Usage-based pricing is the default in software and it is a reasonable default. In a clinic it has a specific failure: it makes every use a small decision, and clinicians respond by not using it.
The behaviour a meter creates
If each reading costs something visible, the agent gets used for the complicated cases and skipped for the routine ones. The routine ones are where the aggregate time actually goes, so the meter removes most of the value it was meant to capture.
What a clinic actually knows
A clinic does not know how many documents it will read next month. It knows how many doctors it has. Pricing per clinician is priced against something the buyer can count, budget and predict.
Being honest about the cost side
An agent that runs through an entire consultation is substantially more expensive to serve than a chatbot answering a question. A seat price set by intuition can easily cost more to deliver than it earns, and the only way to know is to measure real consultations rather than model them.
So the price gets validated against actual usage with actual clinics before anyone builds a plan on it. Anyone quoting a confident per-seat figure for agentic clinical work before doing that is quoting a guess.
Where per-unit pricing does fit
Where the buyer genuinely knows the volume. A laboratory knows roughly how many reports it issues, so per report is a fair unit there. The principle is the same: price against something the customer can count.
This article explains how our technology works. It is health information, not medical advice.