Before you start
- Know what the agent was asked to do and what it is allowed to change.
- Open the run history or approval queue.
- Review outputs against the original input and source context.
Steps


Step 3
Review the output
Read extracted fields, proposed records, generated copy, or actions before approval.

Step 4
Approve, reject, or edit
Approve only useful outputs. Reject or edit results that are incomplete, irrelevant, duplicated, or unsafe.

What happens next
FAQ
Why is output pending approval?
The workflow is configured to require review before changes are applied.
Can I bulk approve agent output?
Only do this after sampling enough results to trust the run. Sensitive actions should stay reviewed.
Let agents research, enrich, or draft work, then review outputs before they affect your workspace.
Learn agents
See it in the productExplore AI Agents