When AI pricing became the hottest topic in London

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 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 models

Most 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" - that's exactly the right attitude!

We're planning more events like this, because when founders get together to solve real problems, magic happens.