Usage Caps
How AI runs are counted and what happens when you reach your monthly limit.
What is a "run"?
Your plan includes a monthly allowance of AI runs. Each metered AI operation reserves one run before it is processed. The allowance is shared across the whole platform, not held separately per feature.
| Action | Counts as a run? |
|---|---|
| Advisor question | Yes — one run per question |
| Policy Hub gap review | Yes — one run per gap-review request |
| Template parsing / KPI extraction | Yes — parsing a template is AI work and is metered |
| AI report narrative | Yes — metered when a report's narrative is generated |
| Data submission or GP-data file that needs the external document parser | Yes — one run when the parser is invoked; returned if the parser fails before returning a result. An unchanged file your organisation has already parsed successfully is served from the stored result and does not use another run |
| Data Quality Check | No — unless it runs on an uploaded file that needs the external document parser (the row above then applies); re-running the check on an unchanged file that was already parsed successfully does not use another run |
| Uploading, mapping, approving, downloading | No — these steps themselves consume nothing |
Because template parsing spends from the same allowance, a heavy set-up month — many templates uploaded at once — draws down the same pool your Advisor questions come from. Plan accordingly.
Monthly limits by plan
| Plan | Monthly runs |
|---|---|
| Bot Only | 500 runs/month — web Advisor not available on this plan |
| Policy Hub | 2,000 runs/month |
| Full | 10,000 runs/month |
Note — Bot Only plan: AI interactions on the Bot Only plan are delivered through the integrated WhatsApp channel. The web-based Advisor and Policy Hub are not included on this plan. The 500-run monthly allowance covers those WhatsApp interactions.
Your run count resets on the first of each month. You can see current usage in your fund's workspace and in the Advisor when you are approaching the cap.
When you reach the cap
Once your monthly runs are used up, metered AI work pauses until the next period:
- Advisor — new questions are declined with a notice that the monthly AI run limit is reached.
- Policy Hub — new gap reviews cannot be generated.
- Template parsing and report narrative — the automatic end-of-cycle chain checks the allowance before spending, and records that it stopped because the cap was reached rather than failing silently.
- Data submission and GP-data files — a file that needs the external document parser is not sent to it; the upload is refused with a notice that the AI usage cap is reached. An unchanged file your organisation has already parsed successfully is still served from the stored result, because that spends nothing.
The cap is checked before a run is spent, so a request that would exceed it is refused rather than half-processed. Runs already in progress are unaffected, and your full allowance returns on the first of the next month.
Increasing your runs
If you need additional runs during the current month, contact the Nia team. They can review your plan and discuss options for your fund.