Skip to content

AI budgets

Cap how much a workspace spends on AI each month. Set two separate limits: one for AI nodes inside workflow runs, one for the AI builder chat. Open Settings → Workspace → AI Budgets.

The AI Budgets card with the AI Agent nodes, AI Builder and Warning threshold fields, the spend so far this month, and Save AI Budgets.The AI Budgets card with the AI Agent nodes, AI Builder and Warning threshold fields, the spend so far this month, and Save AI Budgets.

The fields

Field What it limits
AI Agent nodes (workflow runs) Monthly spend, in US dollars, by AI nodes running inside workflows
AI Builder (chat) Monthly spend, in US dollars, by the AI builder chat
Warning threshold (%) The share of a limit at which Flero records a warning. Default 80.

Leave a field blank for no limit. Zero also means no limit.

Below the fields, a line shows the spend so far this month, for example: "This month so far: AI Agent nodes $12.40 · AI Builder $3.10 across 4 builds ($0.78 per build)."

Only the workspace owner can save the budgets. Click Save AI Budgets.


What happens at the limit

The two limits are independent. A busy month in one does not stop the other.

Limit reached Effect
AI Agent nodes Flero checks the budget before each AI node runs. When the month's spend has reached the limit, the node fails with "Workspace monthly AI Agent node budget exceeded: $spent / $limit. Increase the AI Agent node budget in workspace settings to continue." The node's error handling then applies.
AI Builder Flero checks before each builder model call. The chat shows a message that the AI Builder budget is exceeded, and the builder stops until the limit is raised or the month rolls over.

Spending resets at the start of each calendar month.


Tips & gotchas

  • Set a limit on any workspace with scheduled AI workflows. A loop that calls an AI node many times can spend quickly while nobody watches.
  • Plan for the failure. A workflow whose AI node hits the limit fails like any other failed node. Set the node's error mode in Node Settings if the workflow must carry on. See Retries & error handling.
  • For per-run caps, add a Token/Cost Guard node in front of expensive AI calls.


Found something out of date? This page lives in the Flero docs content set.