The author describes their experience migrating a large codebase of frontend tests from React Testing Library v13 to v14, involving 970 test files and over 6,000 test cases. Initial attempts to automate the migration using Claude Code CLI led to overwhelming test failures. Realizing the complexity of the update—where APIs became asynchronous and timing behaviors changed—the author pivoted to creating a detailed migration guide with Claude's help. This guide highlighted extensive changes required, revealing that the migration would demand significant manual debugging and adjustments. To manage the transition, the author set up a system to run both versions of the library simultaneously. They created a codemod using jscodeshift to automate parts of the migration process. Iteratively, they tested the codemod on small batches of tests, refining it after identifying edge cases and coverage issues. The migration guide expanded from 4,532 to 7,517 words as they documented their findings and improvements. Despite the advantages of AI assistance, including the ability to debug and fix tests, the author encountered limitations, particularly with maintaining code coverage and managing context during longer tasks. The entire migration took a week and involved 50 pull requests, with each one taking about half an hour. The author emphasized the importance of verifying results and maintaining a vigilant approach instead of relying solely on AI solutions. They acknowledged that while AI tools significantly reduced the time spent on repetitive tasks, traditional software development principles still apply. Overall, the project underscored the tension between automation and the need for developers to stay actively engaged in the process.