How Sơn shipped a food diary, solo
What 30 days of building Perfeat alone taught him about scope and taste.
Sơn gave himself a month to put Perfeat in real hands. One person, no team, no funding round to justify the timeline — just a deadline he set so the thing would actually exist.
Scope is the whole game
The first decision was what not to build. He listed everything a food diary could be, then cut until only one loop remained: log a meal, rate it, find it again later. Each cut made the product clearer, not poorer.
Taste is the moat
With AI doing a lot of the typing, the scarce thing wasn’t code. It was judgment — knowing which version of the meal-logging flow felt right, and being willing to throw away three that didn’t. That part doesn’t delegate.
The diary that shipped is smaller than the one he first imagined, and better for it.