AI Pulse · Business Models & Winners

Kyle Poyar’s 2026 State of B2B Monetization survey asked 230 companies how they price their software. The most common answer, at 37%, was hybrid models that mix seats, usage, and sometimes feature tiers. Pure per-seat pricing is no longer the default, and usage-based keeps climbing. This is not a sudden shift, but it is now the norm, and that changes how you think about buying AI tools.

When a vendor offers hybrid pricing, you need to model what you will actually pay under realistic load. A seat-based contract is predictable, sometimes too predictable if half your licenses sit idle. Usage-based can scale beautifully or surprise you with a bill that doubles when adoption takes off. Hybrid tries to split the difference, but it also means you are negotiating two or three levers at once. The pricing model itself becomes part of the product decision, not just a back-office detail after you have chosen the tool.

This matters more with AI products because usage can be lumpy and hard to forecast. One team runs a thousand queries this month, ten the next. Another uses the API heavily for two weeks during a campaign, then goes quiet. If the vendor charges per API call or per token, you need visibility into who is doing what and when. If they charge per seat but gate the high-volume features behind a usage add-on, you are back to hybrid whether you wanted it or not. The Omni Command Centre approach we build helps companies track which tools are actually earning their cost under whichever model the vendor picked, so you are not flying blind when renewal comes around.

The takeaway is simple. Ask the vendor to show you three real-world scenarios with different usage patterns and what each would cost. If they cannot, the pricing model is still too new or too vague, and you should wait or negotiate a cap.