Pricing
How AI credits work
AI spend is metered, visible and capped by you. Here is the actual arithmetic — no “contact sales for AI pricing”.
01 · The pools
What each plan includes.
Credits are granted per seat, per billing cycle. Unused credits do not roll over.
| Plan | Per seat / month | Included credits / seat | Beyond the pool |
|---|---|---|---|
| Starter | $29 | 5,000 | Hard cap — work stops, no bill |
| Pro | $99 | 20,000 | $0.0030 per credit |
| Agency | $179 | 45,000 | $0.0025 per credit |
Starter is deliberately hard-capped: a solo estimator should never receive a surprise invoice.
02 · The controls
Four ways spend stops.
Every limit refuses with a named code rather than failing quietly or billing on.
hard_cap
The plan has no metered overage. At zero credits, AI calls refuse. Nothing accrues.
overage
You have passed the included pool and set an overage ceiling. At the ceiling, calls refuse.
ai_budget_exceeded
A budget you set — global or per department — has been hit. Warn first, then cap, your choice.
subscription_inactive
The subscription is not in a payable state. Read access continues; metered work stops.
03 · Why it's built this way
Metering you can audit.
Per department
Spend is attributed by department and by model, so the ELV team's usage is not hidden inside a company-wide number.
Per model
Cheap and premium models are metered separately. Routing a task to a local model is visible in the same ledger.
Exportable
The full usage ledger exports to xlsx — hand it to finance without a screenshot.
Alerts before caps
Budgets warn at a threshold before they block, so a cap is never the first thing you hear about.
See the metering working
The demo gallery shows credits, budgets, overage and the block codes as live screens.