What is an Agent Approval Gate?

Agents can plan and execute long sequences of steps on their own, but not every step deserves the same level of trust. Reading a file or drafting a message is low-risk and reversible; sending an email to a customer, executing a payment, or deleting a record is not. An approval gate draws that line explicitly: the agent still does the reasoning and prepares the action, but execution stops and waits for a human (or another authorized agent) to confirm before anything happens outside a sandbox. Well-designed gates show the approver exactly what will happen if they approve — the actual email text, the actual payment amount and recipient — rather than a vague summary, because an approval given without real information isn't meaningful oversight. Gates are usually configurable per action type and per tenant, and they must never silently auto-expire on a short timer, since a stalled approval should block the action, not let it slip through by default.