Recallr

We’re solving a billion-dollar inefficiency in how people study, the Duolingo of active recall. We built the best solution on the market in 1 month. In 12 months, the sky is the limit.

Comment

Inspiration

Most flashcard apps are severely outdated. They lack AI integration (even though users already use them on the side!), have clunky interfaces, are expensive, make it hard to find or create quality decks, and are not optimizing learning process properly. They haven't evolved in years. We saw a clear opportunity to rebuild flashcards from the ground up using modern technology, design and knowledge. Our goal was simple: build the best active recall tool in the world.

What it does

Recallr is an AI-powered flashcard app designed for high-efficiency learning. It uses a user-level personalized, ML-based spaced repetition algorithm to help users retain knowledge faster. It works fully online and offline, and syncs across web, iOS, and Android. Users can generate cards instantly from text, PDFs, or entire exams using AI. It supports decks from other flashcard apps, offers detailed analytics, and even uses voice playback so that users can study while on a walk. Everything is built for speed, simplicity, and real learning.

How we built it

  • Our repositories are planned around being able to: develop locally, develop on Bolt, support dev and prod environments with no changes
  • RevenueCat to support a freemium model (and later... a marketplace)
  • Lingo to handle multiple languages
  • Frontend was built on Bolt using MUI design, so everything is standard and gorgeous out of the box and deployed via Netlify
  • Backend was built with Django (unlike many projects, we opted away from supabase), and deployed on Render
  • Already fully PWA compliant!

Challenges we ran into

  • Personalizing the spaced repetition algorithm without making it too complex, and available offline, and obfuscated.
  • Keeping the interface and repository clean and organized, while adding features.
  • Architecting the backend/frontend so it all works locally, or on Bolt, or on final environments.
  • Integrating Google Login with authentication methods and different domains; handling authentication tokens and cookies (since we have cross domain!)

Accomplishments that we're proud of

  • Built a full product in one month.
  • Made a better flashcard algorithm than anything currently on the market.
  • Integrated AI in a way that saves users time.
  • Kept the design simple. We opted away from the common dashboard design for a reason, and it paid off!
  • Planned a clear path to monetization and growth.

What we learned

  • Authentication is a pain, and must be done properly and carefully
  • Good design makes complex tools easier to use.
  • Building fast is possible but requires lots of planning.
  • In such a short timeframe, clear focus and priorization is the only way to have a good product.

What's next for Recallr

  1. Release AI-generated decks for major exams in the U.S., Canada, India, China, and Brazil.
  2. Start selling to schools and universities with a white-label version.
  3. Launch a marketplace so users can buy and sell decks themselves.
  4. Improve the AI with smarter reviews and better learning paths.
  5. Raise capital to grow faster.

Built With

Share this project:

Updates