Feynomenon AI

Inspiration

We were inspired by the Feynman Technique, a proven learning method where teaching and simplification are used to master complex topics. We realized that AI could replicate this technique — not just to explain, but to identify knowledge gaps and adaptively quiz users — turning mistakes into learning opportunities. We wanted to make that experience fun, interactive, and personalized for everyone, regardless of educational background.

What it does

Feynomenon is an AI-powered tutor that helps you deeply understand any concept using the Feynman Technique. Here's how it works:

• You choose a topic
• The AI explains it simply
• Then, it quizzes you with increasingly challenging questions If you make a mistake, the AI logs it and later reinjects it as a new, analogous question This approach reinforces weak areas and accelerates true comprehension, not just rote memorization.

How we built it

We used the following tools and frameworks:

• Next.js with Tailwind CSS for the frontend
• Node.js/Express backend for handling API requests
• MongoDB Atlas to store user sessions, questions, and errors
• Gemini AI (or OpenAI API as fallback) to generate explanations, questions, and evaluate answers
• Firebase Auth for secure user login
• Custom prompt engineering to simulate the Feynman learning cycle

Challenges we ran into

1. Designing prompt chains that correctly escalated question difficulty and contextualized feedback
2. Implementing real-time feedback loops from user mistakes to generate targeted questions
3. Balancing scope: we had to streamline UI/UX while retaining core Feynman logic
4. Avoiding vague or overly complex LLM-generated questions\

Accomplishments that we're proud of

1. A working prototype that fully supports personalized sessions, adaptive testing, and mistake reinforcement
2. Seamless UI that feels polished and intuitive
3. Scalable architecture — backend and database are ready for broader deployment
4. Strong alignment with SDG 4 (Quality Education) by making learning accessible, personalized, and fun

What we learned

1. How to engineer prompts to guide large language models with precision
2. Structuring learning experiences around educational theory (Feynman Technique) enhances both usability and impact
3. Collaborative problem-solving under time pressure is powerful when everyone brings something to the table

What's next for Feynomenon AI

1. Add voice interaction and speech-to-text for accessibility
2. Integrate progress tracking dashboards with visual analytics
3. Expand to support multiple languages and offline access
4. Partner with educators to create curriculum-aligned topic banks
5. Turn this into a full-fledged learning platform targeting underserved communities

Built With

• claude
• css
• gcloud
• gemini
• javascript
• python
• react
• tailwind