Member-only story
Our Senior Engineer Refused to Become a 10× Engineer With AI. Three Months Later, I Understood Why.
Everyone else was using AI to produce more code and finish more tickets, while he was using the same tools to decide what we should stop building.
Three months after we rolled out AI coding tools across the engineering organization, one of our senior engineers seemed to be getting less value from them than almost everyone around him, even though he was one of the strongest engineers on the team and was already using agents for debugging, testing, repository exploration, SQL analysis, documentation, and repetitive implementation. Other engineers were producing noticeably more code, finishing tickets faster, and turning ideas into prototypes within hours, while his visible output had increased only modestly despite having access to exactly the same tools.
The strange part was that he was not slower when he actually decided to build something, because AI had made him considerably faster at implementation just like everyone else. He was simply spending much more of the time AI saved him questioning requirements, testing whether proposed systems were necessary, and occasionally reaching the conclusion that the best technical solution was not another faster…
