Writing code is cheap now
Summary
Simon Willison argues that advancements in agentic engineering make code creation cheaper, but delivering high-quality, maintainable code remains costly. The post emphasizes evaluating design decisions against the new reality of parallel AI agents, stressing that good code—tested, well-documented, robust, and future-proof—still requires discipline, not automation alone. It also advocates developing new habits, such as asynchronous agent workflows and proactive testing approaches like red/green TDD and adherence to YAGNI.