Why Ontu prices per seat, and still doesn't meter the AI

Per-seat pricing gets a bad reputation because most tools pair it with a second constraint: cap the usage too. Add a person, pay more. Use the AI more, pay more again, or just hit a wall. It's the model that makes teams ration access to the tool that's supposed to make them faster in the first place.

Ontu keeps the two decisions separate. A seat is for a person doing real work in a project: drafting, reviewing, giving feedback. Once someone has a seat, the AI itself isn't metered. There's no token cap, no per-query cost, and no add-on tier to unlock more research, just a fair-use provision that catches extreme outliers rather than getting in the way of normal work.

Seats come in two tiers. Team covers the day-to-day, with full AI access on every project you're on and inline feedback on anything it drafts. Partner adds the firm-wide layer on top: SOP libraries, standing instructions, role templates, analytics, and who else gets a seat. Both are billed annually and priced per seat, rather than negotiated deal by deal.

What you end up with is a plan that scales with headcount the way most software does, without punishing anyone for actually using the AI once they're in.