Prominent developer says coding agents make software engineering harder
Simon Willison, a heavy user of these tools, gives a contrarian judgment; the discipline and knowledge bar for adopting agents is worth comparing against.
ImportanceLocalEvidenceE2 unreplicatedWrite-upQuick
Simon Willison says the more he works with coding agents, the more convinced he is that they make software engineering harder.
Willison is a prominent developer who has long tested and written about LLM tools. In a two-sentence note on his weblog on 24 September, he wrote that these tools can do amazing things, but unlocking their full potential requires "extraordinary discipline and knowledge."
The note offers no argument or data beyond the personal judgment. It runs counter to recent public declarations by other developers of going all-in on AI coding, and reads best as a counterpoint.