AI agents and technology
What is human in the loop? Definition and examples
Also: HITL, human approval
In short
Human in the loop means a person reviews and approves an AI process at defined points before it continues. AI prepares, the person decides. Responsibility and control stay with you, even when AI agents take on a lot of the work.
Why it matters: You use the speed of AI agents without handing over responsibility.
- 01
AI proposes
Draft with source and reasoning
- 02
Person reviews
Sees the exact change
- 03
Person decides
Accept or reject
- 04
System applies
Only approved changes are written
Why does Human in the Loop matter for your company?
AI agents make mistakes, and they make them fast and at scale. Approval at the right points catches them before an offer is sent, a ticket is closed or a document is overwritten. The EU AI Act also requires human oversight for high-risk systems.
What matters is the dose. If a person checks every step, you lose the benefit, and if nobody checks, you lose control. Good approvals sit where a decision has consequences.
Example: The roadmap after the strategy decision
After the decision “Strategy 2027: focus on the DACH mid-market, no US expansion” an AI agent finds many tickets, roadmap entries and documents that still assume the US expansion. It could change all of them itself.
Instead it proposes every change as a draft, with source and reasoning. The owners accept or reject, and only then does the source change. Whatever stays open is visible with its owner.
3 common mistakes to avoid
Approving everything
Whoever has to check every detail soon just clicks through. Approvals belong where errors have consequences.
Approving without context
A person can only check what they understand. Show the source, the reasoning and the exact change.
Unclear ownership
When everyone may approve, nobody feels responsible. Each topic needs a named person.
Step by step
- 1
Name the steps with consequences
What leaves the company, changes data or costs money?
- 2
Assign owners
One person per topic who approves or rejects.
- 3
Show changes as proposals
With source, reasoning and the exact difference to today's state.
- 4
Record decisions
Who approved what, and when? That keeps every change traceable.
Bring one important decision.
In 30 minutes with Marc or Simon we show where your work drifts from the decision today.
Meet the Company BrainRead on in the PM Lab
Articles that go deeper.
Related terms
- AI agent →An AI agent is a program that uses a language model to pursue a goal, plan its own steps and use tools to get there.
- AI governance →AI governance is the set of rules a company uses to decide who may use AI for what, with which data and under whose responsibility.
- Hallucination →A hallucination is a statement by an AI that sounds convincing but is false or unsupported.
- Organisational memory →Organisational memory is what a company knows and keeps about itself: decisions, experience and reasons, independent of individual people.