Every run teaches the next one
When the agent works out how to do something in your app, like creating an order or opening a settings page, it saves that as a reusable step. The next run starts from everything earlier runs learned. Tests get faster, cheaper and more reliable over time.
From 138 to 1,008 reusable steps in ten weeks
Real numbers from testing GreatLab’s app. 789 steps drive the screens, 219 set up data through the API.
138
1 August 2026
492
1 September 2026
734
1 October 2026
1,008
10 October 2026
Simple rules, compounding results
API to set up, screens to test
Data is created through the API because it is fast and reliable. What a user would do is tested through the screens.
Fixed in place
When your app’s screens change and a step stops working, the agent fixes the step itself, so every future run benefits.
Shared with your regression suite
The same steps are used by the pull request runs and the regression suite, so improving one improves both.
Fire your QA backlog,
not your team
Let agents test every pull request, keep your regression suites green and send your team clear reports, videos and bug tickets.