Canonical: https://www.strataigize.com/glossary/human-in-the-loop/
Description: Human-in-the-loop is an automation design in which defined actions, typically anything irreversible, wait for a person’s approval before they run.

[Home](https://www.strataigize.com/) / [Glossary](https://www.strataigize.com/glossary/) / Human-in-the-loop

AI Automation

# Human-in-the-loop.

Definition

## What it _means._

Human-in-the-loop is an automation design in which defined actions wait for a person’s approval before executing, typically anything irreversible: spending money, messaging a client, writing to a production system. The automation does the mechanical work; accountability stays with a named human at the moments that matter.

Worked example

## In _practice._

A reporting agent drafts client updates every week. Drafts wait in an approval surface where a person reviews and releases them. The team verifies that sending requires approval, tests failure paths, and checks release logs. Those controls and their execution support the review requirement; the design alone is not proof that every message was reviewed.

Where this work happens: [AI Workflow Automation](https://www.strataigize.com/services/ai-automation-services/)

Related terms

-   [AI agent](https://www.strataigize.com/glossary/ai-agent/)
-   [AI governance](https://www.strataigize.com/glossary/ai-governance/)

## Put it in _context._

Bring your growth question to a free 30-minute consultation. We can work through what the numbers mean and which changes to consider.

[Book a growth consult →](https://www.strataigize.com/audit/)