---
title: "Supervised autonomy: definition | Rex glossary"
description: "Giving an AI agent independence in stages: it starts by proposing actions for approval and runs on its own only as its work proves reliable."
canonical: https://www.rex.inc/glossary/supervised-autonomy
---

# Supervised autonomy

Giving an AI agent independence in stages: it starts by proposing actions for approval and runs on its own only as its work proves reliable.

A new procedure might begin with a person reviewing every action. As accuracy is confirmed, routine cases run automatically, while sensitive actions, such as write-offs or credit decisions, stay gated for approval.

This lets a team adopt automation without giving up control, and move an agent back to supervised mode if its results slip.

## Further reading

- [Rex achieves SOC 2 Type II→](https://www.rex.inc/blog/rex-achieves-soc-2-type-ii)

## Related terms

- [Human in the loop A design in which an automated system hands defined decisions to a person to review or approve before it acts.](https://www.rex.inc/resources/ai-ar/human-in-the-loop-ar-automation)
- [Agent operating procedure A written procedure that tells an AI agent how to handle a type of work, encoded from the standard operating procedures a finance team already follows.](https://www.rex.inc/glossary/agent-operating-procedure)
- [Audit trail A chronological record of every action taken on an account, who or which agent took it, and why, so decisions can be reviewed later.](https://www.rex.inc/resources/ai-ar/ai-ar-audit-trail-explainability)

Next step

Accelerate your receivables operation with Rex

A short walkthrough of agents working your accounts under your guardrails, with your team on the exceptions.

[Learn more](https://www.rex.inc/learn-more)

[Back to the glossary](https://www.rex.inc/glossary)