“In an evening” sounds like marketing, but if you narrow the task down to one understandable function, it is quite possible. Below is an honest analysis of one such evening: without editing and without “and then a miracle happened.”
Where does speed begin?
Speed comes not from the fact that the model writes code, but from the fact that you don’t waste time on an empty start. A ready-made template, a clear task, a predetermined question “where will it live” - this is what saves the first hour.
- One function, not a “platform”.
- Data and deployment are thought out before the code, not after.
- You check the result immediately, in small pieces.
Where are you wasting your time?
The main trap is to try to expand the idea on the fly. Each “let’s do it again” is worth not five minutes, but the entire evening. The second is to blindly trust what is generated without running it: what looks right does not mean it works.
What's in the bottom line
By the end of the evening, the result is not a “product”, but an honest one prototype, which shows whether the idea is worth continuing. And this is exactly what an MVP is needed for: not to amaze, but to quickly test. Vibe-coding does not replace thinking here - it removes the friction between the idea and the first working version.