Human-in-the-loop

Human-in-the-loop (HITL) is a design where a person reviews, approves, or corrects an AI agent’s actions at defined points rather than letting it run fully autonomously.

HITL is used where errors are costly or irreversible — refunds, legal filings, code merges — to trade some speed for control and accountability.

A well-designed loop escalates only the cases that need a human, preserving most of the cost advantage of automation.