Reply to thread

Message: [QUOTE="DBaaS, post: 90436"] Vibe coding gets you to launch, then reality arrives. Scaling that mess demands actual human engineers. Version one practically wrote itself. Nobody warns you about month three, when a single feature quietly spawns more broken behavior than I can even tally anymore. My stack, for the record, involves Fable alongside Sol. I gave them everything: skills, routing instructions inside Claude.md, an entire regressions.md dedicated to past failures. They read all of it, congratulate themselves, then break identical things again, and cap it off by pasting twenty screenshots that supposedly prove everything works now. It never does. Paying a seasoned human developer stung a little, until he killed the entire thing in sixty minutes flat while I watched. Which raises the question: what are OpenAI and Anthropic actually selling at trillion-dollar valuations? Whatever it is, the coming correction will be brutal. [/QUOTE]