Martin Fowler's link roundup covers several AI-in-software-development topics. Chris Parsons' updated guide on AI-assisted coding emphasizes keeping changes small, building guardrails, and shifting focus from 'how fast can we build' to 'how fast can we verify.' Birgitta Böckeler's harness engineering article and video discuss using computational sensors like static analysis and tests to constrain AI agents. Adam Tornhill revisits function length in the context of agentic programming, arguing that well-named functions defining clear concepts matter even more when AI infers meaning from code structure. Nilay Patel's 'software brain' concept explains why many people distrust AI — the tendency to reduce everything to queryable databases. Fowler also reflects on the importance of precise data definitions for AI effectiveness, and shares Ezra Klein's observations about Silicon Valley's race to make themselves 'legible' to AI, while personally resisting the temptation to outsource his writing to LLMs.
Sort: