Set and interpret spend caps
Spend caps prevent surprise bills. You set them once and Nomic enforces them in the background; users and admins see warnings as caps approach.
Cap levels
You can set caps at three levels:
- Organization — the total your tenant can spend per billing period.
- Per-user — the limit a single user can drive in a period.
- Trial / pilot — a one-time ceiling Nomic and your account contact agree on at the start of a pilot.
Caps are additive in the sense that the most restrictive one wins: a user can never exceed their per-user cap, and the whole org can never exceed the org cap.
Set caps
- Open Admin → Billing → Spend caps.
- Set the org-level cap and (optionally) a default per-user cap.
- Override per-user caps for specific people if a few power users need more headroom.
Caps reset on the billing-period boundary.
What happens when a cap is hit
- At ~80% of cap: admins and the affected user get a warning.
- At 100%: new runs are blocked until the cap is raised or the period resets.
- In-flight runs continue. A workflow already running will finish — Nomic doesn't kill in-progress work to avoid partial / wasted results. This means actual spend can land slightly above the cap if a large run was already underway when the cap was reached.
If you need a hard stop in the middle of a run, cancel it from the run's page; spend stops accruing within a few seconds.
Interpret caps in reports
- "Spend this period" is what's posted so far in the current billing window.
- "Reserved spend" is the headroom blocked by in-flight runs. When a run completes, reserved spend moves into posted spend.
- "Available" is
cap - posted - reserved.
Practical tips
- Start trials with a conservative org cap; raise it once usage patterns are clear.
- Set per-user caps that match the work — power users running drawing reviews need more headroom than occasional users.
- Use Admin → Analytics → Workflows to identify whether a single expensive workflow is driving most of the spend before you change caps.
- If several users hit caps quickly, check whether they are indexing large files or running broad workflows before raising the limits.