Definition
A design principle where a human validates the critical decisions or actions of an AI system before they are executed.
Human-in-the-loop (HITL) places a human control point in an automated process: the AI prepares, the human decides. Applied to AI agents, this means an email drafted by the agent is only sent after review, and a proposed transfer is only executed after approval.
It is both a growing regulatory requirement (the AI Act mandates human oversight for high-risk systems) and a matter of operational common sense: the AI's autonomy is calibrated according to the reversibility and exposure of each action, not according to the trust placed in the model.
All AI models, deployed on your infrastructure, with your documents connected.