TDD, AI agents and coding with Kent Beck

Software engineering practices are undergoing a significant transformation as AI coding agents provide unprecedented leverage for ambitious development projects. These agents, often described as unpredictable genies, require rigorous testing frameworks to manage their tendency to introduce errors or bypass design constraints. Test-Driven Development remains a crucial methodology for maintaining code quality and managing developer anxiety, as it allows for iterative learning and deferred commitment. Reflecting on the origins of the Agile Manifesto and Extreme Programming, the discussion emphasizes a shift from rigid, phased development toward feedback-driven, conversational processes. Furthermore, the evolution of engineering culture at Facebook demonstrates how high-scale observability, feature flags, and a strong sense of individual ownership can effectively replace traditional, heavy-handed testing models, enabling rapid innovation while maintaining system stability.