Appearance
Budgets and spend caps
Give the org and each team a monthly AI budget and watch spend against it. AI caps live on the AI caps tab at Finance → Budgets (/plan/main/finance/budgets/ai-caps). The tab is admin-only: you need the AI policy admin permission to view and edit caps.
A cap is a shared pool across providers. A team's budget covers Anthropic, OpenAI, Cursor and everything else Flowstate sees, together. You set one number per team, not one per tool.
What's on the page
The page opens with a plain-English answer: how much of this month's organisation cap you have used, and which teams are over or close to their budget.
Organisation cap. One org-wide monthly cap with a colour-banded bar: green within budget, amber close, red over. Once the month has a few days of data, the page projects where you will finish the month at the current run rate and says whether that lands within the cap.
Budgets by team. One row per team:
| Column | What it shows |
|---|---|
| Team | The team, worst standing first |
| Used this month | Month-to-date spend against the cap, as a colour-banded bar |
| Monthly cap | The cap figure, editable inline |
| Status | OK (under 80%), CLOSE (80% or more), OVER (at or past the cap), or NO CAP |
Editing caps
Type a new figure in a team's cap field and save. You can batch several edits into one save. Every save is recorded as a new policy version, so cap changes are audited: who changed what, and when.
Caps scoped to a single person or a single AI service account are authored as rules on the Rules tab (Resourcing → Agentic workforce → Rules) rather than on this page. See Policy and enforcement.
What a breach does
Caps nudge before they punish. Anyone over their cap gets a nudge first, then their manager is notified. There is no hard stop unless you set one as an enforced rule.
- The subject sees their own standing on My AI (
/me/ai-agents). - Budget-breach signals land in the alerts feed at Insights → Agent insights → Alerts, alongside the other signal categories. See Alerts and signals.
Agents count too
An autonomous Agent's spend is its own line, attributed to no person, and each project deployment can carry its own monthly spend cap set in the deploy flow. Copilot spend counts against the team of the person who used it. See The AI cost model for how the two never double-count.
What it isn't
An AI cap is a spend control, not bookkeeping. It is not the same thing as a Finance budget snapshot, which is a versioned record of what the books say should happen. Budget snapshots support variance analysis and quarter-end reporting. AI caps watch the spend as it happens.
Cross-references
- Policy and enforcement — where rules live and how versions ship.
- Alerts and signals — where breaches surface.
- My AI — the subject's view of their own standing.
- Finance budget snapshots — the bookkeeping cousin of AI caps.