A daily sports-contest product — accounts, contests, subscriptions, payouts — was written entirely by AI and directed by a single engineer. The Laravel API, the web front end, and the mobile app shipped together. The mobile app was released on the App Store in about four to five months.
Nothing was broken, because nothing existed yet. A product like this — sign-up, contests and markets, a leaderboard, subscriptions, payouts, notifications, support — is normally a human workflow from the first ticket. Specialists write the API. Other specialists write the web UI. Another group writes the mobile app. A store release sits at the end of all three calendars.
That default is the thing this project replaced:
One engineer directed it. AI wrote the application: the backend, the web front end, and the mobile app. The result is a daily contest product a member can join, play, subscribe to, and get paid from — including the app that went to the App Store.
This was an AI workflow from the start. The engineer’s job was the product: what to build, in what order, and whether it was actually the app a member would use. The application code — backend, web, and mobile — was produced by AI.
What that covered:
It sits beside a different kind of proof. A five-year sports platform, already in market on a Laravel API and mobile apps, made the opposite move in a similar window: a large human operation turned over to an agent model and a knowledge graph, inside the product customers already used. Read that story.
The outcome is the scope that made it to the store, and how few people it took to get there. The calendar is the same length we quote for converting a live product.
Four to five months is the window the build ran in. “100%” is how the application was produced: AI wrote the code, one engineer directed it, across the API, the web front end, and the mobile app.
API, web, and mobile, directed by one person and written by AI. Fifteen minutes is enough to see whether the thing you want fits that shape.