https://apps.apple.com/se/app/reelly-scroll-learn/id6737239036

This has been a side project for about a year, have thought about it ever since ChatGPT came out :)

Mark your interests --> get a feed of short articles tailored for that

Free text search: Search on e.g. "Explorers" --> The app crawls Wikipedia and pulls together a short feed (e.g. Columbus, Vasco da Gama etc.,)

I rewrite the Wikipedia articles with GPT-4o, and store them in a database. When a user searches for something not in the database, I use various models to classify what the user is after (how wide the topic is) and pull together a nice selection of articles and images based on the user input. Hitting the right balance of generation speed with article quality was the most challenging part.

Would really appreciate some feedback, especially if you tend to browse Wikipedia a lot, plus if Simo Häyhä rings any bells (a lot of WW1 + 2 articles at the moment)