Human in the loop
Definition
Human in the loop is a setup where an AI system pauses at set points so a person can approve or correct its work before it takes effect. It is used for actions that are hard to undo, such as sending emails to customers or moving money.
By Ronan Liedmeier, fractional head of growth and course author · Published 17 October 2026 · Last updated 17 October 2026
Also called: HITL, human approval step
It lets a team put AI on real tasks before trusting it fully, because mistakes get caught at the approval step instead of by a customer.
How it works
The workflow runs on its own up to a decision point, then shows a person what it is about to do and waits. Good approval screens show the exact change, a preview often called a dry run: the email as it will be sent, or the budget before and after. The person approves or rejects, and the choice is logged.
For the riskiest actions, such as deleting an ad account or a customer record, the safer setup keeps the action human-only. The AI never gets that tool, so there is nothing for a reviewer to approve by accident. Pair approval steps with guardrails that run as code, since a person can only review what reaches them.
Example
A paid media manager has an AI agent that proposes budget changes across her ad campaigns each morning. The agent can read performance data and prepare changes. Any change above a set amount appears in the team chat as a before-and-after with approve and reject buttons, and nothing touches live spend until she clicks.
Common mistake
Treating a confirmation prompt as enough protection for destructive actions. By the fortieth routine approval of the day, reviewers click approve without reading. Save approvals for actions worth a close look.
Learn it in AI Academy
- Guardrails, Permissions & Safety ArchitectureExpert · 8 min
- Simple GuardrailsAdvanced · 15 min
Related terms
- AI guardrails
- AI guardrails are checks and limits placed around an AI system that block or flag unsafe or wrong inputs and outputs before they cause harm. The strongest ones run as code outside the model, so a cleverly worded request can't talk its way past them.
- AI agent
- An AI agent is an AI system that works toward a goal by choosing its own next steps: it picks a tool, uses it, looks at the result and repeats until the task is done. You set the goal and the limits, and the agent works out the path.
- AI workflow
- An AI workflow is a fixed series of steps, set in advance by a person or by code, in which an AI model handles one or more of the steps. The steps run in the same order every time, which makes a workflow easy to test and check.
- AI governance
- AI governance is the set of rules, owners and checks an organization uses to decide which AI uses are allowed, how their risks are handled, and who answers when something goes wrong. In the US, NIST publishes a voluntary framework for it, the AI Risk Management Framework.