i run a small insurance firm, 20 years in that domain, zero in software, and i recently vibe-coded my way into running agent infrastructure for the business, so i should stall at 80%. but I think that the judgment my project actually needed was mostly regarding insurance itself, and making sure the knowledge base it pulls from was sound, so I was able to use my experience on that side to produce what I hope is a useful product
the technical side is of course different and i don't have that judgment, so i try to get by with fresh-session reviews, testing, and keeping it on the simpler side because i know i can't audit the code myself. whether or not that proves to be a winning formula remains to be seen, but either way i think it's reasonable to say that some regular folk will stall at the 80% but others may figure out a way to power through it using caution and reason