The AI Usage Report shows what your AI features are costing and who is using them. Navigate to Third Party > AI Usage Report; viewing it requires the Manage AI Settings permission (View level is enough — Full is only needed for the AI Providers configuration page).
Every AI call in Powercode is logged with its provider, model, token counts, computed cost, the feature that made the call, and its outcome. No prompt or response content is stored — the log is metadata only.
Constrain the report by:
- Date range (From / To)
- Provider — Anthropic Claude, OpenAI, or Google Gemini
- Feature — which AI feature made the call (e.g. ticket summarize, editor rewrites, report generation)
- User
- Headline Totals — total calls, Enterprise cost (spend on the company key), BYOK cost (informational) (spend on users' personal keys, shown for visibility but billed to the users' own accounts), input and output tokens, and the count of errors/blocked calls.
- By User — spend per user, with a Cap consumed indicator showing how much of their monthly enterprise-key cap each user has used.
- By Feature (CallerContext) — spend broken down by which feature made the calls, so you can see whether the money is going to ticket summaries, editor rewrites, or report generation.
- By Provider / Model — spend and token usage per provider and model, useful when deciding which default model to configure.
- Recent Errors & Blocked Calls — the most recent failed calls (provider errors, invalid keys) and blocked calls (cost cap reached, permission denied). This is the first place to look when a user reports an AI action failing.
- Costs are computed at the time of the call from the provider's published per-million-token prices, so the report needs no access to your provider billing account.
- Quota math (the Cap consumed column) counts only successful calls made on the company key. Personal-key (BYOK) usage is logged and shown, but never counts against a cap.