---
title: AI auto-fix (beta)
description: "Automatically repair broken tests with the power of AI.."
icon: "bandage"
---
## Overview
Maintaining a test suite can often take more time than building it.
**AI auto-fix** helps you reclaim that time by automatically identifying and fixing broken tests. The feature detects failed test steps, analyzes the root cause, and proposes context-aware fixes, so your tests stay up-to-date with minimal manual effort.
## How it works
When a test fails, the AI agent analyzes:
- Changes in the application’s UI (e.g., updated element locators or structure)
- Differences between previous passing runs and the latest failure
- Patterns in past fixes to suggest the most reliable correction
If a fix is found, the test is labeled 'auto-fix ready to review' in your latest test report.
You can then review, compare, and approve the fix before it’s applied.
---
## Reviewing and Applying an auto-fix
Follow these steps to review and apply an AI-generated fix:
### Open the latest test report
- Go to Test Reports and open your most recent run
- Failed tests will be listed
- If a test shows the 'auto-fix ready for review' label, a suggested fix is available
### Review the suggested fix
- Open the failed test
- Use the timeline and agent summary to understand what went wrong
- Click 'Review Fix' to open the comparison panel
### Compare test steps
The panel displays two columns:
- Left: Steps from the failed run
- Right: Proposed steps from the AI fix
- ⚪️ Gray = Unaffected step
- 🔴 Red = Outdated or invalid step
- 🟢 Green = Verified step already tested by the AI — applying it should make your test pass again.
### Preview details
Click the chevron icon next to any step to view a detailed preview, including updated locators and actions.
### Approve or decline the fix
- Click 'approve auto-fix' to apply the AI’s changes and automatically update the test.
- Click 'decline auto-fix' if you want to keep the test as is — no changes will be made.