Knowledge base
What is human-in-the-loop?
A design principle: AI does the executing work, but a human keeps the decision moments that matter.
Definition
In one paragraph.
Human-in-the-loop is a design principle: AI does the executing work, but a human keeps the decision moments that matter — approving what goes out, stepping in on doubt, and setting the boundary of what the system may do itself. It is the difference between handing over work and handing over control.
Why it matters.
Where the human sits
Not everywhere, but at the moments that matter: publication, money, irreversible actions.
Trust grows on evidence
The boundary only moves once the system has proven itself — action by action.
Our own policy too
At our agency a human approves everything that goes out. Agents propose; humans decide.
We wrote a full dossier on it, with worked examples. Read Trust & Verify
All terms
Back to the knowledge base overview.