What does the approval policy screen control?
Where the line sits between what the AI workforce does on its own and what waits for a person.
Every agent has a risk ceiling. Below it, the action just happens and is logged. At or above it, the agent stops, writes down what it wants to do and why, and a person answers from Approvals. This page is where an admin sets that ceiling, action by action.
Tightening a line here does not undo work already approved — it only changes what happens the next time that action comes up. Loosening one is worth a second look with Workforce open beside it, so you can see what the agent has actually been asking for lately before you wave more of it through automatically.
This screen only appears for firms with the AI workforce turned on. Without it, every action in the product is a person's, the way it always was.