I prefer the article's own "The Prototype Isn't the Product" title, since that seems to better capture what's being discussed. With AI, it can be ridiculously

I prefer the article's own "The Prototype Isn't the Product" title, since that seems to better capture what's being discussed. With AI, it can be ridiculously quick to vibe code throwaway prototypes and personal use apps where "seems to mostly work" is the quality criteria, but what remains harder and slower is software engineering - building things where quality and maintainability are critical. There is much less discussion of this - we mostly see people wanting to show off some flashy demo they just vibe coded, not discussion of how useful AI was, or wasn't, in working on some large legacy codebase or whatever non-sexy non-trivial task their day-to-day professional work consists of. You are not going to get a 6am sunday call at home because your vibe-coded 3-D game prototype stopped working, but you surely will if there's a bug in the 24x7 production system that was just updated.