Human-in-the-loop
Human-in-the-loop means a person reviews or approves AI outputs or actions before irreversible effects occur.
How it works
Define which actions auto-run and which pause. Show evidence to approvers. Log decisions. Feed rejects back into the evaluation set.
Example
An agent prepares a customer credit; a manager approves in a queue before finance systems change.