AI Test Automation Framework

Most automated tests break for a silly reason: a button moved a few pixels, even though the feature still works fine. That one small change can turn a whole test suite red and cost a developer hours of debugging. This framework fixes that: when a test can't find something on the page, it asks an AI model to look at the page and work out where that element moved to, then carries on the test instead of failing it.

It also writes new test cases on its own. Give it a plain-English description of a feature, like a user story from a ticket, and it generates the Selenium and TestNG code to test it, saving the manual setup that usually eats up a QA engineer's day.

In a verified run, it healed a broken locator with 98% confidence and all three tests passed, both on a local machine and inside GitHub Actions, our continuous integration pipeline.