Phase 3 of my VibeTDD experiment: What happens when I take control of the TDD process and use Claude as an implementation assistant? The results were surprisingly challenging.
After systematically testing AI-led development, human-led TDD, and the test-after trap, I've discovered the real boundaries and opportunities of AI-assisted coding. Here's what actually works and what definitely doesn't.
Before testing human-led TDD, I explored the popular 'AI generates tests for existing code' approach. The results were more concerning than I expected.
Phase 2 of my VibeTDD experiment: What happens when Claude tackles a real-world coding challenge with complex business rules? Spoiler: the results were eye-opening.