Credits, not seats: usage pricing for the agent era
Seat-based pricing made sense when humans clicked buttons. Agents do not. We moved Unyly to a unified credit model — and baked the Sael efficiency advantage straight into the price.
Seat-based SaaS pricing assumes a human at a keyboard. In an agent workflow there is no human in the loop for most calls — one person can drive thousands of tool invocations a day. Charging per seat either bankrupts you or starves the product.
The model
We switched to a unified, Claude-style credit model. Each plan includes a monthly credit allowance that resets every 30 days:
| Plan | Price | Credits / mo |
|---|---|---|
| Free | $0 | 1,000 |
| Pro | $12 | 25,000 |
| Team | $49 | 150,000 |
Heavy users top up beyond the allowance from a dollar wallet. Light users sit on Free. The funnel stays wide and the tail monetizes itself.
The part we like most
Credits let us price the protocol honestly. A call through Sael costs fewer credits than the same call through MCP — because it genuinely consumes less: fewer round-trips, less re-sent context, server-side composition instead of client-side chaining.
| Action | MCP | Sael |
|---|---|---|
| Runner invocation | 10 | 4 |
| Tool call | 2 | 1 |
That is not a discount. It is the measured efficiency of the protocol turned into a number you can see on your bill — and a gentle push toward the better architecture. (We benchmarked it: Sael vs MCP.)
Why usage-based wins
- Predictable COGS. A credit maps to a unit of compute; margin stays controllable.
- Net revenue retention > 100%. Consumption grows inside an account without new sign-ups.
- A real funnel. Free is genuinely usable; you pay when you get value.
Pricing is at unyly.org/#pricing.