Human in the loop
A design where a person approves or corrects what an AI system does before the result takes effect.
What it tells you about a developer
A candidate who can say where their team put the checkpoint and why has thought about what their system does when the model is wrong. That judgment is the work in agent engineering.
Human in the loop names where the approval sits. An agent that drafts an email for someone to send has a human in the loop. An agent that sends the email itself does not. Teams place the checkpoint according to what a mistake costs: code going to production usually keeps a person in the loop, and a first draft of a test often does not.