Pricing in the age of AI: founders dinner Amsterdam

Our first private dinner. Around fifteen tech founders, operators and investors around one table at CACHÉ Amsterdam, and one question: how do you price in the age of AI?

No slides. Just the questions people are actually stuck on.

What came up

• Usage-based pricing is no longer optional, but nobody agrees on the unit. Tokens, actions, outcomes, "agent-equivalents". Pick the wrong one and you either leave money on the table or train customers to use less of the thing you want them to use more of.
• MCP is the new "do we build an API?" debate. Open it up and you become infrastructure. Keep it closed and you risk irrelevance. There is no neutral answer, only a strategic one.
• "We'll just build it ourselves with AI" is the new objection. The answer is not a discount. It is a sharper articulation of what is actually hard: the data, the workflows, the institutional knowledge that does not live in a prompt.
• The price of a human hour is being repriced in real time. If your pricing is anchored to seats or hours worked, that anchor is moving underneath you.
• Should you price for agents? For some businesses agents are the new buyer and pricing has to be legible to a machine. For others humans stay in the loop, and as AI commoditises the easy work the human touch becomes the premium.

What struck us most: the founders in the room were not waiting for the dust to settle. They were rebuilding packaging, rerunning price metrics, and rethinking what "a customer" even means.

This was the first. It will not be the last — the next one is Revenue Rendezvous Amsterdam.