Martin Fowler and Kent Beck shared a stage at the Pragmatic Summit in San Francisco to discuss AI's impact on software development. Between them, they authored four of the field's most influential books: 'Extreme Programming Explained', 'Test-Driven Development', 'Refactoring', and 'Patterns of Enterprise Application Architecture'. Their verdict on AI is not optimistic by default: the adoption curve is faster than any prior shift, including the microprocessor, object-oriented languages, and the internet, and the industry is not ready for it.
The full conversation covers eight distinct topics, and the most pointed ones are not the obvious ones. Beck and Fowler draw a direct parallel between the Agile era and today: misaligned company incentives, predatory vendors selling transformation, and a cohort of mid-level developers who resisted change and paid for it with their careers. That pattern is repeating now. Also worth reading: their specific warnings about large codebases, where AI tools underperform compared to greenfield projects, and the emerging trend they call 're-soloing', where collaborative development reverts to individual work mediated by agents. Fowler also offers a concrete personal signal for burnout: when you notice you are producing negative value, stop.
The piece is worth reading in full for what Beck and Fowler say about metrics and quality. Companies are already measuring pull request frequency instead of outcomes, and the speed pressure from AI is causing deliberate quality trade-offs. Beck, who invented TDD, argues tests are now more critical than ever, not less, precisely because agents generate code that needs verification. The full recording is available on YouTube.
[READ ORIGINAL →]