Overview
CardsGPT started as a weekend coding project between me and my dear friend and researcher Antonio Terpin. We’re both passionate about building products, and we periodically try to do it as a creative exercise and to learn something new.
The problem we wanted to solve was straightforward: Antonio is a researcher at ETH working on his Ph.D., which has required him to learn completely new topics tangential to his usual focus (computer science, deep learning, and robotics). In his day-to-day work, he frequently found himself using language models like GPT to discover new concepts.
However, GPT alone isn’t enough to build something from new concepts. You need a study method to organize the knowledge gained. The language model can’t just be an assistant - you need tools to structure and retain the knowledge learned.
That’s why we created CardsGPT, a Chrome extension that adds a button to the ChatGPT interface that extracts the chat you’re viewing to create flashcards.
Through the mobile app, available for iOS, you can:
- Review your flashcards
- Memorize what you asked the model using spaced repetition
- Track your learning progress
While the project was originally designed for a single user (Antonio), we realized it could be a valuable product for many others, so we decided to release it publicly.
Tech Stack
The tech stack is simple but effective:
- React Native for the mobile app
- NextJS for the backend infrastructure
- OpenAI API for flashcard generation
- Vanilla JS/HTML/CSS for the Chrome extension
