WeeNote (위노트)

01The full story

Min Sang-gi taught in Korean elementary schools for about fourteen years, then spent roughly a year working as a developer. When he saw what AI was becoming he quit and started building. He now runs two products; WeeNote is the one that worked. The idea arrived by accident. On Threads, a school counselling teacher posted that she was drowning in a specific administrative problem, had paid an Excel contractor to build her something, couldn't use what she got, and was asking for help. He messaged her, got on a video call, and recognised two things at once: the solution was not technically hard, and it would help someone immediately. He spent about three days without sleep and roughly $400 on Replit — which handled the database and deploys — and produced a prototype. That first version failed, and the reason is the most useful part of the story. His one user turned out to be the top 1% of her profession: she had spent her own money on custom tooling and knew how to brief a developer. Every feature she asked for was real, but the average counsellor never uses those features. He had built for a unicorn. Worse, when she demonstrated it to colleagues, the reaction was flat — not over features, but over the cloud. Counselling logs contain material so sensitive that some of it is deliberately never written down at all. He explained client-side encryption and keys the server can never decrypt; it didn't matter. The objection was that the log leaves the building over a network. A previous government-led program had died for the same reason. He interviewed more than twenty users. Seventy to eighty percent said the same thing: good product, wrong architecture. So he rebuilt version two as a local desktop app. Payments started arriving on launch day.