existing HR data**: The best measurement strategies don't require new systems. They plug into data you already have. If you run pay equity analytics, you're already measuring cost avoidance. Connect that to agent usage and ROI calculation becomes straightforward.- **Show value continuously**: Don't wait for quarterly business reviews to prove impact. Embed live dashboards showing agent activity, time saved, compliance maintained. Make the invisible work visible in terms customers understand.Traditional HR metricsHow they translate to agentic signalsBusiness impact to demonstrate ("ROI")Time to hireOffers generated per dayFaster hiring cyclesPay equity complianceEquity gaps preventedReduced legal riskOffer acceptance rateRecommendations acceptedBetter talent retentionManual analysis timeHours saved per decisionLower operational costs ## A real implementation: Compensation AI billing company One compensation intelligence platform embedded an AI agent that helps recruiters make offer decisions. The agent analyzes candidate profiles, checks internal pay ranges, reviews market data, ensures equity compliance, and generates recommendations in real time.They integrated an AI billing platform using OpenTelemetry. The setup took days. The system automatically captured traces from the agent's workflows.Here's what they tracked:**Efficiency metrics**: Time to generate offer, number of scenarios analyzed, decisions per recruiter per day**Cost avoidance metrics**: Pay equity gaps prevented, compliance violations avoided, offer rejections reduced**Trust indicators**: Recommendations accepted vs. modified, time spent reviewing suggestionsThe classification engine tagged about 70-80% of signals automatically. Their team refined the rest, mapping specific workflows to business outcomes their customers measured.The key innovation we've seen was connecting the AI agent's value to their existing pay equity product data - when their analytics showed a customer saved $500,000 in pay inflation, and that customer used the AI for 1,000 decisions, they could attribute value directly. ROI became measurable in dollars, not engagement scores.Three things changed:- **Customer conversations got much easier**: Instead of explaining how the agent works, they showed what it accomplished. Dashboards displayed hours saved, decisions accelerated, compliance maintained. Customers saw value in familiar HR metrics.- **Pricing became defensible**: When you show a customer they avoided $500,000 in pay equity adjustments, charging $50,000 becomes reasonable. When you only show 10,000 API calls, any price feels expensive.- **Product priorities were much easier to understand**: By tracking which workflows delivered the most value, they focused engineering on features that mattered. They could see which agent behaviors HR teams trusted and which needed refinement.- ## Three pricing models that work for HR AI Traditional seat-based pricing doesn't match how compensation AI actually delivers value.Three models work better:ModelStructureBest forExampleHybridBase fee + outcome feesCompanies testing AI pricing$5K/month + $50 per analysisValue-based tiersDifferent prices per outcome typeMultiple distinct workflows$100/offer, $500/equity analysisOutcome-based with caps% of savings with min/maxMeasurable cost avoidance5% of savings, $10K-$100K/monthThe best pricing strategy depends on what you can measure reliably and what your customers already track.If your platform connects to existing HR analytics, outcome-based pricing becomes feasible.If not, start with hybrid models and evolve as your measurement improves. ## The infrastructure gap in HR tech Most billing platforms and metering solutions used by HR technology companies were built for traditional SaaS or lean heavily on "credits". They handle seats, usage tiers, and subscription management well, but they can't track what an AI agent does. They can't measure the value a compensation workflow delivers. They can't price based on outcomes because they weren't designed for products that deliver outcomes autonomously.If you're bulding AI-powered HR tools, you need different infrastructure:- **Signal-based tracking**: Capture business events (offers generated, analyses completed) not just technical events (API calls, tokens used)- **Cost attribution**: Understand what each agent interaction actually costs to run, so you can price with healthy margins- **Value visualization**: Show customers what their agent accomplished in metrics they already report to their leadership- **Flexible billing**: Support hybrid models, outcome-based pricing, and revenue sharing that traditional platforms can't handle ## What to do next If you're building AI-powered compensation technology, you face a measurement challenge that becomes a pricing challenge that becomes a growth challenge.Your agent delivers real value. You need to prove it, price for it, and show it continuously to customers.Three steps forward:- **Start measuring outcomes now**: Don't wait until you have perfect data. Begin tracking the signals that matter to HR leaders. Time saved, decisions improved, compliance maintained. Rough numbers beat no numbers.- **Test new pricing models**: Pick your best customer and try hybrid pricing. Base fee plus outcome fees. See how they respond. Refine based on feedback. You don't need to change all contracts at once.- **Connect to existing data**: The best ROI calculations use data customers already trust. If you have analytics showing cost savings, connect that to agent usage. If you track time to hire, attribute improvements to AI-assisted decisions.Don't stick with seat-based pricing, you'll struggle to justify their value as costs scale with usage.**Building AI for HR and compensation?** Learn how AI-native billing infrastructure helps you track costs, prove ROI, and move to outcome-based pricing. [Book a demo with Paid's pricing experts](https://paid.ai/get-started#contact-us) to discuss your specific needs. --- ### [2025-10-13] Should you build or buy an AI agent monetization solution? - URL: https://paid.ai/blog/ai-monetization/build-vs-buy-ai-agent-monetization-solution - Date: 2025-10-13 - Author: Arnon Shimoni - Summary: Building agent billing costs $500K-$2M and takes 6-12 months. Use this decision framework to determine if your SaaS company should build or buy instead. You added agents to your previously seat-based SaaS product. Now you need to bill for them. Should you build billing infrastructure or buy it?This guide helps you decide in under 10 minutes. ## The 60-second decision test Answer these three questions:- **Do you need to bill for agents in the next 90 days?** → YES: Buy a platform. Building takes 6-12 months. → NO: Continue to question 2.- **Can you dedicate 2-3 engineers full-time to billing for a year?** → NO: Buy a platform. You can't build this as a side project. → YES: Continue to question 3.- **Do you have unique margin optimization needs?** → NO: Buy a platform. The ROI doesn't justify building. → YES: You might be a candidate for building. Read on. ## Why agent billing is different Your SaaS billing won't work for agents. Here's why:SaaS billing typically assumesAgent RealityFixed number of seatsVariable agent deployment (2 to 30+ per customer)Predictable usage patternsAutonomous 24/7 operation with usage spikesHumans trigger actionsAgents make decisions independentlyFeatures drive valueOutcomes drive valueBusiness hours consumptionContinuous operation across timezones**The bottom line:** Forcing agents into seat-based pricing leaves money on the table and confuses customers. ## Should you price agents differently than SaaS? Yes. Companies switching to outcome-based agent pricing see 20-60% revenue increases within 6 months.- **Why different pricing works:**One customer cut seat prices 50% but added outcome fees. Revenue jumped 60% in 90 days.Another went from 20% to 60% YoY growth by switching to outcomes. They won deals from seat-based competitors by proving better ROI.- **What customers actually want to pay for:**Not seats. Not API calls, but **outcomes.**Meetings booked. Tickets resolved. Documents analyzed. Results delivered. ## The true cost of building Cost CategoryBuildingBuying Paid**Initial development**$500K - $2M$0**Time to first invoice**6-12 months2-4 weeks**Engineering team**2-3 full-time engineers0 dedicated engineers**Ongoing maintenance**$200K-$500K/yearIncluded**AI provider integrations**You build and maintainAutomatic updates**Tax compliance**Your team monitorsAutomatic updates**Feature development**Months per featureConfigure or request**Year 1 total**$700K - $2.5MPlatform fees only**Year 3 total**$1.1M - $3.5MPlatform fees only ## **Hidden costs of building:** Edge cases consume you. Custom deals, tax changes, failed payments for autonomous agents.Maintenance never ends. AI providers change APIs monthly. You track and implement updates forever.Opportunity cost kills you. Every hour on billing is an hour not improving your agents. ## What you actually need for agent billing CapabilityWhy It MattersBuild YourselfBuy Paid**AI cost tracking**Track spend across OpenAI, Anthropic, Google, etc.Maintain integrations foreverBuilt-in, auto-updated**Outcome-based pricing**Charge for results, not resourcesDesign data model, build engineConfigure without code**Margin visibility**Know profitability by customer/agentBuild analytics infrastructureReal-time dashboards**Value reporting**Prove ROI to customersBuild customer portalDrop-in components**Fast implementation**Start billing quickly6-12 months to MVP2-4 weeks to invoice**Flexible pricing**Change models as you learnRebuild when requirements changeConfigure on the fly ## When to build (Build only if you meet *ALL* these criteria:- **Genuinely unique billing logic**Not "we want custom fields." Your pricing would confuse agent billing experts. Test this by explaining to other agent companies.- **Dedicated billing team**2-3 engineers full-time for years. Not borrowed for sprints.- **Massive scale / extreme margin optimization needs**This is true for more than $50M annually + margin optimization that impacts millions (think Netflix, OpenAI, etc.)- **Billing is competitive advantage**How you charge creates differentiation customers care about. Billing innovation is core to your business.- **Indefinite maintenance commitment**You'll track AI provider changes monthly. New models, pricing updates, API changes.If you checked all five boxes, calculate total cost over three years. Compare to buying. Factor in opportunity cost.Most companies that pass this test still buy when they see the numbers. ## When to buy (>95% of companies) Buy if *any* of these are true:- **Timeline pressure:** Need to bill for agents within the next 3 months.- **Common pricing patterns:** Outcome-based, workflow-based, or hybrid models. Most agents fit these.- **No dedicated resources:** Engineering should focus on product, not billing infrastructure.- **Multi-provider needs:** Agents use OpenAI, Anthropic, Google, specialized APIs.- **Margin visibility required:** Need to understand profitability without building BI infrastructure. ## How to evaluate agent billing platforms Use this checklist when evaluating: ## Must-have capabilities - **☐ AI-native architecture** Built for agents from the ground up, not retrofitted from SaaS billing.- **☐ Agent-specific primitives** Signals, workflows, outcomes. Not just "users" and "API calls."- **☐ Outcome-based pricing** Native support without custom development.- **☐ Real-time cost tracking** Across all AI providers with automatic updates.- **☐ Fast implementation** 2-4 weeks to first invoice, not months of professional services.- **☐ Margin visibility** By customer, agent, and workflow in real-time.- **☐ Flexible pricing models** Outcome, workflow, hybrid without code changes.- **☐ Automatic tax compliance** Updates handled by platform. ## Red flags (automatic disqualification) - ** Forces agents into seat-based pricing**- ** Requires months of professional services**- ** No AI provider cost tracking**- ** Can't support outcome-based pricing**- ** No agent-specific primitives** ## The hybrid trap Some teams try to split the difference. Build part, buy part.I've been there, and this usually fails.**Why hybrid doesn't work:**You maintain integration points between systems. You need deep billing expertise for both. You inherit complexity of building AND constraints of buying. However, edge cases fall into gaps - debugging spans your code and the platform and then extends into RevRec systems, ERPs, and data warehouses for analytics. When changes touch multiple systems, you are trapping yourself in maintenance hell.**The only exception:**Buy *core* billing engine. Build thin customization layers on top through APIs.This requires clear boundaries and discipline, and incredible depths in the business processes both for the now and the future. Most companies let customization spread until they've rebuilt half the platform. ## What Paid offers We built Paid specifically for agent billing after seeing companies waste months building what exists. ## **Launch faster** - Outcome, workflow, hybrid, and FTE pricing models- 2-4 weeks from integration to first invoice- Works with LangChain, CrewAI, major frameworks ## **Manage margins** - Real-time cost tracking across AI providers- Profitability by customer, agent, workflow- Alerts when margins drop ## **Show value** - Customer-facing dashboards- Human value equivalent calculations- Automated value reports for renewals ## **Iterate on pricing** - Change models without migrations- Configure without code- Custom terms preserved automatically ## Some of the frequent questions we've heard ## **Q: Can we build an MVP and add features later?** This is the most common mistake. The MVP ***never*** stays minimal. I've personally spent 3 years building billing systems for SaaS and B2C companies - it will eat up 8-14 engineers' time and you'll spend 3-4 months building, then discover you need tax compliance, dunning logic, outcome attribution, margin tracking, and dozens of other features.Companies going this route end up rebuilding after 6-12 months. By then, you've lost a year of proper monetization. ## **Q: Our billing is unique. Won't a platform limit us?** Paid supports outcome-based, workflow-based, hybrid models, and custom signals. Most "unique" requirements are configurable, not custom.Test this: explain your requirements to other agent companies. If they understand you, a platform can handle it. ## **Q: Should we really price agents differently than SaaS?** **Yes - a thousand times yes!** Companies switching to outcome-based pricing see 20-60% revenue increases within 6 months.Seat-based pricing breaks when customers deploy variable numbers of agents. Outcome-based pricing aligns revenue with value delivered. ## **Q: What about vendor lock-in?** Paid uses standard APIs and exportable data formats. Your usage tracking uses OpenTelemetry.The real lock-in is building yourself. You're locked into your technical decisions, team expertise, and maintenance burden forever. ## **Q: How do we convince our CTO not to build?** Show the numbers:Total cost over 3 years to build and maintain vs buying.Opportunity cost of engineers on billing instead of product.Time to revenue: 2-4 weeks vs 6-12 months.Ask: is billing where we want our best engineers focused? ## **Q: Can we start with a platform and migrate to building later?** Yes. Lower risk than building first.Start billing quickly with Paid. Learn real requirements. If you discover truly unique needs, migrate with real data.Most companies never need to migrate. The platform handles everything as they grow. ## **Q: How does outcome-based pricing affect revenue?** Data shows 20-60% increases within 6 months.Pricing aligns with value. Customers deploy more agents when pricing makes sense. Renewals improve because ROI is clear.One customer cut seat prices 50%, added outcome fees, and saw 60% revenue jump in 90 days. ## **Q: What if we already have SaaS billing?** Your existing SaaS billing probably won't handle agents well. Seat-based systems break with variable agent deployment. Usage-based systems struggle with outcome attribution.Evaluate whether extending makes sense or if purpose-built agent billing is better. Most companies find agent billing different enough to warrant new infrastructure. ## The bottom line **An absolute majority of SaaS companies should buy agent billing infrastructure and not build it. **The opportunity cost is massive. The 5% that should build have genuinely unique models, or billing as competitive advantage.**For everyone else, focus engineering on better agents, not billing logic.**Ready to start billing for agents? --- ### [2025-10-08] Paid + GitLaw: Introducing Legal Contracts Built for AI Agents - URL: https://paid.ai/blog/ai-agents/paid-gitlaw-introducing-legal-contracts-built-for-ai-agents - Date: 2025-10-08 - Author: Arnon Shimoni - Summary: Paid has partnered with GitLaw to create and release an open-source Master Services Agreement (MSA) specifically designed for AI agent companies. This contract addresses the unique challenges that traditional SaaS contracts fail to cover when applied to autonomous, adaptive AI systems. The Agentic MSA clarifies decision responsibility, establishes appropriate liability frameworks, and provides clear language around data ownership and training rights. We've partnered with [GitLaw](https://git.law) to launch something that should have existed from day one: a Master Services Agreement specifically designed for AI agents.Not because we love legal documents (we don’t), but because the contracts most agent companies are using create problems they don't see until something breaks. ## The problem with using SaaS contracts for agents Most AI agent companies are still using SaaS contracts which makes no sense.Your software isn't just sitting there helping someone fill out a form. It's booking meetings, writing code, making decisions. When something goes wrong, who's liable?Standard contracts don't answer that question. They assume software waits for human instructions. Click button, thing happens, done.But your agent operates differently. It decides which prospects to contact. It writes outreach messages. It follows up based on response patterns. It learns from interactions and adjusts behavior over time.Those are autonomous actions. And when your agent does something unexpected, the gap between what your contract says and what your product does creates legal exposure you can't price for. ## Three ways agents break traditional contracts ## **Agents make decisions without approval** Your workflow agent doesn't suggest next steps. It executes them. Sends emails. Updates records. Moves data between systems. No human clicking approve at every stage. ## **Agents act continuously** Traditional software processes tasks one at a time when asked. Agents run 24/7, making hundreds of micro-decisions. Remember the Ford dealership chatbot that hallucinated a free truck offer? That's what happens when autonomous systems operate under contracts written for passive tools. ## **Agents adapt over time** Static software behaves the same way every deployment. Agents learn from context, adjust to patterns, change behavior based on accumulated data. The system you shipped six months ago operates differently today.Your SaaS contract wasn't built for any of this. ## What the Agentic MSA actually covers [Working with Nick and the GitLaw team](https://git.law/news/launching-the-agentic-msa----free-open-and-designed-for-ai-builders), we identified the contract gaps that create the most exposure for agent companies. The new MSA addresses three critical areas: ## **Agent classification and decision responsibility** The contract establishes that your agent functions as a sophisticated tool, not an autonomous employee. When a customer's agent books 500 meetings with the wrong prospect list, the answer to "who approved that?" cannot be "the AI decided."It has to be "the customer deployed the agent with these parameters and maintained oversight responsibility."The MSA includes explicit language in Section 1.2 that protects you from liability for autonomous decisions while clarifying customer responsibility. ## **Liability limitations and risk allocation** AI agents hallucinate. They produce confident outputs that turn out wrong. The MSA includes explicit disclaimers that agent outputs require human verification before material business decisions.It also includes damage caps appropriate for unpredictable systems. Typically 12 months of fees with exclusions for indirect losses. Not being difficult. Acknowledging you can't predict every edge case in software that learns and adapts.Section 7 covers liability limitations with AI-specific disclaimers about output accuracy in Section 4.1. ## **Data ownership and training rights** This kills more deals than any other contract issue. Your agent ingests customer data and generates outputs. You might want to use those interactions to improve your models.Customers panic when they hear that. They imagine their proprietary data training models that help competitors.The MSA establishes that customers own their data and any agent outputs. Then it provides separate, customizable language about using de-identified, aggregated data for training purposes. With clear opt-out options.Most customers accept training use when it's explained clearly. Trying to slip it in through vague language destroys trust.Section 2.1 covers ownership with customizable training permissions in the cover page variables. ## Why this matters for agent monetization At Paid, we solve billing and cost tracking for AI agents. But we kept hearing the same problem before companies even got to pricing.Founders would tell us they couldn't figure out how to charge for agents. Then we'd look at their contracts. They were trying to price outcome-based work using terms written for seat-based software.> "Most AI agent companies are still using SaaS contract language which makes no sense. Your software isn't just sitting there helping someone fill out a form. It's booking meetings, writing code, making decisions. When something goes wrong, who's liable? The standard contracts don't answer that. We kept hearing this from founders, so when GitLaw said they were building an agent-specific MSA, we jumped in. Builders need legal frameworks that match what their agents actually do".Manny Medina, Paid CEOYou can't bill for outcomes if your contract only covers usage. You can't price based on value delivered if your liability framework assumes predictable, passive behavior. You can't protect your margins when the legal foundation doesn't match what your product does.The contract shapes everything that comes after. Get it wrong and your entire business model sits on shaky ground. ## How to use the Agentic MSA The MSA is open source and free to use. You can access it directly in the [GitLaw Community](https://git.law/community) or ask the GitLaw AI Agent to generate a customized version for your specific needs.Because the law around AI agents is evolving rapidly, treat this as a starting point, not a substitute for legal advice. Work with a commercial lawyer to customize it for your situation.The template uses CommonPaper's Software Licensing Agreement and AI Addendum as a foundation, adapted for the unique characteristics of AI agents.Nick and the GitLaw team built this based on patterns from reviewing hundreds of agent contracts. We contributed our research from working with dozens of agent companies on monetization challenges.Together, we're building the infrastructure the agent economy needs. Legal frameworks that match how agents actually work. Billing systems that align pricing with value delivery. Cost tracking that protects margins.Because agents aren't just another SaaS feature. They're a fundamentally different product category that needs different infrastructure. ## What happens next Legal frameworks always lag behind technology. Right now that lag creates real risk for anyone building agents.You can ignore it and hope nothing breaks. Or you can use contracts built for what agents actually do, not what software did ten years ago.Most founders choose hope. The ones who survive choose better infrastructure - and you'll soon find these MSAs baked into Paid's offering too.[→ Access the Agentic MSA now](https://git.law/community/doc/ai-agent-msa-software-license-agreement-md-A9NZrQ?utm_source=paid_blog)[→ Read the announcement on GitLaw](https://git.law/news/launching-the-agentic-msa----free-open-and-designed-for-ai-builders)[→ Listen to our conversation with Nick](https://go.paid.ai/podcast-s2e21) about building legal infrastructure for the agent economy. --- ### [2025-10-03] Sowing the seeds of our $21M raise - URL: https://paid.ai/blog/company/sowing-the-seeds-of-our-21m-raise - Date: 2025-10-03 - Author: Arnon Shimoni - Summary: A reflection on Paid's $21M Series Seed funding announcement week, the difference between press validation and product validation, and what it means to build infrastructure for the AI agent economy at an inflection point. This week we announced our $21M Series Seed led by Lightspeed Venture Partners, with participation from FUSE and existing investor EQT Ventures (meaning total funding now stands at $33.3 million).It's been a few days now. The news has settled. I wanted to take a moment to reflect on what happened and what it means. ## The machinery of announcements Monday morning, [Julie Bort broke the news in TechCrunch](https://techcrunch.com/2025/09/28/paid-the-ai-agent-results-based-billing-startup-from-manny-medina-raises-huge-21m-seed/), followed by Manny talking on [Bloomberg's Daybreak with Tom Mackenzie](https://www.bloomberg.com/news/videos/2025-09-29/uk-has-good-tech-talent-pool-paid-s-manny-medina-says-video).In the evening, [we announced Paid's seed round ourselves on this blog](/blog/company/paid-raises-21-6-million-seed-round)!Tuesday brought a longer conversation with Jon Fortt on Fortt Knox.Jon pushed on the hard questions. How do you price agents when the per-seat model is dead? What happens when agents operate in the background? How do you prevent rogue agents from burning through budgets? What's the cost structure when LLM tokens are expensive?They're real issues companies are facing right now as they deploy agents. And honestly, hearing them articulated by someone outside our bubble confirmed what we've been seeing: this is a genuine problem that needs solving.Seeing Paid on the Nasdaq tower in Times Square was surreal. You work on something for months in a small office in London, and suddenly it's 50 feet tall in the middle of Manhattan. ## What we are asked during this week Jon Fortt asked Manny the questions we've been wrestling with internally for months. How do you price agents when they operate in the background? What happens when LLM costs spike unexpectedly? How do you prevent rogue agents from burning through budgets?These questions is what happens when you try to monetize something that doesn't behave like traditional software.Hearing them articulated by someone outside our world confirmed something important: we're not solving edge cases. We're solving the core problem that every company deploying agents will eventually hit.The pattern recognition feels right. The timing feels right. The problem is real and getting more urgent. ## This raise is important I've been through funding announcements before. They usually follow a script. Big number, impressive investors, bold vision for the future, the usual stuffI may be biased, but this one feels different to me. Not because of the amount or the logos, but because of what's happening in the market right now.SaaS companies are stuck. Growth has flatlined. Seat-based pricing is breaking because AI agents are replacing the seats. The business model that built a trillion-dollar industry doesn't work when your product eliminates headcount instead of enabling it.We're watching companies struggle with this in real time. They know they need to pivot to agents. They don't know how to monetize them profitably.That's not a future problem. That's happening right now. And the infrastructure to solve it doesn't exist yet.That's what makes this moment interesting. We're not building for what might happen in five years. We're building for what's breaking today. ## What we're actually building The thesis is straightforward: 50% of the workforce will be agents by 2030. The seat-based pricing model is already breaking. Companies need new infrastructure to monetize agents, track costs, and prove ROI.That's what Paid does. Cost tracking that shows you what each agent actually costs to run. Billing that handles outcome-based and hybrid pricing models. Margin management so you know if you're making money. ROI reporting so your customers can see the value.We started by proving this works for agent-first companies. Now we're helping SaaS companies make the transition before their competitors do. ## The work ahead The announcement week is over. Now comes the part that actually matters.We've got the resources and team to build what the AI agent economy needs. But infrastructure is only valuable if it solves real problems for real people.That's what we're focused on now. Not the headlines or the funding amount. Just making sure what we build actually works.The conversations we had last week with founders making this transition were more valuable than any press hit. They're wrestling with pricing models that don't fit their products. They're watching costs spiral without visibility into what's driving them. They're trying to prove ROI to customers when agent work is invisible.Those are the problems worth solving. Those are the conversations worth having.If you're making the transition to agents and need to figure out the economics, let's talk. --- ### [2025-09-29] Paid raises $21.6 million seed to help SaaS companies break free from the seat-based trap - URL: https://paid.ai/blog/company/paid-raises-21-6-million-seed-round - Date: 2025-09-28 - Author: Arnon Shimoni - Summary: Paid raises $21M led by Lightspeed, FUSE & EQT to help SaaS escape seat-based pricing with AI agent billing. 50% of workforce will be agents by 2030 - pricing must evolve. We’ve raised $21M in an oversubscribed Seed round led by Lightspeed Venture Partners, with participation from FUSE and existing investor EQT Ventures. This brings our total funding to $33.3 million.Our mission is to grow the AI agent economy by helping builders get paid for their agents.We founded Paid as a team of leaders who built and scaled category-defining companies: Manny Medina (co-founder of Outreach), Manoj Ganapthy (serial founder, built Salesforce billing), Raj Dosanjh (early Palantir) and Arnon Shimoni (early Pleo).With this new funding, we’re extending to include SaaS companies building AI Agents, helping them escape the seat-based trap and return to growth through AI agents. ## The SaaS-to-agent transition is our duty SaaS growth has stalled. Public SaaS companies are posting single-digit growth at best with many shrinking. Their core business model is breaking down.Seat counts are dropping across the industry. Why? Because AI agents are replacing entire teams - and no customer will pay “per seat” for software that eliminates seats.With 50% of the workforce expected to be AI agents by 2030, seat-based pricing is over. It built a trillion-dollar industry - but clinging to it means flatlined revenue while AI-first competitors capture all the growth.SaaS companies are now quickly pivoting to agents because it's the only path back to growth. The problem is today’s tech stack wasn't built for a world where agents dominate the workforce. ## AI agents are the life raft The route is clear, and smart SaaS companies are already selling AI agents. Early movers are seeing 20-40% revenue increases within 6 months, higher retention rates, and much faster sales cycles.For others, the transition feels impossible - but it is inevitable. ## Paid is the infrastructure for the SaaS-to-Agent shift Paid is the infrastructure layer that makes the SaaS-to-Agent transition possible. We handle five critical pieces: ## **Customer Value Proofs** - Make invisible AI work tangible business outcomes- Launch branded portals in minutes that show agent impact- Build transparency that earns trust and drives renewals ## **Custom Pricing, Fast** - Deploy per-customer pricing in minutes, aligned to value ## **Outcome-Based & Hybrid Models** - Replace seats with outcome- and value-based pricing- Blend subscriptions with performance fees- Enable revenue sharing and success-based models- Do what your current billing system can’t ## **True Cost Tracking** - See what each agent costs to run- Track margins by customer and agent- Optimize resources before they erode profits ## **AI Business Intelligence** - Monitor agent performance with dashboards- Analyze profitability and run “what-if” scenarios- Operate with full visibility in an agent-first world ## The moment is now The SaaS industry is at the biggest inflection point since on-prem to cloud. Companies that make the transition to AI agent business models in the next 12 months will dominate their categories. Those that don't will get left behind.Buyers are already demanding AI that delivers outcomes, not just features and the technology is now ready too. AI agents can now handle complex, multi-step workflows reliably, with cost structures that are becoming predictable enough for serious business planning.This funding accelerates our roadmap built specifically for SaaS companies making the AI agent transition.**If you’re a SaaS CEO**: Let’s talk, we can help you break free from seat-based pricing and get back to growth.**If you’re an application builder**: We’ll set you up to monetize your apps and agents quickly.**If you’re an agent builder**: We’ve got your entire business infrastructure covered - and it’s free.**Join us in building the agent economy.****Build agents, Get Paid.***Paid's platform is available now for SaaS companies ready to make the transition. Get started at **[paid.ai](http://paid.ai/)** or reach out to our team for a consultation on your AI agent strategy.* --- ### [2025-09-23] 10 features AI Agents need in a billing system - URL: https://paid.ai/blog/billing/10-features-ai-agents-need-in-a-billing-system - Date: 2025-09-23 - Author: Arnon Shimoni - Summary: The 10 features every real AI billing platform must have to capture value, margin, and ROI. After over 250 conversations with AI companies and reviewing production deployments at OpenAI, Sierra AI, [Fin.ai](http://fin.ai/), and others, one pattern is clear: **retrofit SaaS billing doesn’t work for AI agents.**Here are 10 features that AI billing platforms need. ## 1. Agent-native pricing models Traditional platforms offer "per-seat" or "usage-based" - but AI agents need four different models - which matches the value they deliver:- **Per-agent (when positioned as digital employee):** "Sarah the SDR Agent" costs $2,400/month.- **Per-workflow:** Each lead qualification process costs $12, whether it uses 100 tokens or 10,000.- **Per-action:** $0.15 per e-mail, $0.45 per document.- **Per-outcome:** Pay only when the agent succeeds.The mix of hybrid models where base subscription plus outcome fees can create sustainable unit economics. ## 2. Signal-Based tracking (beyond metering and token counting) Tokens are implementation details, but customers care about business outcomes.Track what matters to your business:- Meeting booked- Document processed- Issue resolved- Lead qualifiedTraditional metering platforms like Metronome and Orb force you to predefine metrics. Seats and API calls made sense for SaaS. Agents need something smarter.We believe in **[the crystal ball effect](/blog/ai-agents/the-agentic-runtime-is-a-crystal-ball)**[: Your agent runtime reveals patterns you never thought to measure](/blog/ai-agents/the-agentic-runtime-is-a-crystal-ball). Which workflows drive retention? Which API calls predict churn? Which customer behaviors signal expansion opportunities?Wrapping AI and agentic functions with trace calls can turn outcomes into the primary billing currency: ``` await paidClient.initializeTracing(); const paidOpenAiWrapper = new PaidOpenAI(openaiClient); await paidClient.trace("", async () => { return await agent.processCustomerRequest(paidOpenAiWrapper); }, ""); ``` Your comparison should be with SaaS billing where you need to make specific function calls to translate business into tokens. In agent billing, the business signals *are* the billing units. ## 3. Real-time margin visibility Most AI companies know revenue but not costs until the OpenAI or Anthropic bill arrives. That’s too late.Essential margin capabilities an agentic billing system needs:- Cost per workflow, per agent, per customer- Token efficiency monitoring- Alerts on margin shifts- Profitability dashboardsWhen a customer's agent switches from GPT-4o to GPT-5, your margins can crater. You need to know in minutes, not at the end of the month. ## 4. Multi-Vendor cost management Modern agents use multiple providers:- OpenAI for reasoning- Anthropic for document analysis- ElevenLabs for voice synthesis- Pinecone for vector searchAI billing needs to help you track your costs across vendors and updates rates automatically. You can’t know what your agent’s profitability is if you can’t reconcile the separate bills. ## 5. Outcome Attribution for Multi-Agent Workflows When three agents collaborate to book a meeting, who gets credit?- Research Agent finds a lead- Qualification Agent scores it- Outbound Agent books the meetingBilling must support **workflow-level attribution** with configurable revenue-split rules. ## 6. Performance-based pricing adjustments Not all outputs are equal. A 95% accurate document analysis delivers more value than 70%.An AI-native billing system can adjust pricing based on performance multipliers where the system calculates performance scores and adjusts pricing automatically.Billing platforms should support:- Scoring performance- Applying multipliers automatically based on specific customer contracts- Adjust pricing in real time ## 7. Prepaid credit management with smart limits AI workloads spike unpredictably (meaning, Monday morning backlogs can drive 50x cost jumps).Essential controls you will need:- Prepaid credit pools to prevent unpaid usage- Smart throttling when approaching limits- Predictive alerts based on usage patterns- Hierarchical controls (org → team → agent)Even OpenAI shifted to prepaid credits to manage this risk. ## 8. Value receipts (and ROI Dashboards) Customers need to justify agent ROI to their CFO. Traditional billing shows costs. AI billing should show value created.Automatic reports show:- Tasks completed vs. human hours saved- Cost per outcome vs. human baseline- Productivity trends- Department-level ROIExample: *“Agents processed 2,847 tickets, saving 712 hours at $2.15 each vs. $45 for human resolution.”* ## 9. Customer-Specific Agent Pricing Your "Document Analysis Agent" might save Enterprise Corp $50,000/month but only $500/month for a small law firm. Why should they pay the same?Same agent, different pricing:- **Enterprise Corp:** $0.75 per page (high-value contract analysis)- **Mid-Market Legal:** $0.25 per page (routine contract review)- **Small Firm:** $0.10 per page (basic processing)AI billing ties price to customer value, not uniform features. ## 10. Non-Disruptive Integration We know SaaS companies can’t rip out billing. Agent-native platforms must:- Layer alongside existing systems (Stripe, Chargebee, Zuora)- Connect to customer portals, dashboards, CRMs- Work with existing tax + revenue recognitionResult: predictable SaaS revenue stays on current rails, while variable agent value layers in seamlessly. ## Why it matters now 75% of AI companies struggle with billing because they’re forcing human-era tools onto agent-native problems. If you’re building agents, think of this now!- OpenAI spent many months building billing from scratch.- Sierra AI engineered outcome tracking in-house.Yes, you can reinvent the wheel or use a platform built for the agentic economy. Every day of delay compounds how difficult it’ll be to replace.*Paid is the only billing platform built specifically for AI agents. Track costs, measure outcomes, and bill flexibly with simple integration.* --- ### [2025-09-19] EU data act killed ARR - URL: https://paid.ai/blog/ai-monetization/eu-data-act-killed-arr - Date: 2025-09-19 - Author: Arnon Shimoni - Summary: The EU Data Act changes everything for SaaS: customers can now cancel anytime with two months’ notice. ARR isn’t guaranteed anymore, and retention is no longer about contracts, but about operations, renewals, and customer care. ## The EU Just Killed ARR Annual Recurring Revenue (ARR) has been the backbone of SaaS for two decades.It gave investors predictable growth curves. It gave founders leverage in fundraising. It let boards sleep at night.But ARR was never a *really well defined metric*. And now, it may be practically dead in Europe.With the EU Data Act (effective September 2025), **every SaaS contract with an EU customer becomes a “cancel anytime” subscription.** Customers can walk away with two months’ notice. No excuses, no lock-in.What's good for consumers is now good for companies too.But if ARR is no longer “annual”, but rather optional - is it actually good for companies? ## What the law says The EU Data Act is broad, covering connected products, IoT devices, and critically for SaaS: **data processing services**. Buried in the law is a simple obligation:- Providers must allow customers to terminate contracts at any time.- Notice periods can’t exceed two months.- Contractual or technical “barriers to switching” are prohibited.The usual trick that SaaS companies rely on, offering “discounted” three-year deals in exchange for lock-in, is all but gone. Long-tail penalties for early exit? Off the table.Investors may still talk about ARR multiples, but in Europe, **that revenue is a rolling suggestion, not a contractual guarantee.** ## Involuntary churn is now a much bigger problem Most founders are reacting to this by arguing about pricing models. Do we move to usage-based? Do we invent new discount mechanics? Do we pivot to prepaid credits?But there’s a quieter, nastier risk nobody is preparing for: **involuntary churn. **That’s when services terminate *without the customer consciously deciding to leave.*This would happen when credit cards expire and not updated, but now imagine this:- The finance team is buried in year-end close.- The admin who manages your tool is on vacation.- Your reminder email gets lost in an overloaded inbox.If your vendor has not designed the entitlements/contract correctly - this could result in a subscription terminated. Access revoked. Business workflows broken. It’s the equivalent of your electricity being cut off because you didn’t click “confirm payment” on a reminder email.Here’s what this looks like in real life assuming there are no grace periods correctly implemented:- Customer forgets to renew (because they’re busy running their business).- Your system auto-terminates access (because compliance demands it).- Their team shows up Monday morning and can’t log in.- They call you screaming.- You scramble to restore access.That’s not just lost revenue. That’s a **trust breach**. It turns one forgetful renewal into a multi-department escalation. And it’s happening at the exact moment SaaS teams are already battling rising churn rates.You now need to start considering grace periods and a whole lot more. ## Practically, what you need to start considering To survive, you’ll need to make sure your product has renewals around **business reality, not contract lock-in.**That means:- **Smarter notifications** - Sequenced, multi-channel reminders. Not one lonely email. You may need to pick up the phone!- **Realistic grace periods** - Think weeks, not 72 hours. Enterprises don’t move on startup timelines.- **Tiered service degradation** - Restrict new actions, but don’t shut down existing workflows overnight.- **Emergency restoration protocols** - That VP *will* call your CEO if their team is blocked, and you may not be able to escape it.- **Audit trails for compliance** - You’ll need to prove you gave fair notice before shutting anything off.It’s table stakes if you want to retain enterprise accounts under the new rules, because ARR is no longer enforced by contract. It’s enforced by **customer satisfaction.**Going forward, your growth curve depends less on how clever your contracts are, and more on how resilient your operations are when customers forget, delay, or mismanage renewals. ## Compliance as competitive advantage? Most founders will see this as a compliance nightmare. The smart ones will turn it into differentiation.- Build continuity safeguards that competitors skip.- Market “never-lose-access” as a feature.- Treat renewal operations as a core product discipline.Retention, under the EU Data Act, is no longer a passive outcome**.**Welcome to the new normal: ARR as a reflection of customer reality, not contractual fantasy.*Additional reading:*- [Commission publishes Frequently Asked Questions about the Data Act](https://digital-strategy.ec.europa.eu/en/library/commission-publishes-frequently-asked-questions-about-data-act)- [The EU Data Act Just Killed Long SaaS Contracts](https://revenuewizards.com/blog/the-eu-data-act-just-killed-long-saas-contracts) --- ### [2025-09-18] Variable pricing meets fixed budgets: Notes from the trenches - URL: https://paid.ai/blog/ai-monetization/2025-09-18-variable-pricing-meets-fixed-budgets-notes-from-the-trenches - Date: 2025-09-18 - Author: Arnon Shimoni - Summary: Practical insights from AI founders and Ops people on pricing strategies that work: paid 2-week sandboxes, canary metrics, volume packaging, and the year-one-fixed/year-two-outcome playbook. As part of our Paid Roundtable series, we again hosted AI founders and Ops professionals to discuss why CFOs hate AI pricing.Here's what we learned from bleeding on the front lines: ## The Excel issue I started by saying what I've heard from other CFOs - *"I don't know how to put usage components into Excel". *I'm dumbing it down a bit, sure, but that's the gist of it.Think about that. Procurement teams have been using the same forecasting models for 15 years. Fixed seats, fixed costs, nice clean cells. You show up with consumption pricing and their entire planning system breaks.Another founder said another variant of this objection is *"Something better could come out in 3-6 months"*, which probably means *"I have no idea how to budget for this"*. ## The 2x2 emerging pricing archetypes that determine your pricing model Here's the framework that we showed during our discussion:**Attribution (Y-axis):** Can you prove your impact? **Autonomy (X-axis):** Does your agent work independently?High/High = Charge for outcomes High/Low = Workflow consumption pricing Low/High = Human equivalent pricingLow/Low = Stick to seats or usageMost founders are trying to charge for outcomes when they're actually in the Low/Low quadrant. ## Year one fixed, year two outcome Our cofounder Manoj's playbook, which has worked across multiple deals, is to start with a fixed price year one to gather data and build trust. Then, on year two: renegotiate with their own data proving value.The key insight: You're using year one as paid data collection. Every interaction, every outcome, every hour saved gets logged. When renewal comes, you have twelve months of proof. Contracts go from $100K to $500K because the value is undeniable.This solves the CFO's predictability problem while setting you up for premium pricing. ## The 2-week sandbox strategy Andrew from Dialog closes deals without a finished product. Ben from Covecta refined it further:2-4 week sandbox. 3-5 users. Document every hour saved. Build the business case together.Compare this to the typical 6-month PoC. After 2 weeks, they've seen tangible value and remember why they were excited. After 6 months, you're just another vendor in a long evaluation cycle.The critical shift: These are paid sandboxes. If they won't pay for a 2-week pilot, they're not serious about implementation. ## Canary metrics solve attribution Tom Williams from Supercase introduced a concept that resonated, at least for me: canary metrics.You can't always prove your AI drove revenue. But you can measure the indicators that correlate with outcomes:- Drafts created per day- Cases processed per week- Tickets resolved without escalation- Documents reviewed per hourThese are your canaries. They predict the outcome even when direct attribution is impossible.Tom's approach: Start by pricing against these measurable outputs. Show the correlation to business outcomes over time. Graduate to outcome pricing once you've established the relationship with data. ## Volume packs create predictability This model which was brought up solves the Excel problem I raised elegantly:- 10,000 agent tasks upfront- 12-month commitment- Overage charges for excess usage- Tied to customer's business projectionsCFOs get their fixed line item. You get usage-based reality. If they use more, it means they're getting more value and their business is growing.The clever part: You're not capping value, you're packaging it. ## What we took away After an hour of comparing notes, the patterns for me were clear:**Paid sandboxes beat free pilots.** 2-4 weeks of focused value discovery trumps 6 months of tire-kicking.**Canary metrics unlock pricing conversations.** When you can't prove revenue impact, prove the activities that drive it.**Year one is about trust, year two is about value.** Use fixed pricing to gather data, then graduate to outcomes.**Volume packages solve the Excel problem.** Give CFOs a number they can put in a cell while maintaining usage flexibility.The variable cost objection isn't really about cost. It's about predictability and control. Don't fight for the CFO's Excel model – work within it while building toward something better.*Thanks to everyone who's participated in our roundtable events so far.**We run these roundtables every 3 weeks. Sign up to our newsletter to stay on top of events, as well as our **[LinkedIn](https://www.linkedin.com/company/paid-ai/)** and **[Luma events](https://luma.com/paid)**.* --- ### [2025-09-15] The Economics of Agent Monetization: What We Learned from 80 Builders in One Room - URL: https://paid.ai/blog/company/the-economics-of-agent-monetization-what-we-learned-from-80-builders-in-one-room - Date: 2025-09-15 - Author: Arnon Shimoni - Summary: Learn how 80 AI builders confronted the harsh reality of agent economics at our SF event. Discover why tons of AI companies don't know their cost per outcome, how Artisan found features losing money, and the pricing models that actually work for AI agents. Last Thursday night, Lightspeed's San Francisco office turned into a playground for great conversations about AI economics I've witnessed. The topic? How to *actually* make money with AI agents.We expected 80 people. We got 300 registrations. We had to turn people away at the door.Here's what that tells you. Everyone's building agents. Almost nobody knows how to price them profitably. And for the first time, pricing isn't a back-office function anymore. It's a board-level emergency. ## **Monetization has become strategy** AI is forcing every SaaS company to rethink monetization from the ground up. This isn't iterating on pricing tiers. It's rebuilding the entire economic model while the plane is flying.The room had an interesting mix. 40% founders. a bunch of investors. And surprisingly, a few CFOs and RevOps leaders.Five years ago, CFOs didn't show up to product meetups. Now they're in the room because pricing decisions make or break the company before the next board meeting.One RevOps leader I spoke to said something around "we spent three years perfecting our seat-based pricing. Our entire sales comp structure is built on it. Our forecasting models assume it. Now we need to throw it all away and figure out how to charge for outcomes we can barely measure." ## **Our big thoughts** Manny asked the crowd to take big notes for big thoughts.I did see most people writing, and here are mine: ## **1: Your margins are already dead, you just don't know it yet** Lots of people told me they don't really know how they're pricing, or how to approach it.They won't be our customers necessarily, but they don't know if they're even breaking even. Not low margin. Straight negative.You could see founders doing mental math on their own features, as Many showed the margin collapse. ## **2: Seat-based pricing is corporate welfare for AI** This one sparked a bit of a debate. Traditional SaaS charges per seat because humans are the constraint. But what happens when one agent replaces five employees? You're charging for one seat while delivering five people's worth of value.Christian from IFS put it perfectly when talking about their [IFS.ai](http://IFS.ai) offering. They're running industrial AI agents that manage entire fleets of ships. Charging per seat would be like charging for one forklift on a navy vessel when you've automated the entire dockyard. ## **3: Outcome-based isn't optional anymore** Every successful agent company in that room had already moved to some form of outcome pricing. Not because it's trendy. Because it's the only model that aligns incentives when agents can scale infinitely.Sierra charges only for successful task completions. Fin charges $0.99 per AI resolution. Not experiments, but survival strategies... ## **4: Manoj's demo** You had to see it - but Manoj showed how one agent can be monetized in different ways for different customers, despite the core offering be the same.Customer A pays per renewals saved. Customer B pays per outcomes on growth delivered.Same underlying costs. Wildly different margins. ## **What now?** We've had serious pick-up, from companies we never thought would be our customers. One attendee texted me earlier today (at 3am my time!), after running their numbers post-event, they want to reprice and get a grip on margins.That's the difference between building agents and building a business.For those who missed it, we're offering pricing consultations to attendees only, valid through next week. If you're struggling with agent economics, now's the time to fix it - reach out via the QR code you got.The tools exist. The playbooks are emerging. The only question is whether you'll figure out your economics before your runway ends.Build agents. Get Paid. --- ### [2025-09-09] Seat-based pricing killed SaaS. Don’t let it kill agents. - URL: https://paid.ai/blog/ai-monetization/seat-based-pricing-killed-saas-don-t-let-it-kill-agents - Date: 2025-09-09 - Author: Arnon Shimoni - Summary: 75% of AI companies are discovering the same fatal math: agents reduce headcount while delivering more value. Seat-based pricing punishes this success. After 250+ conversations with AI agent companies, we keep hearing the same impossible math problem.**Agents reduce headcount while delivering more value.**Seat-based pricing punishes you for this success. ## The problems from our customer calls One founder building AI agents put it perfectly:> "I am actually reducing headcount to serve the same purpose. So when I have fewer seats, the credits won't make up for the difference, because the credits are attached to a seat. My expansion became a nightmare."Think about that for a second.Your product works. It automates tasks. It replaces headcount. Customers need fewer seats.Your reward? Less revenue.> "Our original 1.0 was kind of more of a SaaS product. We charge per seat per user... people liked it, right? Now we're releasing our 2.0 product, which probably should be more of a consumption product because of the workflows that will support being a little bit more broad, probably achieving more at scale."More scale. Fewer seats. Lower revenue. ## The math isn’t mathing One of our customers sees the trap clearly:> "Right now, if I were to have a seat for every user in an organization... the margin is going to be tremendously good right now because usage is so little. They're just getting used to it. At one point they're going to start to use it so much I can't increase my SaaS pricing that way."The more successful your AI agent becomes, the worse your economics get.Another spelled it out more vividly:> "How am I going to charge seat-based when I'm incurring these costs in the background with LLMs, data providers, everything? I'm incurring these costs to make these agents become alive on my platform. But I'm only charging a per-seat price."Variable costs. Fixed seat revenue. Growing usage. Shrinking headcount.The math doesn't work. ## I think we all know it’s at a breaking point From our conversations, the realization is universal:- **318 mentions** of seat pricing challenges- **75% of AI companies** struggling with this exact paradox- **22 major companies** actively discussing the problemOne company revealed their board's directive: **"*****They've got to move off seat price, because if they remain on seat price, there's a bunch of other tools that look the same."***Another founder was even more direct:> "I fundamentally believe seats are going to die. Because seats are being destroyed when workflow consumption is going up." ## The Bottom Line Seat-based pricing assumes more value = more users, but as we know AI agents deliver more value with fewer users and fewer butts-in-seats.You can't solve this with discounts. You can't solve it with usage credits attached to seats. You can't solve it by ignoring it.As one founder concluded a recent call with us: **"If they don't change, they're gonna get crushed."**Seat-based pricing will not survive the AI agent revolution, but you still can.*Data from 250+ customer conversations with AI agent companies, 2025.* --- ### [2025-09-08] Behind the scenes on our billboards - URL: https://paid.ai/blog/craft/behind-the-scenes-on-our-billboards - Date: 2025-09-08 - Author: Arnon Shimoni - Summary: From 250 customer conversations to 6 SF billboards. The creative process behind our "Build agents, Get Paid" outdoor advertising campaign. One marketing decision turns into 6 billboards, 4 creative directions, and a Figma with 50 different options - that’s what we wrestled with for our recent campaign. ## Why go down this route with billboards? I wanted maximum bang for our buck, not the "spray and pray" approach other people seem to love. We didn't know what works yet, so going all-in felt reckless.Plus, I wasn't interested in the "make people angry for attention" playbook that seems popular these days. That's not us - at least not right now.I contacted a couple of agencies, and most came back with what I'd call "the usual suspects" - every outdoor concept fell into predictable buckets:- Direct developer pain- Feature positioning- Problem-solution format ## **What actually resonates?** As I looked through our customer conversations (around 250 of them!), I found a few patterns that seemed to do the trick.I don’t think developers wake up excited about "cost management". It’s often an afterthought. However, management seems to worry about margins. ## **The creative direction** **During our first round,** we did too much explanation.But it was a bit boring, a bit too many buzzwords if you know what I mean.**During the second round,** it was too pithy and clever."Building a billing UI is like watching paint dry. But less fun" - Better, but requires too much processing time at 65mph.I tried to get smart, but it was too much.**Third round: **What we actually went with- "Build agents, Get Paid" - Clear journey, obvious outcome- "Your everything app for AI Agents" - Positioning without jargon- "It's not the model, it's the margins" - The key insight ## **Why these three won** - "Build agents, Get Paid" - Every developer's actual goal. No fluff.- "Your everything app for AI Agents" - Simple positioning that doesn't require a PhD to understand.- "It's not the model, it's the margins" - The truth bomb that separates successful AI companies from the rest. ## **The format strategy** - Bigger wallscapes and billboards (high dwell time) = "Your everything app" works here!- Smaller formats (drive-by glance) = "Build agents, Get Paid" and "It's not the model, it's the margins" ## Being smart about this Smart budgeting means testing first, scaling second.We're not trying to piss people off for viral moments. We're trying to solve real problems for real developers.The billboards are live across SF now - try and find them!We are, of course, tracking which messages drive the most [event signups for our upcoming event](https://luma.com/0ubzf5xm), as well as overall site visits. Then double down on what actually converts.Anyone else taken the "measure twice, cut once" approach to out of home advertising?Let's see if being strategic beats being loud.p.s. these aren't as expensive as you think they'd be. --- ### [2025-09-04] Chaos is the operating system - URL: https://paid.ai/blog/craft/chaos-is-the-operating-system - Date: 2025-09-04 - Author: Olly Aston - Summary: How chaos beats polish when building at startup speed. A design leader breaks down why messy iteration ships faster than perfect specs, and how controlled chaos becomes your competitive advantage. Most people see the polished interface. Clean layouts. Smooth flows. Details that look inevitable.What they don’t see is the chaos that built it.My brain doesn’t work in straight lines. It works like spaghetti. Fragments tangle - a sketch in Figma, a half-baked ChatGPT prompt, a broken prototype still open in Replit or Lovable. By 3 AM I’ve got nine tools open, three versions of the same idea colliding, tabs I should have closed hours ago. It looks like a mess, but it’s momentum. Patterns start to emerge. Breakthroughs hide in the noise.I design in that chaos. Not because I have to, but because I thrive in it. And because moving at startup speed demands it. If you are chasing perfect, you are already late.This isn’t about being messy for the sake of it. It’s about unlearning the way we were taught to design, because that way doesn’t work at the speed startups move. ## **What I had to break to move faster** I've worked at places that taught me opposite lessons. IBM's Design Thinking gave me discipline - loops, hills, user outcomes, the whole system. Anaplan showed me how to scale that discipline post-IPO with serious research and design systems. Sentieo was the scramble phase where I learned design could unify everything and give a company its backbone. But each time, the foundation that made us good also made us slow.AlphaSense was the wake-up call - bigger teams, smaller influence, frameworks for their own sake, research that meant paying people to pretend to care. hx stripped all that back to high-stakes clarity where you couldn't fake it.The pattern became obvious: every place rewarded polish over progress, but chaos always shipped.Now I know what to break to move faster. ## **The chaos loop at Paid** At Paid, design is not a phase or a handoff. It is a loop. Messy, fast, purposeful.Here’s how it actually works:- We don’t start with polished specs. I throw rough sketches, hacked prototypes, even broken flows into the wild.- Engineers don’t wait for permission. They grab it, tear it apart, build it forward.- Customers don’t wait for research sessions. They stumble into early versions and tell us exactly where it sucks.- Then design circles back — not to polish for polish’s sake, but to harden what works, cut what doesn’t, and lock it into the system.It’s not linear. It’s not clean. But it’s alive. And that’s why it works.Most first versions are wrong. Good. Wrong gets you to right faster than polish ever will.And design isn’t siloed. Everyone is in the loop. Engineers design when they ship code. I design when I break flows in Figma, Replit, Lovable, or even Bolt (which never works, but I’m still a believer). Customers design when they stumble through and tell us it sucks.From the outside it looks chaotic. Inside, it’s our speed advantage.That is Paid’s DNA: messy, fast, but with intent. We don’t design to keep things neat. We design to ship faster, learn faster, and earn trust in what we’re building. ## **Chaos is the speed hack** The tools keep me moving. Music blasting, YouTube in the background...Figma. ChatGPT. Cursor. Replit. Gemini. Lovable. MidJourney. Sora. Veo. Perplexity. Bolt. Some nights it’s all of them at once, tabs stacked to infinity.They don’t make decisions. They give me fifty variations so I can throw forty-nine away before lunch. AI doesn’t have taste. But it has tireless execution. I supply the judgment. They supply the speed.Chaos isn’t careless. Chaos is how you learn faster than your competition can copy. ## **The craft behind the chaos** Fast doesn’t mean careless. The faster you ship, the more details matter.**Typography:** Inter for interface text, JetBrains Mono for code and data. Tested across seventeen combinations. Inter works at 12px on mobile and 48px in a boardroom. JetBrains Mono makes numbers scannable in walls of data. That isn’t aesthetic. That’s business. A CFO parsing billing at 7am without coffee: that is trust.**Colors:** a system built to scale across light, dark, and high contrast. Error red that warns without panic. Success green that feels earned, not patronizing. Handle revenue? These details decide if people trust you or churn.**Icons:** Material Symbols. Not because they are beautiful (they are not) but because they are obvious. Nobody wants to decode your clever custom icons. They just want to find “Settings” and get on with it. Every second spent decoding design is a second they’re not getting value.All of these decisions come together in Onyx: Paid’s design system.Onyx is not just a library of tokens and components. It is the foundation that unifies typography, color, motion, and states into a system that scales as fast as we move. Chaos creates the concepts. Onyx locks them in. That’s how we stay fast without breaking trust.These are not design flourishes. They are revenue decisions. ## **Who survives the chaos?** This way of working is not for everyone. If you need comfort, polish, or perfect roadmaps, you will hate it here.At our speed, perfect is poison. Clean kills. Process drags.We don’t wait. We don’t polish. We ship. We learn. We move.Chaos is not noise. Chaos is fuel. It is the unfair advantage.Design is not decoration. Design is the architecture. It is how the thing stands up when everything else is breaking.So choose: do you want the illusion of tidy design for yesterday’s problem, or the chaos that actually wins today?We chose chaos. And that’s why Paid will win. --- ### [2025-09-03] Are you selling agents the way customers want to buy? - URL: https://paid.ai/blog/ai-monetization/are-you-selling-agents-the-way-customers-want-to-buy - Date: 2025-09-03 - Author: Arnon Shimoni - Summary: Instead of asking CEOs about their AI pricing strategy, I ask one simple question: 'How are your reps selling it?' The answer reveals everything. We've stopped asking CEOs about their AI pricing strategy. It's the wrong question.Instead, we ask one simple thing: "How are your reps actually selling it?"The answer tells us everything we need to know about whether they'll survive the next twelve months.- **"Custom deals."** Translation: We have no idea what we're doing, so every conversation is a one-off negotiation that takes six weeks to approve.- **"We're still figuring it out."** Translation: Our reps are making it up as they go, probably losing deals while RevOps scrambles to build something, anything, that works.- **"Same as our other products."** Translation: We're forcing AI agents into seat-based pricing and hoping nobody notices the fundamental mismatch.The pricing strategy question is constraining. It assumes you have one. But asking how reps are selling reveals the truth: positioning problems, process gaps, and the painful reality that most companies have no repeatable playbook for selling autonomous software.Your reps aren't incompetent. They're improvising!Every deal becomes a custom negotiation because there's no standard way to sell something that doesn't fit in a seat ## Your sales aren’t the villain We’ve heard CEOs complain "why we're still selling seats for autonomous agents" - and it's tempting to point fingers at sales but this really isn’t a sales problem.Your reps very likely are *not* lazy. They're stuck with the only model your systems can actually operationalize.Your deal desk/RevOps is probably stretched thin.Some numbers say there’s one RevOps person for every 108 employees. That one person isn't building modern billing infrastructure but focusing on the tactical approvals, spreadsheets, and Slack pings from Sales asking, "Can we please just do usage-based pricing for this one deal?"The answer is always no. Not because RevOps doesn't want to help. Because the infrastructure doesn't exist.So your sales team defaults to what's safe and won't require six weeks of change management. ## What it looks like for your numbers I think we’ve all seen what happens when you try to squeeze agents into old SaaS pricing boxes.- **GitHub Copilot** charges $10/month but heavy users cost Microsoft $80/month. Average loss: $20 per user. Annual bleeding: $240 million. But it’s fine because Microsoft subsidizes that.- **OpenAI** brings in $4 billion in revenue but burns $5 billion in compute.- **Chegg** lost 90% of its market cap when students realized AI could answer their homework questions. Sadly went from education platform to cautionary tale in months.- **Stack Overflow** watched questions drop 64% and engagement fall 25% because developers ask ChatGPT instead of humans now. ## Selling agents is harder than it looks At face value "just charge for usage" or "sell outcomes" sounds simple. We know we’ve been preaching it - but it is actually a bit difficult.Some AI agents send 50 emails, analyzes 200 accounts, makes 13 LinkedIn touches, then books one meeting. So which action should the customer pay for?If you’re not sure, your customer surely won’t know and you definitely haven’t set it up right.A [BCG report called Rethinking B2B Software Pricing in the Agentic AI Era](https://web-assets.bcg.com/pdf-src/prod-live/rethinking-b2b-software-pricing-in-the-era-of-ai.pdf) found that 25% of buyers can't figure out value attribution for AI. Another 24% say outcomes depend on factors completely outside vendor control. ## What you need to get set-up **Agent-level tracking** that knows which customers drive costs and which drive value. Without this, you're flying blind with a flamethrower attached to your bank account.**Attribution that makes sense.** Was it the 50th email or the 13th LinkedIn touch that closed the deal? Without proper tracking, you're just making expensive guesses.**Hybrid pricing flexibility.** Base fees plus usage plus outcomes. You need all three to bridge from seats to value-based models. Your current billing system offers exactly none.**Real-time margin visibility.** LLM costs swing by 70 percentage points across accounts. One customer is profitable. Another is killing your P&L. You have no idea which is which. ## Who's Actually Getting This Right While everyone else bleeds, three companies cracked the code:- **Clay** achieved 6x revenue growth and a $1.25B valuation with credit-based pricing. Customers buy credits upfront with no surprise bills, and probably not much margin destruction.- **Intercom** generates tens of millions charging $0.99 per resolution. They defined "resolved" clearly: customer marks it done or doesn't return in 48 hours. No philosophy degree required!- **Sierra** hit $20M ARR in under a year by pricing directly on outcomes, with cost savings and revenue generated. That’s real business impact.None of them jumped straight from seats to outcomes. That could destroy their businesses.Instead, they built hybrid models on the path to outcomes. They tested attribution systems, and transitioned in stages. They also built the infrastructure first, then figured out pricing, then launched.You're probably doing it backward. ## The human-emotional layer Your sales leaders feel trapped selling something they know doesn't align with customer value. They hear "we want usage-based pricing" ten times a day and have to respond with "let me see what we can do" knowing the answer is nothing (or a month long delay to get that hacked into the system)Then RevOps leaders feel guilty blocking creative pricing ideas because the systems can't support them. They're not trying to be the bad guy. They're just out of duct tape and prayers.Then the CEO feels the pressure of investor promises about "AI monetization" while staring at ops infrastructure from 2015. Board meetings are getting awkward.Underneath it all is this quiet fear: What if we launch and the whole model collapses, or we’re universally panned for thec hange?Chegg proves it can happen. Stack Overflow proves it is happening.Your model will break, the only question is when. ## The bridge from here to there You can't flip from seats to usage overnight - and yet you can't stay where you are either.Start with hybrid models. Keep a base platform fee (procurement understands this) but add usage components for actual agent work. Your sales team can still forecast something. Customers get flexibility.Pick one value metric that matters.Build the infrastructure before you need it. Not after your sales team has already lost fifty deals to competitors who got there first. ## The Question That Matters > **“How will you offer to your customers a way to pay that matches the value they receive?” **That’s it.If you can't offer that, someone else will. And probably already is.Your infrastructure may be failing you, which leads your teams to be frustrated.Until you can track, attribute, and flex pricing around agent value, you're fighting with one hand tied behind your back.So let me ask again. Are you selling agents the way customers want to buy them? --- ### [2025-09-02] The FDE hype is real. So is the confusion - URL: https://paid.ai/blog/company/the-fde-hype-is-real-so-is-the-confusion - Date: 2025-09-01 - Author: Arnon Shimoni - Summary: What happens when ex-Palantir, Ironclad vets, and AI founders compare FDE battle scars? Raw insights on hiring, deploying, and surviving forward deployed engineering. Last week, 10 founders and operators joined a Zoom call to discuss forward deployed engineers.Here’s what we learned from each other. ## The big question we didn’t actually *fully* agree on Shyam from Boomerang opened with what everyone was thinking: *"Can someone explain what an FDE actually is?"*The answer was something along:> "It's about ownership. A Salesforce architect executes a statement of work. An FDE makes sure the customer doesn't churn and buys more".Jay from Casper Studio simplified it further: "Does the definition even matter? Just hire people who can code and talk to customers. Call them whatever makes them feel good". ## How to hire these majestic FDE unicorns *"A Harvard CS grad who did improv, is funny, dynamic, single, and can travel 48 weeks a year"* is not easy to hire, but we found better fishing holes:- Customer support reps handling complex tickets- Sales engineers tired of sandboxes- Technical BDRs who actually understand the product- Failed founders who still need revenue ## What an FDE deployment looks like One of our participants closed a Japanese multinational. Their brief to the FDE was "book a ticket to Tokyo. Take the corporate card. Get them live.Lots of problems were unearthed (and solved) by the FDE, like "our OCR doesn't work for Japanese characters written vertically", the US team ran 24-hour support while the FDE fought fires on-site.Turns out half the job was drinking with the client team every night. Technical skills only got you halfway. ## Preventing FDE scope-creep When I asked about scope creep, James flipped it on me:> "If you're worried about scope creep, don't hire FDEs. Scope creep IS the job. You're getting paid to learn from great customers."The key is be ruthless about which customers get this treatment. Meaning, the strategic ones that shape your roadmap and not your small $20K accounts. ## What we took away As with every roundtable we run, I asked for specific takeaways:- "Not all customers deserve FDEs. Only the ones you can learn from."- "PM skills matter more than coding when building new features."- "Look for talent in CS and BizOps, not just engineering."- "Sales engineers are underrated FDE candidates."Simple, tactical next steps! ## FDEs aren’t a unique challenge Every founder thought their FDE challenges were unique. They weren't.We're all fighting the same battle - proving value for something that's never existed before, in a language buyers don't understand.The FDE model isn't a deployment strategy. It's an admission that our products aren't ready and our customers don't know what they want.The only way forward is embedding smart people who figure it out in real-time.And maybe that's okay --- ### [2025-08-27] The agentic runtime is a crystal ball - URL: https://paid.ai/blog/ai-agents/the-agentic-runtime-is-a-crystal-ball - Date: 2025-08-27 - Author: Arnon Shimoni - Summary: The most expensive agentic operations were their most valuable customers. Your agent's runtime is revealing which features matter, who'll upgrade, and what your business actually does. The cost layer is your intelligence layer. Two weeks ago, I was on a customer call that was supposed to be about margin alerts which is a feature we’re building. Standard stuff. They wanted to know when their agent costs spiked.Then their engineer shared their screen and said something that's been rattling around in my brain ever since:> "Look at this. Every time our agent handles a refund request, it pulls the customer's entire purchase history, checks seven different policies, and generates a legal audit trail. But when it suggests a product recommendation? Just a simple embedding search. The refund costs us $3.40. The upsell costs $0.03. We charge the same for both..." ## Back to why we built paid Originally, we built Paid to solve a simple problem: AI agents are black holes for money. You need to know what things cost before you go bankrupt. Track costs, set limits, optimize margins. Basic unit economics.But this customer had figured out something else entirely which really made a difference - by instrumenting their costs, they'd accidentally instrumented their entire business intelligence. The cost data was just a proxy for something much more interesting: operational complexity.High cost operations = high value interactions. Low cost operations = commodity features.Their most expensive customers were their most loyal. Their cheapest customers were tourists who didn’t stick around. ## This matters more than we thought Paid sits at a unique point in the stack. Owing to [Paid's agentic signals architecture](/blog/billing/introduction-to-agent-signals-the-key-to-billing-ai-agents), we see every token, every API call, every tool invocation. We built this to track costs, but what we're actually tracking is the entire lifecycle of value creation.When your agent handles a customer request, we see:- How many steps it took- Which tools it used- How many retries and fallbacks- The total resource consumptionBut that's just infrastructure. What we're really seeing is:- Problem complexity- Solution sophistication- Value being created- Trust being built (or destroyed)We thought we were building a billing system for AI agents. Turns out we're building the business intelligence layer for agents. ## The pattern emerged After that call, I started looking at our own data differently. Started asking different questions.The patterns are everywhere once you look. I took the ~250 calls I had recorded and browsed them again (with Claude’s help - projects are a LIFESAVER). ## **Pattern 1: Complexity correlates to commitment** Users whose agents run complex, multi-step operations stick around.Sure, the operations are complex, but those complex operations mean they've integrated deeply into actual workflows.They're not testing. They're *depending*. ## **Pattern 2: Cost spikes are often a type of PMF** When customers complain about costs, that's good. It means they're getting value.When they don't use enough to even notice costs? They're already gone, they just haven't canceled yet. ## **Pattern 3: The Margin story can be a distraction** True, everyone's freaking out about negative margins. We do too.But margins only matter if you're pricing on costs.If you price on value, and you can see the value in the operations data, margins become a solved problem. See our [complete guide for AI agent monetization](/blog/ai-monetization/the-complete-guide-to-ai-agent-monetization). ## We’re sitting on a goldmine This is the part where I'm supposed to be humble, but fuck it.We're sitting on something special.Every other tool in the agent stack sees only their slice:- Model providers see tokens in and out- Orchestration frameworks see flow control- Observability tools see errors and latencyWe see the money. And money is the ultimate truth-teller. When someone routes a request through Paid, we don't just track what it costs. We see:- What they were willing to pay for- What complexity they were willing to tolerate- What value justified that costYes, we built a cost tracker.But costs get tracked through signals of operations.And operations are just business logic made visible. ## Did you know that you have this data too? Every AI company is sitting on this data, but many don't know it or can’t access it.They're so focused on the AI race - better models, lower latency, more features - that they're missing the business intelligence goldmine in their own logs.Your agent's runtime is telling you:- Which features actually matter- Which customers will upgrade- Which use cases have product-market fit- What your business actually does (vs what you think it does)But most teams are just looking at costs and trying to make them go down.That's like having a gold mine and only caring about your electricity bill. ## The agent economy needs more intelligence We started Paid because the agentic economy needed basic financial infrastructure. Track costs, manage margins, don't go broke.But what we discovered is that the cost layer is actually the intelligence layer.Every dollar spent is a decision made.Every API call is a value proposition.Every operation is a business process made visible.We built a cost tracker. It turned into a crystal ball.And that customer who started this whole revelation? They've cut costs by 60% while increasing prices by 3x. Not because they optimized their infrastructure. Because they finally understood what their customers actually valued.They're not paying for AI. They're paying for outcomes. The operations data shows you exactly which outcomes matter.That's the real insight: In the agentic economy, your costs aren't your problem. **They're your roadmap.***We're building this into Paid now. If you want to see what your operations are really telling you, we should talk.* --- ### [2025-08-27] The platform advantage playbook: SaaS leaders can win the agent revolution - URL: https://paid.ai/blog/ai-agents/the-platform-advantage-playbook-saas-leaders-can-win-the-agent-revolution - Date: 2025-08-25 - Author: Arnon Shimoni - Summary: Learn how established SaaS companies can leverage their platform advantages to dominate the agent revolution. Insights from leaders at Box, Salesforce, HubSpot & more on hybrid pricing, ARR per employee, and transforming from seats to outcomes. After analyzing conversations with industry titans like Jack Altman, Rob Litterst, Aaron Levie, Pat Grady, and other pioneering leaders, a clear pattern emerges: the transition from SaaS to agents represents the biggest opportunity for established platforms since the cloud migration.Here's your strategic playbook for navigating this shift. ## The seat model is dying, but it’s evolving Rob Litterst from PricingSaaS (formerly HubSpot) frames it perfectly: "The seat model is still so well suited for the human part of the interaction of agents". He also believes seats are no longer value levers.**Action item for leaders:** Creating a hybrid model where human creativity meets agent productivity is the key here.The winning formula is a hybrid pricing model, with base platform access fees + outcome or action-based agent pricing.Think of it as your "value ladder":- Tier 1 is DIY with tools- Tier 2 is assisted with AI- Tier 3 is fully automated outcomes.Confluence's new "automation run rules" metric shows how traditional SaaS companies can take baby steps toward outcome-based pricing without shocking their customer base. ## Your platform is your moat if you use it right Aaron Levie's insight from Box is crucial: "AI agents give you a reset moment... there's probably two to three times more categories now than there was in the SaaS world, because AI agents much more approximate what a person does".He also says you'll find 10x more opportunities than just replacing a cost center:For SaaS leaders, this means your existing platform isn't a liability but rather your launching pad. You have:- **Data gravity**: Years of customer data that new entrants can't replicate- **Trust relationships**: Established security, compliance, and procurement approvals- **Integration depth**: Already embedded in customer workflowsAs Salesforce's AgentForce and Intercom's outcome-based pricing experiments show, incumbents who move boldly can leverage these advantages to own the agent transformation in their categories. ## The ARR Per Employee Revolution Amos Bar-Joseph's has a different take on ARR/employee: "The North Star for an autonomous business is ARR per employee". Meaning, not just a metric but rather a complete change in how the company should be set up.Jack Altman observes companies hitting $10M ARR with 5 people instead of 50-100 in his investments.One great example of that, from Paid’s customers, is Jim Cutillo from Alpha7x who demonstrates this with their outcome-based pricing model: "We're aligned right out of the gate. We don't charge seat pricing... we're truly making money only if we're saving them".**Action item for leaders:** Start tracking and optimizing ARR per employee now. Set targets that assume 10x productivity gains. Build compensation models that reward efficiency over headcount growth. ## The innovator's solution: cannibalize yourself first Shawn Harris from Coworked, one of our first podcast guests articulates the challenge perfectly: "When you are building something like this, the value that you are creating is far greater than the value that you're currently capturing from your SaaS solution. And so if you're charging eight bucks a month per seat or 15 or 20 or 25. Literally, the value that an agentic solution is creating could be 10 to 100X those values". He goes on to explain the dilemma: "And so do you do that and then go back to those same customers go, Hey, instead of it being 15 bucks, it's now 1500 or it's whatever 150, even if you wanted to kind of bring it a little bit closer per month per seat and see what they say to you".But look at Microsoft's successful transition from licensed software to SaaS. Painful but necessary.Bryce Maddock from TaskUs (a BPO with over 60,000 employees) is actively using AI agents to disrupt their own BPO business: "We're trying to live the innovator's dilemma, disrupting ourselves and going out and discovering new revenue streams by actually cannibalizing our old revenue streams".**Action item for leaders:**- Create a separate AI/agent division with different metrics- Price agents at 10-20% of human equivalent cost initially- Let early adopters experiment while maintaining your base business- Use your platform advantage to ensure agents integrate seamlessly ## Speed is a competitive advantage Wade Foster from Zapier emphasizes obsessive customer focus and speed: "How much do you obsess about your customers and how quickly do you work to serve them? You better hope you're near the top of the list on that dimension".Pat Grady from Sequoia notes that the "uncomfortable truth is the greatest moat by far is the founders". Their speed of execution and willingness to iterate. For established SaaS companies, this means:- Ship agent features weekly, not quarterly- Let customers vote with usage, not committees- Accept that 80% quality delivered today beats 100% delivered next year ## Bridging the credit model Adam Schoenfeld's approach at Keyplay shows the path: Platform access (traditional pricing) + credit-based agent work. This allows you to:- Maintain predictable revenue from base subscriptions- Scale revenue with value delivered through agents- Avoid the binary choice between seats and outcomesRob Litterst again advocates for hybrid models: "A base fee for capabilities and then a consumption variable fee". This gives customers predictability while allowing unlimited upside as agent usage grows. ## Playing Offense With Your Platform Advantage Jaspar Carmichael-Jack's success with Artisan shows how fast the ground is shifting.But established platforms have unique offensive plays:**1. Distribution Power**: Your existing customer base is your agent testing ground. As Nick Mehta from Gainsight notes, you're moving from "selling tools to selling outcomes" – and your customers already trust you.**2. Category Expansion**: Use agents to enter adjacent categories. Your CRM can become sales coaching, your marketing platform can become content creation, your support tool can become customer success.**3. Enterprise Readiness**: Sahil Mansuri reminds us that "if you've sold to the Fortune 500, you know you need a shitload of salespeople." Your enterprise relationships and procurement processes are massive advantages that agent-native startups will take years to build. ## The practical SaaS to AI roadmap **Month 1: Foundation**- Launch pricing experiments with friendlies- Start tracking ARR per employee as a north star metric**Month 2-3: Acceleration**- Ship an agent that handles a complete workflow end-to-end- Price it at 20% of human equivalent cost- Let power users push boundaries while maintaining base platform**Month 3-4: Expansion**- Open agent marketplace for third-party developers- Launch outcome-based pricing pilots- This is important: double down on what's working, kill what isn't**Month 5-6: Transformation**- Announce company-wide shift to ARR per employee targets- Position as the "platform + agents" leader in your category ## You don’t have to pick a side The agent revolution isn't about choosing sides between SaaS and AI.It's about leveraging your platform advantages to deliver unprecedented value. As Aaron Levie puts it, we're in the "2008 moment" of this disruption, the patterns are clear, the architecture is emerging, and the winners will be those who move fast while leveraging their existing strengths.Your platform, your data, your customer relationships - these aren't relics of the SaaS era. **They're the foundation of your agent strategy**. The companies that win won't be those with the best AI, but those who best combine human creativity, platform leverage, and agent automation to deliver outcomes that matter.The revolution is here. Your platform advantage is real. The only question is: how fast will you move? --- ### [2025-08-20] The SaaS Platform Play of the Decade: Your Agent Economy Advantage - URL: https://paid.ai/blog/ai-agents/the-saas-platform-play-of-the-decade-your-agent-economy-advantage - Date: 2025-08-20 - Author: Arnon Shimoni - Summary: Discover how SaaS companies can capture the $34B agent economy opportunity by becoming the infrastructure platform for AI agents. Learn the 3-layer strategy that turns AI disruption into 10x revenue growth through identity management, orchestration, and intelligence dashboards. Salesforce unlocked $34 billion in revenue (FY2024) selling a database with forms on top with a new business model - that being SaaS. Now they're about to do it again but this time, the opportunity is 10x bigger.Sure, lots of competition on the agent front - but Salesforce is building the platform every agent will run on. They're not competing with AI startups. They're collecting rent from all of them with AgentForce.And you can do the same thing. ## Platforms always win Microsoft didn't fight the mobile revolution. Well, they tried for a bit - but then they became a cloud infrastructure every mobile app runs on. Azure is now their biggest revenue driver - bigger than Windows ever was.Amazon didn't compete with every online store. They built the platform every store runs on. AWS makes more profit than their entire retail business.The pattern is - when a new technology creates chaos, build the infrastructure that brings order.AI agents are creating chaos. You get to bring the order. ## Your unfair infrastructure advantage Here's what every AI startup is discovering the hard way: Agents need exactly what you already provide. **Distribution.**- **Identity management.** Every agent needs to prove who it is.- **Permission systems.** Every agent needs to know what it can access.- **Audit trails.** Every agent action needs to be logged and tracked.- **Integration layers.** Every agent needs to connect to existing systems.- **Compliance frameworks.** Every agent needs to operate within regulations - even if those regulations aren’t clear yet.You've spent decades perfecting these capabilities for humans. Agents need the same systems, just at 100x the scale.You now get to scale what you already dominate. ## This would be your 10x revenue multiplier Workday [manages 70 million humans](https://fortune.com/2025/04/01/workday-fortune-500-top-5-business-lessons/) across their customer base today.By 2027, those same companies will deploy 700 million AI agents, ten digital workers for every human employee.Same customers. Same trust relationships. Same enterprise contracts.But 10x the seats to manage.Each of those agents needs:- Identity verification ($3/month)- Access management ($5/month)- Performance monitoring ($10/month)- Compliance reporting ($15/month)That's not disruption. That's the biggest expansion opportunity in your company's history. ## A three layer platform strategy ## Layer 1: The Identity Foundation Your existing user management system becomes the identity layer for the entire agent workforce.Every AI agent gets registered, authenticated, and tracked through your platform. Just like human employees, but automated.**Revenue opportunity:** $5-15 per agent per month for identity and access management. ## Layer 2: The Orchestration Engine Instead of agents calling APIs directly, they route through your platform. You become the nervous system connecting every agent to every system.This isn't just billing—it's governance, compliance, and optimization at enterprise scale.**Revenue opportunity:** $0.01-0.10 per agent action, with millions of actions per day per customer. ## Layer 3: The Intelligence Dashboard Enterprises need to understand what their digital workforce is doing. You provide the visibility, analytics, and control plane.Think of it as "Manager Tools for AI"—helping humans oversee, direct, and optimize their agent teams.**Revenue opportunity:** $25-100 per human manager per month for agent oversight tools. ## Companies already doing the right thing **Salesforce** launched AgentForce to become the platform where all sales agents operate. They're aiming to orchestrate every B2B transaction in the economy.**Slack** announced dedicated workspaces for AI agents (built on AgentForce). That means your bots need licenses too. They're positioning as the communication infrastructure for human-agent collaboration.**Microsoft** turned Copilot into an Azure revenue driver. Every AI interaction in Office creates billable cloud consumption.They're not fighting AI and they’re not saying all their agents will be home-grown - but they're becoming the infrastructure that makes AI possible at enterprise scale. ## The trust multiplier effect We hear from the slower enterprise companies that they will spend the equivalent of $2m in getting products like Slack or Teams approved for internal use in time spent, RFCs, etc.. Two million dollars. For internal comms.You think they're going to repeat that security review for every AI startup? For every new agent platform?No. They're *eventually* going to say: "Use whatever agents you want, but they have to run through our approved platforms."You have all your audits, certifications - you have their trust and the earned contracts.Read that again: That's not just ***a*** competitive advantage - it’s your extremely defensive moat in the agent economy. ## The three-person company of 2027 Here's what Amos Bar-Joseph told us on our podcast: "An autonomous business doesn't have 10 different roles under the go-to-market umbrella. It has only one - revenue creator". In his autonomous business concept, the future company has just three types of people:- **Agent Creators** build AI armies to amplify human potential. They're not simple prompt engineers or even context engineers, they're digital workforce architects.- **Product Creators** handle development and architecture. They build the systems that agents enhance and humans depend on.- **Revenue Creators** own sticky revenue growth. Not sales, not marketing, not customer success but pure revenue creation and expansion.That's it. Three roles. Everything else gets automated or eliminated.Your SaaS platform becomes the infrastructure that makes this possible. You're not just serving today's bloated org charts - you're enabling tomorrow's lean, autonomous businesses. ## Your platform future as a SaaS company By 2027, you won't just be a SaaS company. You'll be:**The Operating System** for enterprise AI - every agent runs on your infrastructure.**The Control Tower** for digital workforces - every agent action gets managed through your systems. Your people will manage multiple agents. [Paid has a "control tower" feature to help manage multiple agents from a variety of vendors.](/product/cost-tracking)**The Revenue Engine** for the agent economy - every successful AI deployment expands your platform. [Paid can help here too with extensive agentic monetization models](/product/billing).Yes, you need to disrupt yourself or someone else will disrupt you. Become the infrastructure that defines the next decade of enterprise software.We're living through a transition right now and you can still act. You already have everything you need. The trust, the relationships, the infrastructure, the expertise.You just need to think bigger.Welcome to the platform play of the decade.***Your*** agent economy starts now. --- ### [2025-08-18] Your AI Agents Are Losing Money Every Second They Run - URL: https://paid.ai/blog/ai-monetization/your-ai-agents-are-losing-money - Date: 2025-08-18 - Author: Arnon Shimoni - Summary: AI companies are scaling into bankruptcy with 50-60% gross margins while traditional SaaS enjoys 70-80%. Most AI-native companies fly blind on unit economics, optimizing for growth without understanding which agents generate profit. You built an AI product. Your customers love it. Usage is exploding.Then your AWS bill arrives.That customer paying $500/month? They just burned $6,000 in compute. Your "successful" AI feature is now your fastest path to bankruptcy.This isn't hypothetical. It's happening right now to AI companies that can't answer one simple question: **What does each agent interaction actually cost?** ## Flying blind? You’re not alone Traditional SaaS taught us that more usage equals more value. Add another user, marginal cost approaches zero. Beautiful unit economics.AI flipped this completely.Every agent interaction costs real money. Every prompt burns compute. Every workflow triggers cascading API calls. And unlike SaaS, these costs compound exponentially with usage.**The Math That's Killing AI Companies:**Business ModelMonthly RevenueMonthly CostsGross MarginUnit EconomicsTraditional SaaS$1,000$5095%Profitable at scaleAI Company (Light Usage)$1,000$40060%SustainableAI Company (Average Usage)$1,000$1,200-20%Losing moneyAI Company (Heavy Usage)$1,000$3,000-200%Bankruptcy pathYou're literally paying customers to use your product.It’s true that even most AI companies track revenue religiously but have zero visibility into agent-level costs. They know what customers pay but they have no idea what customers cost.This creates three deadly problems:**1. You can't identify which agents hemorrhage money**Some agents run simple workflows. Others orchestrate complex multi-model chains. Without granular cost tracking, they all look the same on your dashboard.**2. You can't price anything correctly**How do you price a feature when you don't know if it costs $0.01 or $10 per use? Most teams guess. Then they scale. Then they discover they guessed wrong.**3. You can't optimize what you can't measure**That agent making 50 API calls? Maybe it only needs 5. But without visibility, you'll never know you're burning money on redundant operations. ## The Midjourney model: $18M per employee through margin mastery While most AI companies struggle with razor-thin margins, outliers like Midjourney demonstrate what's possible when you achieve surgical precision in cost management: **$18 million in revenue per employee.**This isn't luck. This is the result of understanding exactly what each AI operation costs and optimizing relentlessly around those economics. Midjourney's extraordinary efficiency comes from treating every inference, every compute cycle, every API call as a measurable economic event.The companies achieving these extraordinary efficiency levels share one critical capability: **real-time visibility into their cost structure at granular levels.** ## When every interaction is an economic event Here's what makes AI economics uniquely dangerous:**Variable Costs at Every Layer:**Cost ComponentRange per UnitPricing UnitPredictabilityLLM Inference$0.015-$0.060Per 1M tokensMediumVector DB Queries$0.001-$0.010Per searchHighExternal APIs$0.01-$1.00Per callLowGPU Compute$0.10-$5.00Per hourMediumMemory/Storage$0.05-$0.20Per GB/monthHighEmbeddings$0.0001-$0.002Per 1K tokensHighFine-tuned Models$0.120-$0.360Per 1M tokensMedium**The Multiplication Effect:**One customer conversation might trigger:- 3 LLM calls (context, processing, response)- 5 database queries (RAG retrieval)- 2 external API calls (data enrichment)- 10 vector searches (similarity matching)Workflow StepService UsedCost per CallCalls per ConversationTotal CostContext LoadingGPT-4$0.031$0.03RAG RetrievalPinecone$0.005$0.01External DataThird-party API$0.052$0.10ProcessingGPT-4$0.061$0.06Vector SearchEmbedding API$0.0010$0.01Response GenerationGPT-4$0.121$0.12Post-processingClaude 3$0.081$0.08Conversation MemoryStorage$0.021$0.02Total--22$0.43Suddenly that "simple" chat costs $0.43. Customer has 1,000 chats daily? That's $430/day in direct costs. On a $50/month plan. ## The Agentic Margin Ratio: Your only metric that really matters for agents Forget ARR. Forget growth rate.If you can't calculate your *Agentic Margin Ratio* (AMR), you're running blind.The AMR is defined as the profit of your agent divided by it's total revenue, or **AMR = (Agent Revenue - Agent Costs) / Agent Revenue**In order to calculate your AMR, you need to know:- Exact compute costs per agent interaction- API consumption by workflow- Infrastructure allocation by feature- Token usage patterns by customer segmentMost companies can't answer any of these. They're optimizing for growth while their economics implode underneath.The AMR benchmark example from one Paid's customers:Workflow StepService UsedCost per CallCalls per ConversationTotal CostContext LoadingGPT-4$0.031$0.03RAG RetrievalPinecone$0.005$0.01External DataThird-party API$0.052$0.10ProcessingGPT-4$0.061$0.06Vector SearchEmbedding API$0.0010$0.01Response GenerationGPT-4$0.121$0.12Post-processingClaude 3$0.081$0.08Conversation MemoryStorage$0.021$0.02Total--22$0.43 ## Cost tracking matters for agents Companies that survive the AI transition share one capability: they know what every agent interaction costs in real-time.This is, like with other measures, a spectrum. You likely have something through some of your providers (e.g., a monthly aggregate from OpenAI's dashboard) - but you don't have real-time per-customer.That's fine - but you need to understand where you are and where you need to get to.Maturity LevelTracking CapabilityOptimization SpeedTypical MarginsSurvival RateLevel 0No trackingNever-50% to -200%Level 1Monthly aggregatesQuarterly-20% to 0%25%Level 2Daily reportsMonthly0% to 20%50%Level 3Hourly dashboardsWeekly20% to 40%75%Level 4Real-time per interactionDaily40% to 60%90%Level 5Predictive + real-timeContinuous60%+95%This means:- **Granular instrumentation** of every model call- **Cost attribution** to specific customers and workflows- **Real-time dashboards** showing margin by feature- **Automated alerts** when costs spike- **Optimization loops** that reduce expense systematicallyWithout this, you're not running a business. You're running a charity for your cloud providers. ## Surgical precision in cost management is the future of agentic monetization While your competitors fly blind into margin destruction, Paid provides the surgical precision in cost tracking that separates survivors from casualties in the AI economy. ## Real-time cost tracking across your agentic stack **Your AI agents are spending money every second they run.**Paid's agentic monetization stack tracks it all:- **Granular LLM monitoring** across OpenAI, Anthropic, Mistral, ElevenLabs, Vapi and dozens of other providers- **API cost attribution** for every third-party integration- **Infrastructure mapping** connecting compute costs to specific agents- **Vendor bloat elimination** by identifying unused providersWe know you can't optimize what you can't measure, so we also provide- **Real-time profit calculation per agent and per customer**- **Agentic margin ratio tracking** with industry benchmarks- **Cost spike alerts** before they destroy your unit economics, captured automatically through our Open Telemetry-based SDK wrappers- **Workflow profitability analysis** to optimize your highest-value features- **Vendor cost comparison** to negotiate better rates ## **Your margin discipline is everything** While your competitors scale blindly into bankruptcy, you can scale with the confidence that comes from knowing exactly what every agent costs and exactly what every customer pays. --- ### [2025-08-13] The SaaS competitor's agent is coming - URL: https://paid.ai/blog/ai-monetization/the-saas-competitors-agent-is-coming - Date: 2025-08-13 - Author: Arnon Shimoni - Summary: It's do or die We’ve been hearing from around the world - SaaS boards want an "AI strategy". Now you’re adding chatbots, adding “AI” to landing pages, and pitching "AI-powered insights."Meanwhile, three university dropouts in a trenchcoat are building an agent that replaces your entire product. And they'll be charging $9 for what you charge $900.Six months from now, your enterprise customers will ask why they need 47 DocuSign seats when an agent can handle every contract. Why they need ZoomInfo's $30k package when an agent can enrich leads for pennies. Why they need RingCentral at all when an AI handles every call.You'll have a great answer about security, compliance, and enterprise features.They won't care. ## The math that will kill your SaaS Let's use DocuSign as an example.It’s a great product with $2.7 billion in revenue. Average enterprise contract estimated at $48,000/year, or so I’m told.Here's what a document signing agent costs to run:- PDF parsing: $0.20- Signature verification: $0.001- Workflow orchestration: $0.003- Audit trail generation: $0.001- Total cost per document: $0.007A startup can charge $99/month for unlimited signatures and still have 94% margins.Your customers aren't stupid. They can do this math too. Even if they pay for other things than just the PDF signing aspect. ## **Chart 1: The 90% Price Collapse Is Already Here** ## **Your moat could sink you** > "But we have enterprise features! Compliance! Integrations!"Sure, cool story.Let’s look at some examples:- **ZoomInfo's moat**: 300 million contact profiles, painstakingly verified.- **Agent replacement**: Real-time LinkedIn scraping + email verification. Cost: $0.03 per contact.- **Calendly's moat**: Beautiful scheduling UI, calendar integrations.- **Agent replacement**: "Find me time with John next week." Done. No UI needed.- **Zendesk’s moat**: Decade of conversation data, sophisticated routing.- **Agent replacement**: Agent that actually solves problems instead of routing to humans.This moat could end up protecting you but also drowning you.The AI-native competitor doesn't need to rebuild your features. They need to solve your customer's problem. Those aren't the same thing. ## **AI enhancement is a lie** Every SaaS company has the same playbook right now:- Add ChatGPT to existing workflows- Call it "AI-powered"- Increase prices 20%- Hope nobody notices it's lipstick on a pigThis isn't transformation. It's decoration.Company typeTheir "AI" featureAgent alternativeCustomer satisfactionCRMSuggested email templatesFully autonomous SDR12% vs 84%HR softwareCV keyword matchingComplete hiring agent23% vs 91%AnalyticsAI insightsAnswers any question instantly31% vs 88%Project managementSmart task assignmentSelf-completing projects18% vs 79%Support deskSuggested responsesResolves tickets autonomously28% vs 93%If your AI innovation is suggested email templates, while a 5-person startup built an agent that completely replaces sales outreach, …You get the picture. ## **The cannibalization dilemma** Here's your actual choice:**Option A**: Protect your existing revenue model- Keep charging per seat- Add incremental AI features- Maintain 70% margins- Die slowly as customers churn to agents**Option B**: Blow up your own business model- Build agents that replace your product- Charge 90% less- Destroy your own margins- Maybe survive the transition**There is no Option C. This is do or die.**The brutal truth? If you're not willing to make your own product obsolete, someone else will do it for you. And they're already building. ## **What Salesforce understands that you may not** Marc Benioff launched Agentforce. Not "Salesforce with AI", not "AI-powered CRM"… Agents. That replace human work. That threaten their own seat-based model.They're charging $2 per conversation instead of $125 per user per month.Salesforce told the market: "Our own pricing model is obsolete"They're cannibalizing a $30 billion revenue stream. Because they know if they don't, someone else will.Your categoryAgent startups buildingFunding raisedTime to feature parityIf you're Calendly14 companies$47M totalAlready thereIf you're DocuSign8 companies$31M total3 monthsIf you're HubSpot31 companies$280M total6 monthsIf you're Monday.com12 companies$67M total9 monthsIf you're SalesSorce43 companies$520M total12-18 monthsIf you're Zendesk22 companies$156M totalAlready there ## **How we see the market in 3 types** I see the market split across three types:**Type 1: The Deniers** (45% of market)"AI is just hype. Our enterprise customers value stability."Timeline to irrelevance: 18 months**Type 2: The Decorators** (40% of market)"Look, we added AI! There's a chatbot now!"Timeline to irrelevance: 24 months**Type 3: The Destroyers** (15% of market)"We're building agents that make our current product worthless."Timeline to relevance: IndefiniteWhere are you on this path? ## **If you're an incumbent SaaS - here's your escape route** Before you panic and fire your entire product team, I have some math for you.Your board may keep asking about protecting seat revenue and ARR, but they're asking the wrong question.- **Seats in your customer's org**: These are going down. Every quarter. Forever.- **Tasks they need completed**: Going up. Exponentially. No ceiling.A 50-person company used to have 40 potential Salesforce seats. **Maximum**.That same company now has 50,000 potential customer interactions that need handling. 10,000 documents to process. 100,000 data points to analyze.The constraint has flipped. You're no longer limited by headcount. You're limited by imagination. ## **What would work for you:** **Step 1: Stop Counting Seats, Start Counting Outcomes**Salesforce didn't launch "Agentforce" as a feature, but as a shift from "per user per month" to "per conversation." That's not a pricing change!Your version:- DocuSign: Stop charging per user. **Charge per contract value processed.**- ZoomInfo: Stop charging per seat. **Charge per qualified opportunity generated.**- Intercom: Stop charging per agent. **Charge per issue resolved.**The beautiful part? A 50-person company might have had 3 support agents. But they have 3,000 support tickets. You just 1000x'd your addressable market.Think in outcomes!**Step 2: Become the orchestration layer and embed in their workflows**You can't compete with agents on cost. But you have something they don't: **Trust**, **compliance**, and a decade of edge cases.Orchestrate the agents or AI…For example:- Let startups build specialized agents- You become the governance layer- Charge for orchestration, compliance, and quality assurance- Take a cut of every agent transaction through your platformThink in terms of the App Store, not apps. Think Shopify, not shops….**Step 3: Price for Abundance, Not Scarcity**The old model assumed *scarcity*:- Limited seats = premium pricing- More users = more revenue- Growth tied to customer headcountThe new model assumes *abundance*:- Unlimited tasks = volume pricing- More automation = more revenue- Growth tied to customer successHere's what this looks like:Old modelNew modelRevenue multiple10 seats x $100 = $1,00010,000 tasks x $0.50 = $5,0005x50 seats x $100 = $5,000100,000 tasks x $0.20 = $20,0004x200 seats x $100 = $20,0001M tasks x $0.10 = $100,0005x ## **Agents aren’t your replacement - they’re your multiplication factor** Every agent that replaces a human creates 10x more work that needs governing, monitoring, and orchestrating. Every automated workflow creates 10x more data that needs analyzing. Every AI interaction creates 10x more complexity that needs managing.The companies that win won't be the ones that fight agents. They'll be the ones that help customers deploy 1,000 agents safely. ## **Get it together** **Accept reality now**- Run the unit economics on outcome-based pricing- Identify your top 10 customers' task volumes- Model revenue at $0.10-1.00 per outcome**Pick your battlefield**- Choose: Build agents, orchestrate agents, or govern agents- Launch a pilot with your most innovative customer- Price it at 10x current per-seat equivalent**Burn the boats**- Announce the new model publicly- Give customers 12 months to transition- Show Wall Street the multiplication math ## **The Choice Is Still Binary** But it's not the choice you think.> You don’t get to keep seats. You need to embrace abundance, and not defend scarcity.*Seats are finite. Tasks are infinite.*You can be Blockbuster, clutching your late fees while Netflix ships DVDs.Or you can be Netflix, burning your DVD business to build streaming.There's no middle ground. No hedging. No "wait and see."Build the agent that kills your product. Or watch someone else do it.Choose wisely.What's your move? --- ### [2025-08-11] You need AI cost tracking, and you should start now - URL: https://paid.ai/blog/ai-monetization/you-need-ai-cost-tracking - Date: 2025-08-11 - Author: Arnon Shimoni - Summary: "We basically look at our monthly OpenAI bill and apply a distribution factor" > "We basically look at our monthly OpenAI bill and apply a distribution factor."Sound familiar? You're not alone.I analyzed hundreds of our customer conversations about AI agent monetization. The second most common pain point after pricing? *Cost tracking.* Or more accurately, the complete lack of it.Way too many people have no idea what their AI is racking up in costs, and you see it on Reddit, X, and in communities everywhere.According to CloudZero, the average monthly AI spend is jumping from $62,964 to $85,521 in 2025 - a 36% increase - but I actually think they’re underestimating it. ## Why it gets harder at 10 customers Imagine yourself in this person’s shoes who told us they track globally:> *"We have separate API keys per customer so that we can track usage per customer. But it doesn't give you the dollar figure. It gives you tokens or usage. Then we just basically look at our monthly bill at OpenAI and apply a distribution factor."*This works when you have 5 customers. Maybe even 10. But here's what happens at scale:- **At 5 customers:** You're checking OpenAI dashboard daily. Life is good.- **At 10 customers:** You've got separate API keys. Maybe a spreadsheet. Still manageable.- **At 50 customers:** Welcome to the Spreadsheet Spiral of Death.By the time you figure out a customer is unprofitable, they've already burned through months of margin. ## The API key problems: What OpenAI and Anthropic won’t tell you about usage These foundation model providers’s usage dashboard is built for developers, not businesses.One founder told us:> "There are all these vendor tools, right? And certain discounts based off bundles. Pricing is very fluid, and so that cost may change. But right now it seems like it's a static view."**What you get:**- Token counts by API key- Monthly aggregate spend- Basic usage graphs**What you really need:**- Cost *per* *customer, per action, per outcome*- Real-time *margin alerts*- Workflow-level *profitability*- Which features are *margin killers*The gap between what you need and what you get? That's where companies die. ## Building a simple cost attribution Before you spin up a massive data warehouse project - get practical. You need cost visibility TODAY, not in 3 months.Here's the progression that works: ## Level 1: Just ship it Start tracking these 5 signals immediately - you can pull this off in an hour!In Python it’ll look something like:**Pro tip from Paid:** Even a CSV beats nothing. Perfect is the enemy of shipped. ## Level 2: Attribution magic You wrap every AI call:In Javascript, it’ll look something like: ## Level 3: Get a purpose-built system > *"… this is crushing us. Our cost has gone up, and it's driving our margins down."*We’ve seen customers go from 70% margins to 40% because ONE customer was "very chatty" with their voice agent.This is where [Paid’s AI cost tracking and margin management](/product/cost-tracking) can help.Don’t build this yourself - let someone else handle the details and stay up-to-date. ## One of our customers had negative margins… A founder walked into our office (figuratively - we met online) and told us “we're bleeding money and I don't *really* know why."They have a clean UI (not shadcn!), happy customers, $299/month pricing that looked profitable on paper. They were celebrating hitting 100 customers when their CTO dropped a "we're losing money on every single call"…**We helped dig in** to their voice bills where most customers had totally normal usage - but then we found a really chatty customer.- Average customer: 2,000 voice tokens/month ($8 in costs)- Chatty customer: 187,000 voice tokens/month ($748 in costs)- The $299/month became -$449/month**The fix **was to introduce complexity factors. More complex workflows cost more, and use the higher-value LLMs. Once a certain threshold has been reached, the models were downgraded. ## Your path forward Every day that you operate without cost visibility is a day you're potentially losing money. But you don't need a perfect system to start.**Today:** Implement basic signal tracking. Even a CSV is better than nothing.**In a few weeks:** Add customer-level attribution. Know who's killing your margins.**In a couple of months:** Graduate to real-time monitoring. Catch problems before they compound.Stop guessing based on your vendor bills - start tracking, start managing, start profiting.Your margins will thank you. --- ### [2025-08-07] AI pilots can be a nightmare - URL: https://paid.ai/blog/ai-agents/ai-pilots-can-be-a-nightmare - Date: 2025-08-07 - Author: Arnon Shimoni - Summary: When founders admit their pilots aren't great - real solutions emerge. Here's what nobody talks about when running an AI startup: We're all secretly running the same failed experiment. Over and over. And calling it "customer validation".Yesterday I discovered I wasn't alone in this particular flavor of founder hell. Neither were the other 15 people staring back at me through Zoom, each nursing their own pilot horror stories like war wounds. ## The question that started everything "Quick show of hands," I asked, already knowing the answer. "Who has a pilot that's been running for more than 30 days?"The silence hit first. Then, slowly, hands started creeping up. We weren’t judging, but recognizing. ## When "Interesting" becomes a four letter word > “Interesting is that red flag at least for us and talking about the mechanics that's besides the point. Let's course correct and identify the business problem and why that is needed to be solved now rather than later.”(This rings familiar from [The Mom Test](https://www.momtestbook.com/))Martin from Signify had just finished a paid pilot - actually satisfied every success criteria in the contract. The business team loved it. Then IT showed up with "different opinions," and suddenly the work became a very expensive case study.Chris from Quarq jumped in with a metaphor - It's like customers wanting to get in shape before going to the gym. They're preparing to prepare. Meanwhile, we're supposed to be the personal trainer saying 'No, you come as you are. That's the whole point’. ## The $200K solution that we don’t know how to price Others had a different dilemma: Replacing analysts who cost $200K a year. But when the price is named, they look at me like I’m crazy.This seems obvious until you realize what's actually happening. We're not selling software anymore. We're selling organizational transformation, and nobody (not us, and not them) knows what that's worth.Krishna added "I tell prospects they'll make $1,000 more per month. They look at me like 'okay, but why would I want that?' The value is so obvious we literally can't explain it."The counterintuitive truth is that making the ROI too clear actually makes it less believable. ## The Two Words That Changed Everything > "As soon as we said the word AI, because we're in a highly regulated business, we deal with business with banks and large financial institutions. Then I got down this track of, we don't even have an AI governance strategy yet in place for something like that. Right. So I kind of got off the rails because I was calling an AI agent. I literally changed it to a digital worker. Didn't even talk about AI."Wait, that's it? Just terminology?Jim continued: "They had 15 people doing an OFAC check. So that's checking people... They had 15 people in a room doing this every day. Right. And with our agent, they can do it with three people." ## The Uncomfortable Mirror Marcus had been observing from a different angle -he helps enterprises evaluate AI solutions. His perspective was interesting:> "You know what your buyers are really thinking? They have no idea what their organization looks like in 12 months if they adopt your solution. You're not selling them software. You're selling them a future they can't visualize. And futures are expensive."We've all been there. Painting visions of transformation while our prospects are just trying to make it through Q4. ## Workflows beat solutions > "Really trying to demonstrate how you own the workflow, not the solution. Like the solution, the solutions are a dime a dozen out there".This resonated with me as this is something we’ve spoken about before: ## My final takeaway Let me tell you what I personally learned hosting this beautiful mess:Every founder thinks their pilot problems are unique. They're not. We're all fighting the same battle - proving value for something that's never existed before while speaking a language our buyers don't understand.Sure, you can build better tech - but you need to learn to translate that to value, from CTO to CFO. From features to headcount. From "AI transformation" to "digital workers".Marcus said it best as we were wrapping up:> "Stop selling them vision. Sell them headcount reduction. The vision comes after they see the results."I leave you with this final clip - AI and agentic pilots are probably just normal pilots.As people started dropping off, the chat had some of the commitments:- "Tell our customer which person should be the POC owner"- "Qualify the potential pilot better by looking at questions and asking better questions"- "Qualify the Jokers"- "Longer view/plan how to operationalize it throughout the years instead of (problem, solution)"Nobody tells you about building AI companies: The technology is the easy part. The hard part is admitting our go-to-market is broken and finding others brave enough to fix it with you.The pilot trap is real. **But apparently, so is the escape route. **You just need 15 other founders to help you find it! --- ### [2025-07-29] Why “Per Seat” pricing is failing for AI agents (and what's replacing it) - URL: https://paid.ai/blog/ai-monetization/why-per-seat-pricing-is-failing-for-ai-agents - Date: 2025-07-29 - Author: Arnon Shimoni - Summary: Field Notes: new data from 178 calls with agent builders shows how the best teams price their AI agents + what’s killing early deals. **Every week, we talk to founders and revenue leaders building AI agents.**They’re replacing analysts, SDRs, ops managers, even engineers with products that run continuously and get smarter over time.But when it comes to pricing? Most teams are stuck using SaaS playbooks.That means:- 📉 Flat fees.- 👥 Per-seat pricing.- 🧾 Arbitrary implementation charges.Our data shows that it’s not working.We just had [Rob Litterst ](https://open.substack.com/users/8095891-rob-litterst?utm_source=mentions)on our podcast, and he said the same things:I’ve been in + reviewed **178 recorded conversations** with buyers across agent-powered products - from devtools to finance—and mapped where pricing friction shows up and what high-performing teams do differently.Here’s what I found. ## **What our data says** I tracked **conversation density** across key themes like pricing, value attribution, and margin visibility. Here's what stood out:> **Key spike themes**:- “What are we paying for?”- “How do we prove the agent works?”- “Can we scale pricing with usage?”These patterns repeat across sectors and maturity stages, with almost even splits so these problems meet everyone similarly! ## **The pricing patterns that don’t work** **Per-seat pricing** confused nearly every buyer.They’d ask things like:> “Is this per user or per agent?”> “What happens if it runs 24/7?”> “How do I know I’m not overpaying?”My takeaway: **Flat fees** made early deals easier but blocked expansion.In dozens of renewals, companies delivered $500K to $1M+ in agent output… and were still charging the same $25K/year contract.***That disconnect kills your pricing power!*** ## **The models that do work** The teams that consistently closed and expanded used **hybrid pricing**:Pricing componentWhat it coversSetup feeIntegration and onboardingBase feePlatform access (predictable billing)Usage meteringActual agent work performedOutcome bonus% of savings, revenue or cost avoidedA good example looks like:- $8K/month base (e.g., platform fee)- $20 per strategy executed, $0.50 per task completed- 9% of savings > $100K (revenue share)Line itemFlat fee invoiceValue invoicePlatform access$15,000$8,000/moStrategies backtested-$1,080 (54 @ $20)Cost savings bonus-$9,999 (9% of $100K)**Total****$15,000****$18,080**Customers understood it. They trusted it. **They paid more.** ## **What actually works** From the best teams in our network, here’s what we recommend:✅ **Show the work**Track what the agent does. Log signals. **Share it with customers.****I can’t overstate how important showing the work is.**✅ **Anchor pricing to outcomes when possible**Frame the price in terms of cost avoided, time saved, or money made.✅ **Use a hybrid model**Blend base fees with metered usage and outcome-linked bonuses.✅ **Tie invoices to value**Your invoice should *prove* the agent delivered ROI. If it doesn't, pricing is guesswork. ## **💬 tl;dr** Most agent pricing doesn’t reflect the product. You're not selling seats. You're selling output, work, and value.If you’re using SaaS-style pricing for agents, you’re leaving money on the table and confusing your customers in the process.Need help instrumenting usage, tracking value, or rewriting your pricing model?That’s what we do at Paid.We’ve helped dozens of teams price and prove the value of their agents without adding complexity for their end customers. --- ### [2025-07-23] AI Agents are everywhere: A field guide to what's actually working - URL: https://paid.ai/blog/ai-agents/ai-agents-are-everywhere - Date: 2025-07-23 - Author: Arnon Shimoni - Summary: After working with hundreds of agent companies at Paid, we've seen a clear pattern emerge. The most successful agents aren't trying to do everything for everyone. They're solving specific problems for specific industries. If you're building an AI agent, you're not alone.After working with hundreds of agent companies at Paid, we've seen a clear pattern emerge. The most successful agents aren't trying to do everything for everyone. They're solving specific problems for specific industries.Here's what's actually working in the wild. ## Industry-Specific Powerhouses These agents don't just add AI features to existing workflows. They completely automate entire job functions.**Mortgage processing** companies like [Alpha7x](https://alpha7x.com/) handle the full pipeline from application to approval. One agent, end-to-end automation.**Insurance operations** have massive appetite for agents. Claims processing, underwriting decisions, policy renewals. Companies like [Gradient AI](https://www.gradientai.com/) and [Quandri](https://quandri.io/) are replacing entire departments.**Healthcare workflows** present huge opportunities. Surgery center management, clinical documentation, patient scheduling. The regulatory complexity actually helps create moats.**Legal automation** is heating up fast. Contract generation, document review, paralegal research. [Lawhive](https://labs.lawhive.co.uk/) processes thousands of legal documents daily with their agents.> 💡 Pick an industry with expensive labor doing repetitive work. Build agents that deliver the same outcomes at 70% lower cost. ## Voice Agents Are Taking Off Voice technology finally works well enough for production use. We're seeing agents handle complex conversations across industries.**Restaurant ordering** agents like [Kea](https://kea.ai/) take phone orders during rush hours. No more missed calls during peak dinner time.**Logistics negotiations** through [HappyRobot](https://happyrobot.ai/). Their agents call truckers, negotiate rates, book loads. They're processing thousands of calls daily.**Multilingual customer service** with companies like [Vozy](https://www.vozy.ai/en). One agent speaks 15 languages fluently. Try hiring humans for that.**Automotive dealerships** use [Fuzey](https://www.getfuzey.com/) for customer communication. Service reminders, appointment scheduling, follow-up calls.> 💡 Voice agents work because customers already expect phone interactions in these industries. The friction is lower than you'd think. ## Sales Operations Revolution Sales teams were early adopters. The workflows are predictable, the ROI is clear.**AI SDRs** from [Artisan](https://www.artisan.co/) and [Sailes](https://sailes.com/) handle prospecting, outreach, and follow-up. They're booking qualified meetings at scale.**Sales intelligence** agents like [Aomni](https://www.aomni.com/) provide real-time prospect research. No more generic email templates.**Training simulations** through [Hyperbound](https://www.hyperbound.ai/) let reps practice with AI buyers. They're getting reps ready for complex enterprise deals.> 💡 These agents integrate with existing CRM workflows. Sales teams don't need to change how they work. ## Process Automation Specialists Some of the most profitable agents handle very specific business processes.**Government contracting** with [American AI Logistics](https://americanailogistics.com/). They automate proposal generation for federal bids. One agent, millions in contract value.**Debt collection** through [Bircle](https://bircle.ai/). Multi-channel recovery across email, SMS, and voice. Better outcomes than human collectors.**Procurement automation** with companies like [Kavida](https://www.kavida.ai/). Purchase order processing, vendor management, approval workflows.**Fleet management** through [RubyFleet](https://rubyfleet.io/). Route optimization, maintenance scheduling, driver communication.> 💡 These agents succeed because they solve painful, expensive problems that businesses already understand. ## The Data Intelligence Layer Smart companies are building agents that make sense of business data.**Causal AI** from [Causalens](https://causalens.com/) helps understand cause-and-effect in business metrics. Not just correlation.**Explainable AI** through [ExplainX](https://www.explainx.ai/) makes AI decisions transparent. Critical for regulated industries.**Data storytelling** agents automatically generate reports and insights from raw data.> 💡 The opportunity is huge. Most companies are drowning in data but starving for insights. ## Security and Compliance Security agents are growing fast. The use cases are clear, the stakes are high.**Autonomous penetration testing** with [XBOW](https://xbow.com/). Continuous security assessment without human pen testers.**AI governance** through [Symmetry Systems](https://www.symmetry-systems.com/). Monitor how AI is being used across the organization.**Threat intelligence** agents detect disinformation and security threats in real-time.> 💡 Security budgets are large, and CISOs understand the ROI of automation. ## Developer Tools Technical teams are building agents to automate their own work.**Code generation** through [Engine Labs](https://www.enginelabs.ai/) and [Qodo](https://www.qodo.ai/). Agents that write, test, and deploy code.**Browser automation** with [Magical](https://www.getmagical.com/) and [OpenAI Operator](https://operator.chatgpt.com/). Universal workflow automation across any web application.**No-code platforms** like [OmniMind](https://omnimind.ai/) democratize AI development for non-technical teams.> 💡 Developers building tools for developers creates a natural feedback loop for improvement. ## Sustainability and GreenTech Environmental compliance is driving agent adoption.**Net zero reporting** with [Twin4Green](https://twin4green.com/). Energy management and sustainability metrics.**Solar installation design** through [SuntropyAI](https://suntropy.es/). Automated renewable energy planning.**Digital twins** with IoT integration for real-time environmental monitoring.> 💡 Regulatory requirements make these agents must-haves, not nice-to-haves. ## Marketing and Advertising Marketing agents are solving attribution and personalization challenges.**Privacy-first advertising** with [Crumbless](https://www.crumbless.ai/). Cookieless targeting that actually works.**Programmatic job advertising** through [Wonderkind](https://www.wonderkind.com/). AI-generated recruitment campaigns.**Cold email automation** at scale with [Salesforge](https://www.salesforge.ai/). Personalized outreach without the spam.> 💡 Marketing budgets are under pressure. Agents that deliver measurable ROI get renewed. ## What we’ve learned After working with hundreds of agent companies at Paid, three patterns stand out:**Narrow focus wins** - the most successful agents solve one specific problem extremely well. Don't try to be everything to everyone.**Industry expertise matters **- the best agents are built by teams who understand their target industry deeply. Technical excellence isn't enough.**Pricing drives adoption** - agents priced like tools get treated like tools. Agents priced like outcomes get renewed like essential services.> 💡 The agent economy is real. These companies are generating serious revenue by automating work that used to require humans. --- ### [2025-07-16] When AI pricing became the hottest topic in London - URL: https://paid.ai/blog/company/when-ai-pricing-became-the-hottest-topic-in-london - Date: 2025-07-16 - Author: Arnon Shimoni - Summary: When founders get together to solve real problems, magic happens. Last Tuesday night we tried something different.We invited AI founders, investors, and operators to squeeze into a room at Lightspeed as part of [Generative London](https://events.lsvp.com/lightspeed/rsvp/register?e=generative-london-paid-ai) to talk about the one thing nobody wants to discuss but everyone's stressed about: ***pricing***.We expected a few engaged people. We got 75+.People were literally standing against the walls. We almost had to turn folks away. Apparently billing can be exciting when it's this broken. ## The room was split pretty nicely **40% founders** wrestling with impossible pricing decisions**60% investors and operators** trying to value these new business modelsMost everyone had the same question: how do you price something that's never existed before? ## What we learned from a room full of stressed founders ## Lesson 1: Pricing is the new product-market fit Your AI's capabilities don't matter if you can't prove the value you delivered. This is where most companies are failing. We watched founders nod along as this truth hit home. ## Lesson 2: The "cheaper AI = better margins" myth is dangerous The room erupted when this came up. Founders are learning the hard way that induced demand and model upgrade pressure actually compress margins as inference costs drop.It's like Netflix getting cheaper - people didn't watch the same shows and save money, they binged everything. Same with AI models. Everyone wants the latest greatest, which costs way more. ## Lesson 3: Customer trust erodes fast with bad pricing When customers pay for zero value delivered, you're not just losing money. You're losing trust. And trust is harder to rebuild than code. ## Lesson 4: Traditional SaaS billing is officially dead for AI "If you're charging per seat, you're a dead man walking" - Manny's words that got quoted in every LinkedIn post after.The shift from predictable revenue to variable outcome-based pricing is the biggest business model change since SaaS killed perpetual licenses. ## Lesson 5: Four pricing models are actually working Per-agent, per-action, per-workflow, and per-outcome. Everything else is just experimentation that's not scaling. ## Lesson 6: Outcome-based pricing isn't a fad It's survival. Agentic startups aren't just selling software. They're automating labor. Which means they're not pulling from IT budgets. They're tapping into labor budgets.As one founder put it: "Outcome-based pricing does what good economics always aims for - aligned incentives." ## The live demo that almost broke everything Our CTO Raj vibe-coded our embeddable UI blocks on stage. In front of 75 engineers who can spot bugs faster than you can type them.Building a dashboard from scratch while everyone watches? Let's call it "an experience."But it worked. People saw how you can drop billing infrastructure into any app in minutes, not months. ## Why this event mattered to us Three people stayed past 9pm to get pricing advice. War stories were shared. Actual solutions were discussed.If you're building an AI agent and pricing feels like an impossible puzzle, you're definitely not alone. The entire industry is figuring this out together.Traditional billing wasn't built for this new world. That's exactly why we exist.Thanks to everyone who came, especially those who helped answer questions and brought great energy. Special shoutout to [Johnny Wing from the kitchen industry who said "I'm here to be the dumbest guy in the room"](https://www.linkedin.com/feed/update/urn:li:activity:7349079055171760128) - that's exactly the right attitude!We're planning more events like this, because when founders get together to solve real problems, magic happens. --- ### [2025-07-10] How to make your AI agent irreplaceable - URL: https://paid.ai/blog/ai-agents/how-to-make-your-ai-agent-irreplaceable - Date: 2025-07-10 - Author: Arnon Shimoni - Summary: 5 Strategies to build agentic AI that sticks. When Andreesen Horowitz released their findings from surveying 100 Enterprise CIOs about their AI purchasing behavior, the headline sounded like great news for anyone selling AI:*“AI spend blew past high expectations and is here to stay,”* boasted [the article by a16z](https://a16z.com/ai-enterprise-2025/).Cue the confetti. But read a little deeper, and the mood shifts.Enterprise companies aren’t shopping for the *best* solution and then making a purchase. They’re trying a handful of options, then ripping out the ones that don’t stick. This is [Vibe Revenue 101](/blog/ai-monetization/vibe-revenue-a-mirage-of-ai-success).Vibe revenue means it’s easy to get a tryout, but it’s hard to make the final cut.A recent study by the [S&P Global says “Generative AI experiences rapid adoption, but with mixed outcomes”](https://www.spglobal.com/market-intelligence/en/news-insights/research/ai-experiences-rapid-adoption-but-with-mixed-outcomes-highlights-from-vote-ai-machine-learning).At [Paid](/), we’ve consulted with 200+ agentic companies to see what sticks - and we’re building our own product to be impossible to rip out.If you’re building agentic AI, this is your playbook for making sure you don’t get replaced next quarter. ## Easy come. Easy go. **Q:** What kind of AI agent has the highest churn rate?**A:** The one you can rip out without consequence.Lots of point solutions fall into this category - your AI notetaker, research assistant, and productivity agents. Chances are if “plug and play” describes your agent, it’s a dime a dozen, fully commoditized in the market.But of course, you didn’t build an agent just to get replaced in 3 months. If your product is easy to install, it’s easy to delete - like ripping off a band-aid.However, if it’s deeply embedded, ripping it out shuts down the whole system - like ripping out a brainstem.Our research into the state of AI agents shows there’s a spectrum from commoditized to irreplaceable. Three stages from bandaid to brainstem. ## Level 1: Surface integration - Simple tasks: Records calls. Uploads to CRM. Answers FAQs.- Doesn’t change how the business operates. Just layers on top.- Easy to adopt (but easy to forget)> **Reality check:**Most competitors can vibe-code this in 2–3 months. If your agent lives here, your moat is as thin as a UI preference. ## Level 2: Workflow integration - Powers critical flows like pricing pages, CPQ, onboarding, and checkout.- Replacing it would break processes and require real engineering time.- Customers won’t switch unless they have to.> **Reality check:**Ripping this out costs 2–6 months of dev time. Enough pain to think twice, but not a dealbreaker. ## Level 3: *Deep* workflow integration - like a brainstem - It’s not just part of the system. It *is* the system.- Integrated across departments, workflows, and lots of touchpoints.- Starts defining how work gets done.> **Reality check:**Replacing this means 6–18 months of re-engineering.This is what successful enterprise AI deployments are built on.The words “change management” end any conversation about switching to a competitor. ## What does Level 3 Integration look like in practice? The most successful agentic AI companies we’re working with are building agents with deep workflow integrations that run part of their customer’s business.Surface integrations get swapped every budget cycle.Deep workflows survive org changes, reorgs, and procurement reviews.We built Paid to operate at that level because we knew from day one: agents that don’t integrate, don’t survive. ## How to make your agent stick Here’s how to design for deep integration from day one. ## 1. Redesign the workflow, not just the task **The goal: **turn a helpful tool into operational glue.If your agent automates one step, it’s easy to swap.Map the full workflow and look for upstream and downstream touchpoints you can own. ## 2. Go deeper than the UI **At Paid:** We became part of finance operations, not just a reporting layer.CRM syncs and Slack alerts aren’t integration.Design your agent to impact how decisions are made, how data is processed, or how money moves. ## 3. Integrate Like an ERP ERP systems are notoriously hard to rip out because they’re so embedded. They’ve taken down entire companies.Build integration points that touch multiple departments and business rules, so removing you means re-architecting workflows.Think “operational spine,” not “bolt-on feature.” ## 4. Become the Source of Truth The most irreplaceable agents generate data no one else does, then make that data essential.Find the metric your customer’s leadership can’t live without, and make your agent the only source.**At Paid:** We own *agentic ROI* — the hard numbers that prove value beyond vibes. We track real-time margins, usage analytics, and cross-functional impact that powers retention and expansion. ## 5. Expand Inside the Org Single-team adoption is fragile. Cross-org reliance is sticky.Build dashboards, reports, or triggers that serve sales, finance, CS, and ops — not just your champion.As more companies deploy multi-agent workflows, your agent’s ability to integrate, collaborate, and sync across use cases becomes a retention moat. ## Make It Stick Or Get Swapped In the agentic economy, tryouts are easy to get. Contracts are hard to keep.Your customer isn’t loyal to your interface, your model, or your brand.They’re loyal to the work they’d have to redo if you vanished.**That’s what you’re competing with.** --- ### [2025-07-07] Metering for AI: Why Everyone's Solving the Wrong Problem - URL: https://paid.ai/blog/billing/metering-for-ai-why-everyone-s-solving-the-wrong-problem - Date: 2025-07-07 - Author: Arnon Shimoni - Summary: "How do we handle metering for our AI agents? The token tracking is getting complex, and our costs are all over the place." I've had this conversation at least 10 times in the past few months:"How do we handle metering for our AI agents? The token tracking is getting complex, and our costs are all over the place."My answer quite often surprises people: **Metering isn't your problem.**Not because it's technically impossible, but because you’ve not understood what you should be measuring. ## The token obsession I’ve seen this so many times:- They use token-heavy frameworks- They spend months building sophisticated token tracking systems- They create complex cost allocation algorithms- Sometimes, they obsess over GPU utilization metrics- They end up having to build real-time consumption dashboards to alert their customers to usageThen then they wonder why customers hate their pricing and churn after pilots.**Yes, the technical solution was good, but the problem was wrong.** ## AI metering is much simpler for agents (not harder) I’ve built a few billing systems, so let me tell you. AI agent metering is fundamentally easier than traditional SaaS or even other AI systems because:**Lower volume, cleaner signals:**- Agents have discrete, countable outcomes (not continuous behavior)- Agents operate at roughly human-scale frequencies (hundreds of actions, not millions of events)- Agents tend to have clear success/failure states- Agents, in some verticals, have obvious attribution (this agent did this task) - for example in customer supportCompare this to traditional AI metering challenges:- High-volume event streams requiring real-time processing- Complex user session tracking across devices- Ambiguous value attribution across features- Edge cases around partial usage ## The problem being solved: Input vs. Output confusion Everyone gets stuck on metering because they’re coming from a mindset that’s copying what they’ve seen with OpenAI or other foundational models.Sometimes, it’s where consumption correlates with value.But AI agents are very different.**In SaaS:** More usage generally = more value- More seats = more users getting value- More storage = more data preserved- More API calls = more functionality accessed**In AI:** Consumption has zero correlation with value- Email A: 5,000 tokens → Perfect sales email- Email B: 25,000 tokens → Same quality email (with much longer context window)Should Email B cost 5x more? Absolutely not. ## The Technical Architecture You Actually Need If metering isn't the hard problem, what is?**Outcome attribution**This is where the real challenge lies, in my opinion.The hard problems are:- Defining clear success criteria- Handling partial successes (do you still charge for part of a workflow?)- Attribution across multi-step workflows- Measuring quality and building guardrailsToken counting? That's the easy part.But that’s why so many companies end up counting tokens. It’s easier. ## But what about costs? LLM tokens match up to costs! You absolutely should track consumption - for internal optimization, not customer billing.Paid lets you do both in one go. You can track costs and charge your customer at the same time.The price of the outcome will be based on a separate pricing you agree with your customers, not directly tied to a token cost. ``` paid = Paid::Client.new( [...] ) paid.usage.record_usage( signal: { agent_id: "agent_1", event_name: "document_processed", customer_id: customer_id data: { 'llm_tokens': 15000, 'llm_cost': 0.045, 'infrastructure_cost': 0.023, 'processing_time': 45.2, } } ) ``` Think of it like a restaurant: You pay $12 for a pizza, while the restaurant tracks labor costs and rent obsessively. But they don't charge you $2.50 for flour + $0.30 for tomatoes + $4.20 for chef time. ## The Bottom Line You shouldn’t build your own elaborate metering systems for AI agents.Agent behaviour makes it easier than traditional SaaS, but the business model makes it irrelevant to meter.The companies we see succeed the most spend time on the success criteria, not on real-time token consumption tracking - and definitely not on GPU inference time.Your customers don't care about your infrastructure costs. They care about the business problems you solve.Meter for margins. Bill for outcomes! --- ### [2025-07-03] How to do AI Agent billing in 2025 - URL: https://paid.ai/blog/billing/how-to-do-ai-agent-billing-in-2025 - Date: 2025-07-03 - Author: Arnon Shimoni - Summary: The Agentic Playbook If you feel like you can’t find the right answers to agent pricing, it’s because they’re being written in real time. We’re in the midst of the biggest shift to disrupt the market in 25 years, and every founder we speak with has the same questions. Luckily, we’re obsessed enough with billing to compile all the answers.Agentic AI is software, and for the past 25 years, software has been SaaS. So, it has been (understandably) difficult for the world to accept that*** Agentic AI is not SaaS.***Agentic AI is to SaaS what SaaS was to CD-ROM. The rapid change you’re seeing in approach to pricing is indicative of the market waking up to this distinction.AI Agents priced as SaaS makes no more sense than SaaS being priced as CD-ROM.*Outcome based pricing billed on AI native platforms* is the final destination everyone will eventually get to. We’ve said it since the beginning, and we’re watching the debates finally start to tip in our direction. Since we spent the last 6 months interviewing over 160 agentic founders, we were a little early to the party.The good news is we’ve had plenty of time to sort out the details for you. ## What pricing models actually work for AI Agents? Though SaaS was born in 2000, Windows was still sold on a shelf in a box until 2012. Not because Benioff was wrong, but because he was early. Physical products in retail stores was the monetization model the software industry knew and had access to.Today’s Office Max equivalent, “the devil you know” so to speak, is a category of billing platforms that predates the first agentic company. Billing agentic AI on these platforms is like selling Salesforce at Office Max today.> Stripe, Chargebee, and Zuora weren't built for the AI economy. They’re built on traditional SaaS assumptions that break with agents.**Seat-based pricing:** Agents don't need "seats" because one user can deploy dozens of agents**Fixed SKU structures:** Can't handle dynamic, context-dependent agent workflows**Usage = API calls:** Misses the actual work your agents perform**Monthly billing cycles: **Agents deliver value continuously, not in subscription intervals75% of AI agent companies we analyzed struggle with pricing because they're forcing agent economics into SaaS billing boxes. They’re the devil you know, so you try to make it work. (Spoiler: it doesn’t work!)***Instead of pricing like SaaS, price like labor. That’s what agents are doing.*****Per workflow completed **(document processed, call handled, analysis delivered)**Per outcome achieved** (meeting booked, issue resolved, lead qualified)**Per unit of business value** (cost saved, revenue generated, time recovered)As Rob Litterst from PricingSaaS puts it: "AI agents are making it so that you actually can do jobs... SaaS could never actually do a job for you." The shift from selling tools to selling outcomes is already happening, with companies like Intercom and Salesforce leading the charge.FrameworkBest forRevenue modelRisk/rewardImplementation complexityCompanies using this modelAgent-based- FTE replacement- Enterprise deployments- Predictable- Recurring revenueLow/standardSimple- 11x- Harvey- Salesforce (hybrid)- Hubspot (hybrid)Action-based- Variable workloads- BPO replacement- Pay-per-use- Usage-based pricing- Consumption-based pricingLow/standardSimple- Bland- Parloa- HappyRobotWorkflow-based- Process automation- Multi-step tasksUsage based with minimumsMedium/mid-highModerate- Rox- Salesforce- Artisan- IcertisOutcome-based- Measurable results- Performance guaranteesSuccess basedHigh/highComplex- Zendesk- Intercom- Airhelp- Chargeflow- SierraWe broke down each of these models in detail in [The Complete Guide to AI Agent Monetization](/blog/ai-monetization/the-complete-guide-to-ai-agent-monetization). We’ve seen every one of these models implemented and have tracked what works and what doesn’t across all 160+ companies we’ve consulted.Every founder starts with usage-based pricing. Half realize it’s killing their margin within the first 30 days. ## How do you make AI Agents profitable? The biggest killer of agentic companies is **margin erosion**, *not* lack of revenue.ARR has always been the lifeblood of SaaS. Since cost of delivery held steady at scale, ARR and EBITDA had a strong positive direct correlation.That correlation flies out the window with Agentic AI. Costs can spike unpredictably and margins can vanish overnight if your customers write too much to your agent.**Here's what breaks traditional margin tracking:**- **Model costs fluctuate**: GPT-4o costs 33% more than GPT-4.1, but customers pay the same.- **Usage varies wildly**: One customer's "simple" task burns 10x the tokens you expected.- **Hidden infrastructure costs**: Voice processing, image analysis, vector databases - all invisible to traditional billing.Stripe and friends help you collect money. They don't help you spot a **bleeding margin** when your inference costs explode.We’ve seen founders celebrate pilot wins… only to panic when infra bills hit 5 figures with no matching revenue. Adam Schoenfeld candidly shared with us: *"I do think it's going to be something we have to deal with because... there is a version of the future where the costs actually go up, where like the agents have a lot more different tools and those tools have their own costs and there's all this variability between customers."*This margin compression is happening everywhere, and most don't catch it until it's too late.**The metrics that actually matter for agent profitability:**- **Agentic Margin (AM)**: Revenue minus *all* agent operating costs per customer- **Agentic Margin Ratio (AMR)**: Your true profit percentage after AI infrastructure- **Task Monetization Ratio (TMR)**: What percentage of agent work actually generates revenue**Real example:** One customer told us their "profitable" AI support agent was actually *losing* $0.40 per conversation after accounting for model costs, voice processing, and infrastructure. They were scaling their way to bankruptcy.It’s gutting to realize your most successful agent is actually hemorrhaging money. That’s why we wrote a complete guide to margin management: [The Agentic Margin: What It Costs vs. What You Earn from AI Assistants](/blog/ai-monetization/the-agentic-margin-what-it-costs-vs-what-you-earn). We want to help you catch the bleed before you scale it. ## How do you grow and scale an agentic company? Does it feel like all your customers are blown away by what your agent can do, but then start to go cold after a few months? Most of the founders we spoke with saw a major drop-off 6 to 9 months after launch. It can be super scary, but there is a path forward.Vibe Revenue is a new challenge that software never faced in Office Max or SaaS, but is giving agentic companies a damning false hope.AI Agents are so hot right now everyone wants to try them… once. Then, never give you money ever again. The result is an initial influx of money followed by a steep drop off 6 to 9 months after launch. That’s ***Vibe Revenue.***As Sequoia's Pat Grady perfectly captures this phenomenon: "The difference between having a magic moment and solving an end-to-end workflow... That's the difference between the vibe revenue and the real revenue."Initial contracts come up for renewal. Novelty wears off. Customers evaluate ROI with cold, hard metrics. Without genuine momentum, you're facing the renewal cliff without a paraglider. We’ve watched it happen like clockwork to companies of all sizes.**The warning signs you're in vibe territory:**- High initial adoption but shallow engagement depth- Customers trying your product vs. *depending* on it- Usage metrics that look great but don't translate to business outcomes- Contracts that feel more like "experiments" than strategic investments**What separates vibe from value?**- **Deep workflow integration** (not just cool features): Each integration point creates another reason to renew- **Expanding value over time**: Agents handling increasingly complex, important tasks- **Measurable business impact**: Clear ROI that executives can defend to the board**Real example:** One customer told us their AI agent went from "nice-to-have demo" to "mission-critical infrastructure" when they shifted from showcasing features to demonstrating $50K/month in cost savings with detailed attribution.We wrote extensive instructions on how to prove outcome and sell results in [Vibe Revenue: A Mirage of AI Success.](/blog/ai-monetization/vibe-revenue-a-mirage-of-ai-success) ## The AI economy doesn’t fit on SaaS infrastructure If you’re building agents that work 24/7 to deliver measurable business outcomes and replace entire job functions, you’re not building SaaS. You can’t bill for SaaS.AI agent companies using outcome-based pricing see 4-8x higher contract values than those trapped in SaaS models. They convert pilots faster, retain customers longer, and scale without the margin compression that kills their competitors. --- ### [2025-06-24] AI Agent Monetization FAQ: The Basics - URL: https://paid.ai/blog/ai-monetization/ai-agent-monetization-faq-the-basics - Date: 2025-06-24 - Author: Arnon Shimoni - Summary: Basic concepts you should know about agentic monetization and AI monetization, including how to price agents and how to measure value. ## AI monetization: Basic Concepts ## The Four Pricing Models that work for agents ## Pricing Strategy ## FAQ --- ### [2025-06-22] How do you handle pricing predictability for Agentic AI? - URL: https://paid.ai/blog/ai-monetization/how-do-you-handle-pricing-predictability-for-agentic-ai - Date: 2025-06-22 - Author: Arnon Shimoni - Summary: Give your customers predictable pricing without compromising on profitability with our guide At Paid, we’ve spoken with more agentic AI founders than anyone else. They all think they’re the only ones secretly guessing at pricing.Even the ones who don’t think they’re mimicking SaaS playbooks are often doing so by accident. And when we show them how their pricing kills [Agentic Margin](/blog/ai-monetization/the-agentic-margin-what-it-costs-vs-what-you-earn), the same explanation always comes up:> "My customers want predictable pricing. It needs to be the same every month."That feels reasonable… until you ask:> “*Why* do customers want predictable pricing in the first place?” ## SaaS Trained Buyers to Expect Predictability In the SaaS era, pricing predictability made sense. Infrastructure costs were relatively fixed. Whether users logged in once a week or 100 times a day, the bill stayed steady. Predictability was easy to offer, and customers came to expect it.Agentic AI doesn’t work like that. Every action your agent takes, every prompt, API call, speech synthesis, avatar render incurs variable cost. And those costs don’t scale linearly with usage. They spike with task complexity and change as user behavior evolves.Unlike SaaS, usage can’t always be tied to a single user because in some cases, the agent is *replacing* the user.Trying to force predictability in this model is like trying to sell Uber rides on a Netflix subscription. Netflix is a fixed fee because it’s a fixed cost. Charge someone $12/month for Uber and they’ll cost you $800 in driver payouts. ## It’s Hard to Kill the Status Quo Whether or not we want to admit it, founders are all too happy to acquiesce simply because it’s easier. Measuring and tracking the costs associated with agentic work is difficult because the systems for tracking them haven’t caught up to demand - as Steve Jobs also found out when they were manufacturing the Apple II.The tools needed to track and price agentic labor correctly are scarce. While your typical SaaS app consumes resources rather predictably (like monthly subscriptions), AI agents are more like having a consultant who might spend 10 minutes or 3 hours on the same task, depending on how the conversation unfolds.And the length of the conversation matters too. The costs are all over the place: one simple customer support interaction might cost $0.02 with a cheap model, but if the agent gets confused and starts looping through different models, calling external APIs, and building up massive context windows, that same interaction could suddenly cost $0.50 (that’s 25x more, and that’s not a mistake).If you consider how every AI provider prices differently (OpenAI's tokens aren't the same as Anthropic's, Google’s way of charging is even more confusing somehow), and you've got a real problem where you just need to guess because building proper usage tracking feels impossible.Yes, there are tools that are emerging to track it (like LangSmith for deep workflow analysis or AICosts.ai for multi-provider visibility) but most companies are still flying blind.So, agentic founders are falling back on what their SaaS predecessors did before them and pricing “the way we’ve always done it.” ## Why Subscription Pricing Kills Agentic Companies > Predictable losses are worse than unpredictable wins. No CFO is turning down 300% ROI because the invoice varied by 15%.On the face of the business, subscription pricing devalues agentic work because it minimizes the results the agent produces.On the backend, if your costs fluctuate but your revenue doesn’t, margin becomes a mystery.It’s a lose-lose.You burn your AMR (Agentic Margin Ratio) while positioning yourself as a commodity.> Agentic Margin ÷ Revenue Generated by Agent = Agentic Margin RatioYou end up subsidizing power users and overcharging light ones.You lose the ability to optimize because your pricing isn’t connected to what your agents actually do.In trying to give customers peace of mind, you’ve sacrificed your own clarity and control. Your differentiation from your competitors evaporates and you have no moat. ## Predictable ≠ Flat Your customers don’t need the same invoice every month, but they do need protection from “fear of spikes.”Candid conversations with agentic builders have helped us uncover viable solutions for pricing predictability - that don’t break the Agentic Margin.**1. Flat Fee + Outcome component**Some buyers want simplicity. Others want performance alignment. Give them both.Offer two clear options:- **Higher flat fee** (for peace of mind)- **Lower flat fee + success fee** (for ROI alignment)**2. Capped Outcome pricing**The biggest fear with variable pricing is runaway costs. Fix it with a cap (aka ceiling)!Set your outcome based pricing with a **monthly cap (e.g. 150% of flat rate equivalent)**. This gives you upside while giving customers a ceiling they can plan against.Bonus: Use **past customer data to simulate 12 months** of invoices under each model. Share that with your customers to prove the value you generate.**3. Hybrid: Upfront Now, Success Later**For seasonal businesses, like education or hospitality, timing matters. Don’t force revenue when outcomes won’t happen.Try this combo:- **Implementation Fee paid upfront** (cash now, revenue in slow season)- **Success Fee paid later** (value-based, more revenue when work ramps)Or, build in **credits**: Let them “earn” outcomes in their slow season and “spend” them in their busy season. ## Objections You’ll Hear (and How to Solve Them) > Q: "What if our usage is seasonal?"A: You put in some safeguards. Allow pauses, place a cap. Predictability doesn’t mean *monthly uniformity.*> Q: "Procurement won’t approve variable billing."A: Frame variable elements as performance-based. Bundle them with a fixed platform fee. This isn’t too different from paying your sales people a commission.> Q: "We don’t have enough data to model this yet."A: Yes so start now! Track usage at the task, workflow, and outcome level. The sooner you collect signals, the sooner you can price confidently.> Q: "What if usage grows way beyond what we modeled?"A: Great. That means value is growing. This is an excellent signal, not something to be afraid of. This means you can offer a discount or renegotiate the contract. ## Where Paid Fits In Wherever you are on the pricing journey, fixed, hybrid, or outcome-based, Paid helps you:- Track real-time signals from agent usage- Measure cost and Agentic Margin at the task and outcome level- Design billing models that reflect real delivered valueWe don’t force you into a pricing philosophy. We help you operationalize the one that matches how your agents work.Predictability is a feature. **Alignment is a strategy!**If you’re selling value, your pricing should prove it and Paid helps you get there. --- ### [2025-05-31] The complete guide to AI agent monetization - URL: https://paid.ai/blog/ai-monetization/the-complete-guide-to-ai-agent-monetization - Date: 2025-05-31 - Author: Arnon Shimoni - Summary: Transform your AI agents from cost centers to revenue generators with Paid's proven monetization frameworks. I don’t need to tell you again, but the AI agent economy is exploding.This year, tens of thousands of businesses have deployed millions of AI agents across every industry.Most companies (over 75%!) struggle with a critical question: **How do we turn these AI tools into *****sustainable***** revenue streams?**At [Paid](/), we've pioneered the infrastructure that powers monetization for the world's most successful AI agents.This guide shares our proven frameworks to help you design, implement, and scale your agent monetization strategy.So what should you consider when monetizing an AI agent? ## The four AI Agent pricing models dominating the market We've identified four fundamental monetization frameworks proven by 60+ AI agent companies, each suited to different agent types and business models:1. 👤 **Agent-Based Framework**: The FTE Replacement Model2. ⚡ **Action-Based Framework**: The Consumption Model (also called Usage based)3. ⚙️ **Workflow-Based Framework**: The Process Automation Model4. 🎯 **Outcome-Based Framework**: The Results-Based ModelThese frameworks are easy to compare because they are sufficiently different. ## Comparison at a Glance - Usage based vs. Outcome based FrameworkBest forRevenue modelRisk/rewardImplementation complexityCompanies using this modelAgent-based- FTE replacement- Enterprise deployments- Predictable- Recurring revenueLow/standardSimple- 11x- Harvey- Salesforce (hybrid)- Hubspot (hybrid)Action-based- Variable workloads- BPO replacement- Pay-per-use- Usage-based pricing- Consumption-based pricingLow/standardSimple- Bland- Parloa- HappyRobotWorkflow-based- Process automation- Multi-step tasksUsage based with minimumsMedium/mid-highModerate- Rox- Salesforce- Artisan- IcertisOutcome-based- Measurable results- Performance guaranteesSuccess basedHigh/highComplex- Zendesk- Intercom- Airhelp- Chargeflow- Sierra ## Decision framework: choose your model Use this decision framework to select the optimal pricing model for your AI agent.Should you be looking at usage based, agent based, or outcome based pricing? ## Key questions to ask: - **What budget am I targeting?**Headcount budget (10x larger) → *Agent-based*- BPO/outsourcing budget → *Action-based*- ROI/performance budget → *Outcome-based*- **How commoditized is my offering?**Unique capabilities → *Any model works*- Standard features → *Avoid the action-based model*- **Can I prove my value?**Clear attribution → *Outcome-based*- Process efficiency → *Workflow-based*- Time savings → *Agent-based*Let’s look a bit deeper into each of the models, their pros and cons, and how they work: ## Agent-based pricing Agent-based pricing treats AI agents like digital employees where customers pay a fixed recurring fee for each agent deployed.This model positions agents as FTE replacements, tapping into headcount budgets rather than IT budgets.Companies like **11x**, **Harvey**, and **Vivun** have proven this approach can command premium pricing by directly competing with human hiring costs. ## How It Works ## Design Principles - **Position as Digital Employee**: Frame your agent as a FTE replacement- **Tap Headcount Budgets**: Target 10x larger budgets than traditional software- **Demonstrate Clear ROI**: Show savings vs. $60,000/year junior employee- **Bundle Capabilities**: Add features to justify premium pricing and resist commoditization ## Implementation Guide ## Step 1: Define Your Agent Tiers - **Starter**: Basic functionality, single user- **Professional**: Full features, team access- **Enterprise**: Custom limits, SLA guarantees ## Step 2: Set Pricing Anchors - Research competitor SaaS pricing in your domain- Price 20-30% below equivalent human cost- Build in margin for infrastructure and support ## Step 3: Configure in Paid ## Real-World Examples from our customers ## Example 1: Legal Document Review agent **Agent**: AI Legal Assistant for contracts- **Functionality**: Reviews contracts, identifies risks, suggests edits- **Pricing Model**:Starter: $3,000/month (up to 50 contracts)- Professional: $8,000/month (up to 200 contracts)- Enterprise: $20,000/month (unlimited contracts, API access)- **Why it works**: Law firms have predictable document volumes and prefer fixed costs for budgeting ## Example 2: Customer Success agent **Agent**: Churn protection / retention AI- **Functionality**: Monitors customer health, predicts churn, automates interventions- **Pricing Model**:Small Business: $1,500/month (up to 500 customers)- Growth: $5,000/month (up to 2,500 customers)- Scale: $15,000/month (up to 10,000 customers)- **Why it works**: SaaS companies want predictable costs tied to their customer base size ## When to choose agent-based pricing - ✅ Your agent performs a comprehensive set of tasks replacing a job function- ✅ Customers want budget predictability from headcount allocations- ✅ You can demonstrate clear FTE replacement value- ✅ Competition prices on seat-based models- ✅ You want to avoid race-to-bottom pricing pressure⚠️ **Future-proofing tip**: As LLM costs drop 10-100x, shift your value prop from "cheaper than human" to "vastly more capable than human" ## Action-based pricing Action-based pricing charges customers for every discrete action their agents perform.Used by agentic companies like **Bland** and **Parloa**, this consumption model mirrors cloud infrastructure and BPO pricing.It's transparent but vulnerable to commoditization as AI costs decline. ## How It Works ## Pricing principles - **Transparent Consumption**: Direct correlation between usage and cost- **Low Barrier to Entry**: Customers only pay for what they use- **BPO Competition**: Target the $900/employee BPO market- **Volume Discounts**: Incentivize higher usage with tiered pricing ## Implementation guide ## Step 1: Calculate unit economics - Determine base LLM/infrastructure costs- Add 50-300% margin based on value delivered- Build in cushion for future cost reductions ## Step 2: Define billable actions ## Real-World Examples ## Example 1: AI Voice Agent (Bland.ai Model) **Agent**: Call AI- **Functionality**: Handles inbound customer service calls- **Pricing Model**:$0.12/minute for inbound calls- $0.18/minute for outbound calls- Volume discounts: 10% off at 10,000 minutes/month- **Why it works**: Direct competition with call centers at 70% lower cost ## Example 2: Document Processing Agent **Agent**: Document parsing AI- **Functionality**: Extracts data from invoices, receipts, contracts- **Pricing Model**:$0.10 per page processed- $0.02 per data field extracted- Bulk pricing: $500 for 10,000 pages/month- **Why it works**: Clear unit economics vs. manual data entry costs ## When to Choose Action-Based - ✅ Competing directly with BPOs or call centers- ✅ Highly variable usage patterns- ✅ Customers want "pay only for what you use"- ✅ Simple, discrete, measurable actions- ✅ Testing market fit with low commitment⚠️ **Warning**: This model faces the highest pricing pressure as AI costs plummet. Plan to transition to workflow or outcome-based pricing within 12-18 months. ## Workflow-based pricing Workflow-based pricing charges for complete sequences of agent actions that deliver specific intermediate outcomes.Companies like **Rox**, **Salesforce**, and **Artisan** use this model to balance between pure consumption and outcome pricing. Each workflow represents a meaningful business process with clear deliverables. ## How It Works ## Design Principles - **Process-Level Value**: Price complete workflows, not individual actions- **Complexity Protection**: Complex workflows resist commoditization- **Clear Deliverables**: Each workflow produces measurable intermediate outcomes- **Margin Management**: Monitor workflow costs to avoid negative margins on complex processes ## Implementation Guide ## Step 1: Map your workflows You want to identify the actions and assign some values to the steps.- Identify all discrete actions your agent performs- Estimate resource consumption per workflow- Assign business value to each workflow ## Step 2: Design pricing structure Here, you set prices to cover your margins and costs.- Set base platform fee (covers overhead)- Price each workflow based on:Computational cost- Business value delivered- Market alternatives ## Step 3: Add commitment tiers Commitment tiers and included quantities are easy ways to force a minimum revenue, but they can backfire if you don’t explain to your customer what they get.Here’s how you set them in Paid: ## Real-World Examples ## Example 1: Sales Development Representative (SDR) Agent **Agent**: Automated SDRHybrid setup:- **Base platform fee**: $3,000/month (platform access)- **Workflow pricing**:Lead Research: $2 per lead profiled- Email Personalization: $1 per email crafted- LinkedIn Outreach: $3 per connection request- Meeting Booking: $8 per meeting scheduled- **Optional commitment packages**:Starter: 500 leads/month minimum ($5,000 guaranteed revenue)- Growth: 2,000 leads/month minimum ($2,500 guaranteed revenue)- Scale: 10,000 leads/month minimum ($5,000 guaranteed revenue)- **Why it works**: Sales teams can start small and scale with success ## Example 2: Financial Analysis Agent **Agent**: Automated CFO AssistantHybrid setup:- **Base platform fee**: $5,000/month- **Workflow Pricing**:Report creation: 20 included, $100 for every overage- Cash Flow Forecast: $250 per forecast- Budget vs Actual Report: $50 per department- Board Deck Generation: $500 per deck- Real-time Dashboard Update: $25 per refresh- **Optional volume discounts**: 20% off after 50 workflows/month- **Why it works**: Finance teams have varying needs throughout the month/quarter ## When to Choose Workflow-Based - ✅ Your agent executes multi-step processes with clear deliverables- ✅ Workflows are standardized but complex enough to avoid commoditization- ✅ You can demonstrate ROI for each workflow type- ✅ Different workflows deliver different business value- ✅ You want pricing flexibility between action and outcome models⚠️ **Watch Out For**:- Complex workflows (document parsing, security scans) risk negative margins. Make sure you segment your workflows for length or complexity to avoid having your margins crushed.Monitor costs carefully to maintain profitability! ## Outcome-based pricing Outcome-based pricing represents the pinnacle of value-aligned pricing and the model most resistant to commoditization.Companies like **Zendesk**, **Intercom**, **Airhelp**, and **Chargeflow** charge only for successful business results. As we note in [our frameworks article with Kyle Poyar,](https://www.growthunhinged.com/p/ai-agent-pricing-framework), this model will likely dominate as AI costs plummet.**It's the only model that completely decouples pricing from underlying technology costs.** ## How It Works This model is similar to the workflow-based pricing model. Even though you can combine it with a platform fee and others, you should avoid pricing individual “attempts”.Has there been success? Bill.No success? Don’t charge. ## Design Principles - **Results-only focus**: Charge only for achieved outcomes, not attempts.- **Clear attribution**: Develop robust methodologies to prove your impact- **Shared risk AND reward**: Include performance guarantees or success bonuses, like revenue share.- **Premium positioning**: Command highest prices through guaranteed results- **Future-proof**: This model is the most resistant to AI commoditization ## Implementation guide This model is the trickiest to get right, so follow carefully! ## Step 1: Define success metrics - These must be objectively measurable- Directly tied to business value for best success ## Step 2: Calculate Risk-Adjusted Pricing - Estimate success rate of your agents- Add risk premium (typically 30-50% with most of our customers)- Include a base platform fee to cover operational costs if no success is reachedSetup in Paid is similar to the workflow setup, but prioritizes outcomes and not just actions. ## Real-World Examples ## Example 1: Recruiting Agent **Agent**: Automated Recruiter- **Base Fee**: $2,000/month (platform access, unlimited searches)- **Outcome Pricing**:Qualified Candidate Submitted: $500- Interview Scheduled: $1,000- Offer Accepted: $5,000 or 15% of first-year salary- **Success Metrics**:Interview: Candidate completes first round with hiring manager- Offer Accepted: Candidate signs offer letter- **Why it works**: Recruiting is already outcome-based; AI agent follows industry model ## Example 2: E-commerce Optimization Agent **Agent**: Conversion Rate Optimizer- **Base Fee**: $500/month (A/B testing infrastructure)- **Outcome Pricing**:Conversion Rate Improvement: $2,000 per percentage point- Revenue Increase: 5% of incremental revenue- Cart Abandonment Reduction: $50 per recovered cart- **Success Metrics**:Conversion: Measured via integrated analytics, 30-day attribution- Revenue: Year-over-year comparison, seasonally adjusted- Cart Recovery: Completed purchase within 7 days of intervention- **Caps**: Maximum $50,000/month to prevent runaway costs- **Why it works**: Direct tie to revenue makes ROI crystal clear ## When should you choose Outcome-based pricing? - ✅ Success can be clearly measured and attributed- ✅ Your agent consistently delivers quantifiable results- ✅ Market already expects outcome-based pricing (e.g., recruiting, sales)- ✅ You want maximum pricing power and differentiation- ✅ You're confident in your agent's performance💡 **Tip**: This is the most future-proof model. As AI costs approach zero, outcome-based pricing maintains margins by focusing on value delivered, not resources consumed. ## Paid’s best practices for agent monetization ## 1. Start simple, evolve sophisticated - Launch with agent-based pricing- Add usage components as you learn- Introduce outcomes once proven ## 2. Transparency builds trust - Real-time usage dashboards- Clear billing breakdowns and value receipts- Proactive cost alerts ## 3. Price for value (not cost) - Research alternative solutions- Understand customer budgets- Price below human equivalent- Leave room for discounting ## 4. Monitor and iterate - experiment frequently - Track key metrics:Customer acquisition cost- Lifetime value- Churn rate by pricing model- Usage patterns- A/B test pricing changes- Survey customers quarterly - not just those who churn --- ### [2025-05-18] Our playbook for launching Paid.ai - URL: https://paid.ai/blog/company/our-playbook-for-launching-paid-ai - Date: 2025-05-18 - Author: Arnon Shimoni - Summary: We announced when we were ready, generated >1000 signups and the momentum continued for weeks. So you want to launch a company in 2025? Let me tell you how we did it at Paid, where we raised €10M to revolutionize AI agent monetization.This isn't your typical "we got covered in TechCrunch and everyone clapped" story - this is my playbook, complete with what worked, what didn't, and what we had to scramble to fix mid-launch. ## The plan We approached our launch as a strategic chip to be played at exactly the right moment. As everyone should.You only have one chance to launch.We decided to launch with a party for our investors and several of our customers. We set the date to our March 25th launch party date, and then created clear milestones “working backwards”:- **March 25, 2025**: Launch party (with actual humans in the same room!)- **March 24, 2025**: We will have a PR and a TechCrunch article- **March 20, 2025**: Launch day assets finalized (intro video, images, banners, etc.)- **March 17, 2025**: Website and brand guidelines completion- **February 28, 2025**: Friends and investors identified and ready to be invited- **February 13-14, 2025**: Tour party locales, figure out size of party- **February 12, 2025**: Begin preparing content (photos, our manifesto)Our story is simple but powerful: AI agents are transforming software, but there's a massive disconnect in how they're monetized. Traditional SaaS pricing doesn't work for agent-based businesses, and we built Paid to solve this problem.Our friends and advisors pushed us to be more provocative: "The SaaS model is dead and not coming back." We identified clear "villains" in our story:- Hidden costs (LLMs, credits) that can't be managed- VCs demanding traditional MRR/ARR metrics for non-traditional businesses- Outdated seat-based pricing that makes zero sense for agents**📊 The Launch Process Explained**We organized our launch into five key verticals: ## Brand & Content Development - We prioritized having a clear and relatable manifesto that we can share, that will resonate with people. It went through 4 or 5 iterations before being finalized. We recorded it on the morning before the launch!- 30 seconds of pure value prop - a professional intro video, to bridge the gap between the messaging and the product itself (took 3 weeks to produce).Our animator [Rasmus Johan](https://rasmusjohan.com/) worked fast and we had a video in record time!- Brand guidelines and visual assets (to be completed within ## Strategic Thinking: Our three core value propositions: **M**onetization, **M**argins, and **M**omentum needed to be communicated consistently everywhere. The manifesto became our north star.For our video content, we focused on a simple but powerful message: "How do you monetize an AI Agent business? There is no blueprint for how to monetize AI Agents." ## What We Deprioritized: - Extensive podcast production pre-launch. Even though we were running with [Agent Talk](https://agenttalk.substack.com/), it’s production value will remain “just OK” for now. It was also a bit harder to get all of the guests we wanted before we launched - since they weren’t aware of the context behind what we’re doing.- “Sales enablement” material and long-form blogs will have to wait. No white papers or one-pagers yet! ## Online Presence ## What We Prioritized: - Website completion by March 17, a week before the launch. That’ll give us enough time to work out any bugs. [Litebox](https://litebox.ai/) were super helpful here!- CRM setup - we offloaded this to a contractor who knows how to do this fast!- Social media account creation (LinkedIn, X, Meta)- Launch day assets *finished* by March 20 so that we can just hit “Go!”- All of our employees know their role during the launch- Day-by-day messaging for launch week and subsequent weeks ## Strategic Thinking: Each founding team member had a unique perspective on why they joined Paid.We wanted to make sure those were captured in our messaging.We all think about things differently, and the different perspectives reach the respective networks differently. ## What We Deprioritized: - Automated marketing funnels with more than 2 steps.We decided to have a very simple flow that *could* drop some people, but wouldn’t overwhelm us or our users. A simple signup form is all we went with.- Multi-page website. A one-pager will have to do for now! ## PR & OutreachWhat We Prioritized: - PR firm engagement (Our investors EQT and Sequoia were super helpful in arranging the PR work)- Direct outreach to TechCrunch to get our story out there- Having all of the press briefing material prepared ahead of time- Podcast outreach strategy (20VC, Training Data-Sequoia, etc.)- Comprehensive Q&A preparation ## Strategic Thinking: We deliberately leaned into our networks (especially Manny’s) to distribute our message. We had specific angles for each podcast and news outlet.Some examples:- "Why great AI Agents will be about monetization"- "Building the GDP for the AI economy"- "Why AI Agent Monetization is the Next Big Platform Play" ## What We Deprioritized: - Broader press outreach. We preferred quality over quantity!- Using a "podcast house" to record podcasts more professionally. Guest quality mattered more. ## Event Planning (March 25 Launch Party) ## What We Prioritized: - Venue selection - it had to be *tight*- Little delightful details (a glowy sign, nice food and drinks, churro stand!)- Demo flow, videos, and exact messaging practiced- A "who's who" guide for the team, so that we could connect faster- Having a photographer ([Adam Kang](https://www.adamkang.com/)) on-site to capture the event- Creating a little bit of FOMO around us 🙇‍♂️ ## Strategic Thinking: The physical gathering needed to embody our brand values. We wanted a space where natural connections could form between investors and design partners.We arranged our own demo and launch around our early customer’s products.We gave them room to demo their wares, interleaved with ours - to keep everything interesting.We practiced our demo flow during the 2 days before the party to get the flow *just right.* ## **What I never got to finish:** - **Swag/merch** — I'm still annoyed about this one - there just wasn’t enough time to create good merch! I did get some stickers though. ## **👥 Engagement & Content Strategy** ## **What We Prioritized:** - Building a comprehensive list of friends and investors- Creating "what to say" materials for them, and telling them what we expected of them- Writing personalized posts for founding team members- Detailed hour-by-hour messaging plan for launch week:Tuesday: TechCrunch article + company announcement- Wednesday: Manifesto video + podcast release- Thursday: Our messaging framework- Friday: Post about how the response to the launch was- Monday: "Vibe Revenue" blog post - it was such a good podcast!- Tuesday: Product preview- Wednesday: Usual podcast release ## **Strategic Thinking:** We focused on "owning the week" with a carefully orchestrated content strategy.For our waitlist strategy, we designed a clear approach:- Set clear expectations with "Join the beta" messaging- Plan for a monthly digest email for the entire waitlist- Provide additional documentation to higher-quality leads (“3-4 stars”)- Maintain low barriers to entry for sign-ups, enriching data later- Target building a waitlist of thousands with 10% (500) as high-quality leads ## **What We Deprioritized:** - Community building (pushed to post-launch)- SEO and amplification strategies- Pay-to-play podcast opportunities ## **🏃‍♂️ Launch Day: What Actually Happened** For the actual launch execution, we followed these principles:- **Pre-preparation is everything**: All assets were completed in advance. Launch day was for engaging, not creating!- **Coordination timing**: All communications were carefully sequenced with the PR release as the trigger.- **All hands on deck**: Our entire team blocked the launch day for active engagement.- **Rapid response**: We committed to quick responses to every comment and message.But here's what actually happened: The PR release got postponed by a day, and everything happened within the same 3 hours.We still got **OVERWHELMED**.Meeting bookings were flooding our calendars while we were preparing for our launch party, and while we were trying to stay on top of our socials.The response was so big that we had to extend our "break time" before booking a meeting to 7 days to avoid being fully booked on the same day as our TechCrunch article went live.Even with this adjustment, we were fully booked for the next 2-3 weeks.Then, we had the party and all of the guests arrive at the same time.It was fun but it was not ideal.Our demos went swimmingly however!Party was excellent though! ## **📈 The Numbers: Holy Smokes!** The 24 hours of our launch exceeded expectations:- 15.8K visitors, 47K views (7,500 on first day, 5,800 on second day)- 2,700 new LinkedIn followers, 15,000 impressions- Manny's announcement post: 532,270 impressions, 3,280 reactions, 511 comments, 63 reposts- YouTube: 2,202 views, 14.9 hours watched- Inbound: >650 signups, >140 scheduled meetings- 45 attendees at our launch party, with 20 more on a waitlistWe’d reach >1000 signups within a week. ## **💸 The "Vibe Revenue" Concept** One of our most successful post-launch content pieces was about "vibe revenue" - a term coined by Pat Grady during a conversation with us. It refers to the initial surge of adoption and revenue when launching something new in AI, driven more by excitement than sustainable value creation.This concept resonated deeply because it highlighted a critical challenge in the AI space: the difference between initial excitement and long-term value. We identified three critical elements for sustainable momentum:- **Workflow integration** - not just features- **Expansion** - initial results aren't enough- **Outcome sharing** - don't trust usage metrics alone ## **🧠 What I learned from our launch** I mean, some of these should be obvious, but still.- **Timing trumps speed**: Wait to announce until you can capitalize on momentum.- **Quality over quantity**: Do *fewer* things exceptionally well.- **Personal activation beats mass communication**: One-on-one outreach to key supporters creates far more impact. Use your networks!- **Vision must match reality**: Carefully calibrate your messaging.- **Prepare for the "vibe revenue" phase**: Acknowledge that initial excitement will happen, but don't be fooled by it.- **Messaging cohesion is critical**: As our marketing advisor emphasized, "tighten messaging, more consistently harp on the same thing" and avoid "random acts of marketing".- **Plan for success**: We would use contractors to help get things to a good state. Brand design, website building, videos - these will have to be outsourced. This is not free - the time spent managing contractors was still significant.- **Leverage founder networks strategically**: Manny's connections were invaluable for amplifying our message. ## What I Would Do Differently Next Time - **Complete the swag**: Physical brand touchpoints matter (yeah, still annoyed).- **Assume your launch will go gangbusters. Prepare the nurture!**: Have a plan for maintaining engagement with waitlisted companies. We had enough, but we can still do better.- **Prepare more post-launch content in advance**: Having a content reservoir ready would have helped capitalize on the initial surge.- **Better prepare for overwhelming response**: We were fully booked for 2-3 weeks - having more capacity for inbound interest would have been valuable.- **Create a more structured task tracking system**: At some point I had to re-do the whole launch plan, because I was missing a high level overview of the individual tracks. It was in my head, not in an easy-to-follow list.I ended up with a very simplified checklist in Notion, with a calendar viewThis playbook represents our journey, with all its successes and lessons learned.While all this planning was crucial, **being able to adapt on the fly was equally important**. Having backup plans is key. No big launch goes perfectly according to plan! ## A checklist for surviving a launch in 2025 Finally, here’s a checklist you can follow and use, specifically for the final 48 hours: ## Team Readiness - All team members have blocked calendar for launch day- Clear roles assigned (who responds to comments on socials, who handles meeting bookings)- Everyone practiced their elevator pitch and key talking points- Team has access to all final assets and messaging documents- "Who's who" guide distributed to team for networking events ## **🎨 Content & Assets** - Website fully tested and live- PR release approved and scheduled- Social announcement posts drafted for all team members- Launch video finalized and uploaded to all platforms- Screenshots/images prepared for social sharing- Demo flow practiced at least 3 times ## Response Management - Meeting booking buffer time set (suggest 5-7 days as we learned!)- Auto-responders configured for high-volume channels- CRM tags created for tracking launch sources- Template responses prepared for common questions- Prioritization system for inbound requests established ## Tracking & Measurement - Analytics tracking confirmed working- Baseline metrics captured (followers, subscribers, etc.)- URL parameters set for different traffic sources- Screenshot tool ready for capturing moments/milestones- Designated person for collecting key stats throughout day ## Contingency Plans - Backup plan if PR gets delayed- Technical support contact if website issues occur- Alternative messaging if certain angles don't resonate- Capacity plan if response exceeds expectations (it likely will!) ## Post-Launch Momentum - Week 1 content calendar finalized and scheduled- Follow-up plan for key sign-ups and leads- Thank you messages prepared for supporters and amplifiers- Team debrief scheduled for 24-48 hours post-launch- Metrics review planned for 7 days post-launchRemember - while all this planning was crucial, being able to adapt on the fly was equally important. No launch goes perfectly according to plan, but having this checklist should help ensure that when things inevitably shift, you'll still capture the momentum of your big moment! --- ### [2025-05-08] 3 value frameworks: What we've learned from 120+ AI agent companies - URL: https://paid.ai/blog/ai-monetization/3-value-frameworks - Date: 2025-05-08 - Author: Arnon Shimoni - Summary: In recent conversations with founders building everything from warehouse robots to AI sales reps, one question keeps coming up: "How do I price my AI agent to actually capture its value?" Most of the people we speak to don’t know how to price their agents.They don’t know how because there are no frameworks or playbooks.After talking to over 120 AI agent companies and analyzing millions of agentic data points in just our first month at Paid, we've identified three frameworks that separate the companies thriving from those watching their margins get crushed.Whether you're building warehouse robots like our recent prospect, AI SDRs bombarding LinkedIn and outbound, or document processors handling the boring stuff nobody wants to do... these frameworks apply across the board.FrameworkWhat it measuresPricing impactWhen to useCommon pitfall** Direct human replacement**Labor costs saved by automationThis is your pricing floorAnytime. Most customers understand this.Stopping here and underpricing.**️ Enhanced performance delta**Additional specialized staff needed to match AI capabilities2-5x higher than basic replacement valueWhen your AI delivers insights, consistency of scale a single human can't matchFailing to quantify the "superpowers"** Unlocked Value Potential**New capabilities and outcomes previously impossible5-10x — highest possible valueWhen you can clearly attribute new revenue or savings to your agentAttribution problems ## Framework 1: The Direct Human Replacement Value > **"If a human was doing it, how much would it cost?"**This is the starting point for AI agent pricing, but too many companies stop here. It's pretty simple: figure out what it would cost to hire humans to do the same stuff your agent does.It's not just about volume of actions.If an AI calls 6793 people today and they all hang up on it, no value is created. But, if it makes only 4 calls and 2 of them open opportunities for the sales team worth $50k each, it just created $100k in value with a teeny tiny percentage of the output.For example, if your AI sales agent opens 10 qualified opportunities monthly and a human SDR (costing you $60K/year) only opens 5, you're delivering 2x the output. At minimum, you've created $60K in annual value.This is your pricing **floor**, not your ceiling.This approach makes sense to a lot of our customers (they get the math), but it limits how much you can charge.**tl;dr:** Start here, but don't stop here. This framework is just the beginning, and if you only use this one, you're leaving money on the table. ## ️ Framework 2: The Enhanced Performance Delta > **"If a human was doing it, and you're doing it better, what's that delta?"**This second framework recognizes that AI doesn't just replace humans, it often blows them out of the water in ways that would require additional specialized people.Our CEO Manny recently explained to an agentic founder recently: "When a BPO makes a call, does a transaction, hangs up, and moves on... your system is doing the same thing, but it's gathering all this information in the background, correlating it to other information, creating clusters of information, giving you insight."For you, your agent might:- Analyze 1,000 conversations and spot patterns no human could see- Generate insights that would normally need a data team- Never miss a detail, never have a bad dayTo get the same value, a company would need to hire analysts. And they're gonna sit in there for a week, and it's gonna cost you $20,000. And you just got that for free!**tl;dr:** Don't just think about the task being automated. Think about the superpowers your AI brings that would otherwise require additional specialized teams. ## Framework 3: The Unlocked Value Potential > **"What value does your agent unlock that wasn't possible before?"**This framework focuses on entirely new stuff your agent makes possible.Think: By seeing all these things, can I take action on these other things? Because that has a lot of value.Examples include:- Spotting problems before they happen- Finding revenue opportunities nobody would have seen- Enabling entirely new business capabilitiesWhile this framework has the highest potential value, it comes with a warning: it has an attribution problem that people get “wrapped around the axle”. Unless you know how to attribute, this could derail you from the conversation that you're trying to have.**tl;dr:** This framework captures the really transformative value of AI but gets tricky when you try to directly attribute it in pricing conversations. ## Turning 3 frameworks into actual pricing So how do you go from understanding value to setting prices?Paid’s data from successful AI companies shows a consistent approach:- **Always price below the full value you're generating, but above your cost**If you really understand the full value you're generating through all three frameworks, you'll probably realize you're constantly underpricing your solution.- **Show the value explicitly in every customer interaction**Our platform makes this possible through value receipts that show:This is the fee, this is the number of activities that you did that you're charging for, etc. More importantly: *This is the human equivalent value activity.*ffThis turns invoices from painful reminders into value demonstrations. You can say to your customer “even though you're paying me this much, **this is the amount of value you got.**"- **Align pricing to outcomes, not inputs**We’ve heard from our prospects and beta customers: "The whole kind of value-based pricing is something that we need to tap more and more into." ## The real-world impact You can drive real results, just like these companies- One AI SDR company doubled their revenue after shifting from platform pricing to charging for high-value activities- A customer service AI provider boosted their margins by 42% after properly accounting for the analysis value their system provided- A document processing company bumped renewal rates from 70% to 95% by showing the full spectrum of value they deliveredAgentic company typeBefore frameworksAfter frameworksKey changes made**AI SDR platform**- $2k/mo per customer- 70% gross margin- $4.5k/mo per customer- 76% gross marginShifted from seat-based to outcome-based pricing (meetings booked)**Customer service AI**- 20% annual churn- 11-month payback period- 8% annual churn- 6-month payback periodAdded value receipt showing analysis savings ($18k/mo) alongside each invoice**Warehouse robotics**- $X/robot hardware fee- $Y maintenance fee- Base fee + % of warehouse savings- Avg. deal size 3.2x higherMoved from "per head replaced" to % of total efficiency gains**Document processing**- $0.25/page processed- 40% AMR- $0.40/page + $X per insight- 68% AMRCreated tiered packaging based on complexity insight value ## Getting started today Try applying these frameworks today:- Map every activity your agent performs to its human equivalent cost- Identify the enhanced capabilities your system provides that would require specialized teams- Document any transformative outcomes your system enables- Create value receipts that make this visible to customersOne of our customers recently said this statement that we love:> “Education about value can't be *your* way. It has to be *their* way... You have to educate them in the words that they understand.”The AI economy is moving fast.We think who properly articulate and capture their true value will thrive, while those stuck in outdated pricing models risk leaving millions on the table (or worse, building businesses with negative agentic margins).Are you getting paid what your agent is actually worth? --- ### [2025-04-16] The Agentic Margin: What It Costs vs. What You Earn from AI Assistants - URL: https://paid.ai/blog/ai-monetization/the-agentic-margin-what-it-costs-vs-what-you-earn - Date: 2025-04-16 - Author: Arnon Shimoni - Summary: A straightforward way to measure whether your AI helpers are actually worth what you're spending on them. When companies build (and deploy) AI agents, they often focus on capabilities first and economics second.This is a bit like hiring employees based solely on their skills without considering their salary requirements. It’s possible, but it’s a recipe for trouble down the line.AI agents consume resources every time they perform tasks. They use computing power, make API calls, LLMs, text-to-speech, avatars, and so much more.All of these elements cost real money - much more than a SaaS. These costs scale not just with usage, but also with the type of inputs in ways that aren't always predictable or linear.**The agentic margin (AM) and agentic margin ratio (AMR) **are practical measures that tells you whether your investment in AI is actually creating value or quietly draining your resources.In our recent conversations we have seen too many companies rush to implement AI agents only to pull back months later when they realize the agents cost more to operate than the value they create.Without tracking agentic margin from day one, these issues often remain hidden. ## What makes agentic margin different from other metrics (including SaaS metrics)? Unlike traditional software that has predictable licensing costs, AI agents have a unique economic profile:- They typically charge by usage (tokens, API calls, compute time)- They consume multiple services simultaneously- Their costs fluctuate based on task complexity- Their usage patterns can change as user behavior evolves ## The Components of Agent Costs To calculate your agentic margin accurately, you need to understand all the ingredients that contribute to your agent's operating costs:- **Language models**: The "thinking" capabilities of your agent- **Voice processing**: Converting speech to text and text to speech- **Visual capabilities**: Understanding or generating images- **Infrastructure**: Servers and data storage- **Development**: Building and improving your agents- **Human oversight**: People who monitor and train agents ## A Tale of Two Support Agents Here is a real example (with details scrubbed) that we encountered when helping a software company evaluate their customer support agents:**Agent A** used meager resources, sticking to text-only responses and simple database lookups. It cost about $0.22 per customer interaction while resolving 65% of tickets without human intervention, while bringing in $5 per resolved ticket in revenue.Agentic Margin (AM): $5 - $0.22 = $4.78. That's an AMR of 95.6%. VERY healthy!**Agent B** used advanced features: voice capabilities, screenshot analysis, and could generate tutorial videos on demand. It cost $3.20 per interaction but resolved 85% of tickets, resulting in $5.50 revenue per resolved ticket.Agentic Margin (AM): $5.50 - $3.20 = $2.30, with an AMR of 31% - much lower than that of Agent A.Even when considering the resolution rate, and assuming only resolved tickets get you paid, you’d still be making a better profit:When building these agents, the company focused only on resolution rates and almost standardized on Agent B before realizing that Agent A's superior agentic margin made it a better value for deploying at a greater scale.It’s also true that as the cost of LLMs and tools go down, Agent B’s margins could improve.That’s why you have to continuously check the health of your agents, with different variants and models. ## Try this today: Making your agents healthier Here are some practical ways to improve your agentic margin:- **Right-size your capabilities**: Match agent features to actual user needs. If you are not sure which features/abilities get used, start tracking sooner rather than later.- **Continuously monitor capabilities:** As models get cheaper or develop new capabilities, you should have a continuous check of your agents’ metrics. You could be underpricing and overdelivering (or vice-versa).- **Cache common responses**: Don't regenerate the same answers repeatedly to prevent expensive API access- **Set sensible guardrails**: Prevent unnecessary use of expensive capabilities. Don’t allow users to paste in very long texts or very large images.- **Monitor usage patterns**: Watch for inefficient processes or unexpected costs. If you can’t monitor this on a per-interaction base, at least try to understand global usage patterns. ## Remember This The companies that succeed with AI won't necessarily be those with the most advanced capabilities, but those who understand the economic realities of deploying AI at scale. By tracking and optimizing your agentic margin, you're not just counting pennies, you're building a sustainable foundation for AI that actually delivers on its promise. --- ### [2025-04-10] Introduction to agent signals: The key to billing AI agents - URL: https://paid.ai/blog/billing/introduction-to-agent-signals-the-key-to-billing-ai-agents - Date: 2025-04-04 - Author: Arnon Shimoni - Summary: As agents take over the discussions, one challenge remains consistent for the businesses building them: how do you effectively monetize agent solutions? As agents take over the discussions, one challenge remains consistent for the businesses building them: how do you effectively monetize agent solutions?At the core of this shift? ***Signals***.Signals are a powerful concept that's transforming how AI agent companies measure, demonstrate, and bill for **value**. ## What are signals? The atomic unit of AI work **Signals** are (usually) discrete, measurable events that occur when an AI agent performs a meaningful action or reaches a significant milestone.In the Paid platform, signals are events that come in when an agent has performed a unit of work.Think of signals as digital stamps (or breadcrumbs!) that mark each valuable action your AI agent takes.A signal gets generated whenever an agent:- Completes a specific task- Reaches a checkpoint in a workflow- Produces an output- Interacts with a user or system- Achieves a meaningful outcomeEach signal captures rich metadata:- Timestamp- Agent identifier- Action type- Context (customer ID, conversation details)- Performance data (tokens consumed, processing time, input lengths, *costs and margins*) ## Why we've built our platform around signals An agent workflow often contains different actions and outcomes that need to be priced.Signals aren't just a technical feature, they’re a business innovation at Paid.There are 5 key reasons why we’re building them into our core: ## 1. Bill on what creates value With signals, you can price and more importantly bill based on:- Individual workflow steps (charge more for high-value steps)- Complete outcomes (only bill when value is delivered)- Volume-based activities (scale with usage)- Hybrid models (combine subscription + outcomes)The workflow image above shows how each step in a content creation process becomes a billable unit, with distinct pricing tied to the specific value delivered. ## 2. Transparency and auditability In a world of AI black boxes, signals create trust. Your customers see exactly what they're paying for.You can show the actual work performed rather than vague summaries and entitlements. ## 3. Value-aligned pricing Signals let you connect pricing directly to business outcomes that matter.Each meaningful action becomes a distinct signal. In the medical receptionist example above - from appointment scheduling to preventing no-shows.This creates multiple potential monetization points that your customers would care about. ## 4. Intelligent Business Decisions Signal data provide insights:- Which agent actions drive the most customer value?- Where are the performance bottlenecks?- Where are we not monetizing something we could?- How do costs compare to revenue for each action across customers and agents? ## 5. Evolving Business Models As your AI capabilities grow, your pricing can evolve without rebuilding your entire infrastructure.Pricing an outcome at $1 that delivers $2 of human-equivalent value creates an obvious win for customers while maintaining healthy margins for you. ## Prepare for the future: Figure out your value signals today Signals vary across different types of AI agent applications.I thought I’d give some practical examples of signals to monetize: ## AI SDRs / Sales AI agents - **Prospecting signals**: New lead identified, contact information verified- **Outreach signals**: Email drafted, email sent, follow-up scheduled- **Engagement signals**: Response received, meeting requested- **Outcome signals**: Meeting booked, deal advanced ## Document processing agent - **Input signals**: Document received, format identified- **Processing signals**: Text extracted, information categorized- **Output signals**: Summary generated, data entered into system- **Quality signals**: Accuracy score, confidence level ## Customer support agents - **Conversation signals**: Chat initiated, intent identified- **Resolution signals**: Answer provided, issue escalated to human- **Satisfaction signals**: Positive feedback received, problem resolved ## Start your signal strategy today Practically, here are things you can do now:- **Map your agent's core value stream** - Identify every interaction that creates tangible value- **Prioritize high-impact signals** - Focus on outcomes that customers would willingly pay for- **Design your signal architecture** - Create consistent naming and metadata conventions- **Start tracking** - Ensure signal emission at each critical step- **Experiment with pricing models** - Test activity-based vs. outcome-based approaches - even in a spreadsheet. Are you leaving money on the table?The question isn't whether to implement signals-based agent monetization, but how quickly you can make them the backbone of your AI business strategy. --- ### [2025-03-31] Vibe Revenue - a mirage of AI success - URL: https://paid.ai/blog/ai-monetization/vibe-revenue-a-mirage-of-ai-success - Date: 2025-03-31 - Author: Arnon Shimoni - Summary: So what happens when the vibes… stops? Pat Grady coined the term “vibe revenue” in a recent conversation with us.Vibe revenue is that initial surge of adoption and revenue when you launch something new and shiny in the AI world - a phenomenon driven more by excitement and novelty than by sustainable value creation.This concept resonated deeply with what we're building at Paid.Let me break down why vibe revenue matters and how it impacts the way we should think about building AI agent businesses.And more importantly, what happens when the vibes… stops? ## First, what exactly is vibe revenue? Vibe revenue is that initial burst of customer adoption and spending that comes from the "wow factor" of new AI technology. It's:- Fast growth that feels magical but may not be sustainable- Driven by curiosity and experimentation, not embedded workflows- Often characterized by poor retention and engagement metrics- A result of people trying something because it's novel, not because it solves their problems deeplyAs Pat put it:> "You go zero to a hundred overnight or whatever number it is. And then you look at the metrics and it turns out your engagement stinks and your retention stinks... It's because everybody's willing to try something... but the difference between having a magic moment and solving an end-to-end workflow in a unique and compelling way, that can be a pretty big difference."(Check out the full podcast episode here, where Pat describes it at around minute 12.)Have no doubt, many many companies will face this trap. ## The Momentum Trap: When Vibe Revenue Evaporates The inflection point will likely hit about 6-9 months after the peak.This is when:- Initial contracts come up for renewal- Novelty has worn off (or competitors caught up)- Customers evaluate ROI with cold, hard metrics- Stakeholders question whether to continue investingThis moment separates the vibe companies from the value companies.Without genuine ***momentum***, you're facing what I call the "renewal cliff" - that sudden drop when customers realize they've been paying for excitement rather than results.Not all is lost though. ## The 80/20 (”Pareto principle”) problem in AI development The 80/20 rule takes on a new meaning in AI. You can get 80% of the way there in minutes, creating that magical demo or first experience that gets people excited.But that last 20% - the part that creates actual stickiness and retention - might take weeks or months of refinement.This creates a dangerous trap for new founders:- You build something that creates initial excitement- You raise money based on early traction metrics- You discover your retention cliff when the novelty wears off- Your business hits a wall when growth flattensWith vibe revenue, you hit a peak very quickly - but the novelty wears off and the revenue drops. After that, a more sustainable upwards momentum is possible. ## What Creates True Momentum in AI Agent Businesses? Having worked with dozens of AI agent companies, I've observed that sustainable momentum comes from three critical elements: ## 1. Not just features. You need workflow integration for long term retention. The difference between a cool demo and a must-have product is how deeply it embeds into daily workflows. Companies retaining momentum have:- Connections to existing tools- Clear "jobs to be done" that happen repeatedly- Evidence of increasing usage depth over time (that means expansion!)**Practically:** Track not just how many people use your product, but how many different workflows they integrate it into. Each integration point creates another reason to renew. ## 2. Initial results are tricky. You have to expand for repeated engagement. Successful AI (and specifically agentic) companies demonstrate a pattern of expanding value:- Each month, the AI or AI agent handles more complex tasks- Users trust it with increasingly important decisions- The value proposition evolves from "cool tool" to "business critical"At Paid, we see this firsthand - customers who measure and demonstrate expanding value maintain 85%+ renewal rates, while those who don't struggle to maintain even 40%.**Practically:** Can you demonstrate with metrics that your models are improving based on usage? Can you prove to your customers that the more they use it, the better they get? ## 3. Don’t trust usage metrics alone. Share what your product does! True momentum comes from shifting the narrative from what their agent does to what outcomes it delivers, like:- Time saved per employee- Revenue generated or costs reduced- Specific business metrics that matter are improveAs Pat noted in our conversation: "The 80-20 might give you a magical moment, but you gotta get that last 20% to get people to stick, and that's where a lot of people are getting stuck."**Practically:** Look at cohort retention - What percentage of day 1 users are still active on day 7, 14, 30? Are people completing valuable workflows or just trying one feature? ## Building this elusive momentum measurement system To avoid the vibe revenue trap, implement a momentum measurement system from day one:- **Value Tracking:** Document specific outcomes achieved for each customer- **Expansion Metrics:** Track how users expand both usage depth and breadthThese metrics serve as early warning systems, helping you identify which accounts are building momentum and which are stuck in vibe territory. ## Escaping the “vibe revenue” Here are our practical recommendations for escaping the vibe revenue:- **Expect the vibe phase** - *Acknowledge* that it will happen, but don't be fooled by it. It won’t last.- **Look beyond top-line growth** - *Analyze* *retention* and engagement deeply.- **Invest in that crucial 20%** - The difference between magic moments and sticky products. *Design* *for* *integrations* from day one.- **Build trust mechanisms** - Make your AI and AI agent's work *transparent and verifiable.* If you can, make it reliable too.- **Find your value receipts** - Define how you measure and communicate value, and do it often. This should include every touchpoint, including your monthly invoices.- **Price for outcomes** - Align your revenue model with the value you create. Now is the time to consider outcome-based pricing. --- ### [2025-03-28] Paid Launch: Transforming the AI Agent Economy - URL: https://paid.ai/blog/company/paid-launch-transforming-the-ai-agent-economy - Date: 2025-03-28 - Author: Arnon Shimoni - Summary: Our launch attracted significant attention This week marked a significant milestone for our team as we officially launched [Paid](/), our platform designed for the AI agent economy.We’re so grateful for the extraordinary response, and we're excited to share the highlights of our journey so far. ## Our launch party On the eve of our launch, we brought together 40 industry leaders, investors, and tech enthusiasts who share our vision for the future of AI monetization. The energy in the room was legitimately *electric* as we demonstrated our platform's capabilities and discussed the potential impact on the emerging AI agent ecosystem.Steve Krenzel from Logic and Fabian Beringer from Vidlab7 showcasing their productsAmong our special guests were Steve Krenzel from Logic and Fabian Beringer from Vidlab7, two of our early customers who have been instrumental in shaping our product.Their insights and feedback have been invaluable, and we're grateful for their continued support.Check out our highlight reel from the party below! ## Overwhelming reception - the numbers from the launch day The numbers tell a compelling story about the agentic market's response to Paid:In just the first 48 hours after launch:- **15,800** unique visitors explored our platform- **47,000** page views were generated- **7,500** visitors on day one alone- Strong global interest with the US, UK, and India leading the wayOur LinkedIn presence exploded:- **2,600** new followers- **15,000** impressions on our first postsOur YouTube is also doing well:- Over **2,200** views- People spending nearly **15 hours** watching our explainer video and demonstrations, and our [podcasts](https://agenttalk.substack.com/).Meanwhile on our personal announcements, the [announcement post from our founder Manny Medina](https://www.linkedin.com/posts/medinism_im-officially-launching-my-new-company-activity-7310316285546938383-D-9u?utm_source=share&utm_medium=member_desktop&rcm=ACoAAAM5ft8Bbw558iJAW6BGuxeeMijzs2LWkG8) achieved remarkable engagement with over half a million impressions alone, and over 670k for the rest of the team members combined. ## Media Spotlight The launch captured the attention of prominent tech and business publications:- **TechCrunch**: ["Outreach founder Manny Medina has a new startup that helps AI agents get paid"](https://techcrunch.com/2025/03/25/outreach-founder-manny-medina-has-a-new-startup-that-helps-ai-agents-get-paid/)- **EU-Startups**: ["Paid raises €10 million to transform AI Agent economy and open beta programme"](https://www.eu-startups.com/2025/03/paid-raises-e10-million-pre-seed-to-transform-ai-agent-economy/)- **GeekWire**: ["Former Outreach CEO Manny Medina launches new startup tackling AI agent monetization"](https://www.geekwire.com/2025/former-outreach-ceo-manny-medina-launches-new-startup-tackling-ai-agent-monetization/)- **PYMNTS**: ["Paid Raises $10.8 Million to Scale Financial Infrastructure for AI Agents"](https://www.pymnts.com/news/artificial-intelligence/2025/paid-raises-10-million-dollars-scale-financial-infrastructure-ai-agents/)We've also been featured in newsletters from [Axios Pro Rata](https://www.axios.com/newsletters/axios-pro-rata-26f450e7-1623-4072-8080-00fe52830b20.html?utm_source=newsletter&utm_medium=email&utm_campaign=newsletter_axiosprorata&stream=top), [WSJ Pro](https://venturecapital.cmail19.com/t/d-e-sidurhd-tutkkkiduu-r/), [DailyAI Brief](https://dailyaibrief.com/news/paid-launches-platform-for-ai-agent-monetization-lrwyxjHn?utm_source=linkedin&utm_medium=social&utm_campaign=ai-brief-x), and [Sifted Daily](https://email.sifted.eu/uk-pledges-400m-to-defence-tech?ecid=ACsprvtgXt_1wFz_H-33RwbjVOH_jF2HWHngFVa7OOqGnmrO8igUe7nDqPxJgC7tP3k-9p5P-kHv&utm_campaign=Sifted%20Daily%20Newsletter&utm_medium=email&_hsenc=p2ANqtz--1gOHzqzK8x-IaQLeYlA-z2iOT-lipV_-pJhy08zrwzBj-eSa2UCLPkIpxC6yib3goQyY7Hyv1PnPj_sk2vSdTUBqM3gxoRJeipepuA0dnJKJpjoM&_hsmi=353742613&utm_content=353742613&utm_source=hs_email), expanding our reach to specialized audiences interested in AI innovation and investment. ## Customer Interest Perhaps most exciting is the influx of potential customers eager to explore what Paid has to offer:- Over **650** new contacts established- More than **140** meetings scheduled - meaning we’re fully booked for the next 2-3 weeks.Interestingly, many of these contacts are companies who are still in stealth mode! ## What's Next for Paid This launch is just the beginning. With our pre-seed funding of €10 million from EQT Ventures and Sequoia Capital, we're positioned to accelerate our development and expand our team.As Manny shared during the launch: "The numbers don't lie: The AI agent industry is about to explode. Three separate sources project a staggering $217 BILLION market by 2035. But this isn't just about future potential anymore."We're deeply grateful to everyone who has supported us on this journey so far. The enthusiasm and interest we've received confirm our belief that Paid is addressing a genuine need in the market.Stay tuned for more updates as we continue to build the financial backbone for the AI agent revolution. --- ### [2025-03-27] The AI Agent Builder's Manifesto - URL: https://paid.ai/blog/ai-agents/the-ai-agent-builders-manifesto - Date: 2025-03-27 - Author: Arnon Shimoni - Summary: The platforms of yesterday weren't built for us. They were built for a world where software supported human workers. **Traditional SaaS software is dead.**Together, we’re moving past building another SaaS tool. We are no longer augmenting existing jobs - we are creating agents that perform entire roles with efficiency that traditional software can never achieve.**We are AI Agent builders.**We are the founders and fighters who code AI Agents. We are built differently, just like the software we produce. We live in every region of the world. We travel light. Our teams are smaller and our reach is global. Our impact to employee ratio is higher. Our expenses are lower. Our ambition and courage is unmatched. We redefine what's possible in every industry and every role. Yet, we stand shackled by antiquated business infrastructure that fails to understand our reality.**The SaaS business model is irrelevant**AI Agents have transformed software. And anyone who tries to call it “Agents as a Service” ends up sounding like an AaaS. To add injury to insult, the traditional SaaS playbook, with its seat licensing and usage-based pricing, harms more than it hurts.- *Why measure “seats” when a single person can deploy dozens of agents performing complete job functions?*- *Why charge for “usage” when our agents autonomously handle complete workflows?***AI Agents have changed how software works**In this transformed world, agents- Take on complete roles and functions, targeting a market far larger than traditional SaaS- Use state-of-the-art language models to solve complex problems, making managing cost a priority- Work 24x7, handling tasks that once required entire teams- Deliver outcomes, not features, but we lack the tools to monetize and demonstrate the true value of this transformationThe platforms of yesterday weren't built for us. They were built for a world where software supported human workers.Today, our creations take on entire functions independently. There is no established playbook for how to run an agentic company.We get to invent a new paradigm. One that provides us with the business infrastructure we deserve:- Monetization - Billing, invoicing, and pricing models that reflect the true value of roles our agents perform- Margin - Solutions that enable us to run a healthy business- Momentum - Financial tools to show our customers the value we bring to their businessesThe potential of our AI agent economy stretches into the trillions. To our fellow AI Agent builders: Our time is now. Let's build this new economy and capture the full value we create.Innovate, create, get paid.*— Manny Medina, Manoj Ganapathy, Arnon Shimoni, Raj Dosanjh* --- ### [2025-03-25] Paid raises €10M to build the business engine for AI Agents - URL: https://paid.ai/blog/company/paid-raises-10m-to-build-the-business - Date: 2025-03-25 - Author: Arnon Shimoni - Summary: We're thrilled to announce that Paid has raised €10M led by EQT, Sequoia Capital and with participation from GTM Fund, Exceptional Capital, and strategic angel investors to build the financial infrastructure powering the next wave of AI. We're thrilled to announce that Paid has raised €10M led by EQT, Sequoia Capital and with participation from GTM Fund, Exceptional Capital, and strategic angel investors to build the financial infrastructure powering the next wave of AI. ## The AI Agent world needs new business infrastructure AI agents are completely transforming software. They're not just tools that help people work; they're taking over entire roles and functions independently.But there's a massive disconnect:While the AI agent market is exploding toward $47B by 2030, most builders are stuck using outdated billing methods designed for traditional SaaS. They're manually invoicing customers, using basic payment processors, or charging per transaction - approaches completely misaligned with the value their agents deliver.When AI agents replace an entire team while consuming expensive LLM resources, seat-based pricing makes zero sense. When software performs complex work worth thousands of dollars per output, transaction-based billing breaks down. ## Introducing Paid Paid is the all-in-one business engine for AI agents that solves these challenges. With just a few lines of code, Paid handles your pricing, subscriptions, margins, billing, and renewals - everything needed to monetize AI agents effectively.We support AI agent builders in three key ways: ## Monetization: Easy Pricing and Packaging Capture the full value of what your agents deliver with flexible pricing models that make sense. Want to charge for outcomes instead of inputs? Need hybrid subscription/performance pricing? Charging per agent? We've got you covered. ## Margins: Pricing and Packaging Simulator Our pricing simulator helps you iterate and experiment with different packages to find the optimal balance. Know exactly what you're spending on tokens and compute for each agent, so you can set prices that ensure healthy profits.Paid helps you track your margins on a customer-by-customer and agent-by-agent level of detail ## Momentum: Client Portal and Value Demonstration Our client portal handles all invoicing while demonstrating the actual work your AI agents have performed. We transform invisible AI value into tangible ROI metrics that make renewal conversations easy and keep competition out. ## Who we are Our founders bring unique experience to this challenge:- **Manny Medina** built [Outreach](http://outreach.io/) from scratch to over 6,000 customers, 220,000 active users, and $250M in ARR. He experienced the AI agent pricing dilemma firsthand.- **Manoj Ganapathy** built InvoiceIT, which was acquired by Steelbrick and later became Salesforce Billing. He has 10+ years of experience building billing systems.- **Arnon Shimoni** built monetization systems at Pleo and Storytel- **Raj Dosanjh** YC Alumni & ex-Palantir ## Early success stories We're launching with several innovative AI Agent builders already using our platform.Companies like [Logic](https://logic.inc/), [Artisan](https://www.artisan.co/), [Vidlab7](https://www.vidlab7.com/), and [HappyRobot](https://www.happyrobot.ai/) are already using Paid's platform to transform how they monetize their AI agents. ## Innovate, create, get paid. The playbook for running an agentic company doesn't exist yet. We're writing it together. ---