TravelLine, a hospitality automation platform, conducted an audit of its 6,500 end-to-end (E2E) tests to prepare for a migration from Java to .NET. The company utilized an AI agent to analyze the test base, aiming to eliminate technical debt and redundant UI checks.
The team configured the agent within OpenCode, using GPT-5.5 and the Model Context Protocol (MCP) to integrate with TestIT. The agent was tasked with classifying tests into three categories: migration to .NET, shifting to integration/API tests, or removal due to obsolescence.
During the first audit phase, 2,600 tests were analyzed. For instance, in the 'Transfers' section, 15 out of 18 E2E tests were identified as redundant UI-based duplicates of existing integration tests and were marked for removal.
In the second phase, the team prioritized tests based on failure rates and execution duration (P85 metric) for the entire 6,500-test database. This approach helped identify 9 critical tests in the 'Reports' section that were both slow and prone to failure.
The entire audit process took approximately two weeks. As a result, 450 outdated screenshot tests were removed, and the team established a clear roadmap for migrating the remaining tests to the new framework.
