Descriptions:
The Pragmatic Engineer hosts a conversation with software engineering legends Martin Fowler and Kent Beck, marking 25 years since the Agile Manifesto, and exploring what AI means for a field they’ve spent decades shaping. The discussion opens with a striking observation: test-driven development — long controversial and still divisive — may be experiencing a renaissance in the AI era. As Beck notes, a practitioner recently told him that 25 years of TDD culture is proving critical now that developers must verify AI-generated code rigorously. When you’re working with a powerful but unpredictable system, the discipline to confirm correct behavior matters more than ever.
Both Fowler and Beck share candid, firsthand accounts of using AI coding tools in practice. Fowler describes his current work on martinfowler.com as an effort to document what practitioners are actually doing — the decisions humans are still making, what reviewing AI output actually entails, and how the conversation with AI tools unfolds in real workflows. Beck admits to initially dismissing AI based on poor early experiences with GitHub Copilot, but credits colleagues like Simon Willis and Thoughtworks peers Mike Mason and Birgitta Böckeler for demonstrating that Claude Code and Gemini can deliver genuine value when used skillfully.
A key insight is that AI dramatically compresses time-to-prototype — from three weeks to three hours — but relocates the bottleneck to design thinking. Programmers who previously had time to develop ideas during long implementation cycles now face a different constraint: interesting design ideas don’t arrive faster just because code does. The conversation frames AI not as a one-time disruption but as a continuous forcing function that keeps reinventing what good software engineering looks like.
📺 Source: The Pragmatic Engineer · Published April 07, 2026
🏷️ Format: Interview
