AI agents can automate work, but full autonomy is rarely the right default. Many production actions still need human judgment, approval, or post-hoc audit because the cost of a mistake is high or the context is incomplete. Human-in-the-loop (HITL) design makes those control points explicit.

Human-in-the-loop systems combine automation with explicit human control points. The agent can do the routine work, but humans review outputs, approve risky actions, resolve ambiguity, and correct mistakes that become training or evaluation data.

This chapter covers approval workflows, confidence-based escalation, asynchronous review queues, correction loops, autonomy levels, and the UX patterns that keep human review useful instead of performative.