Feynman Flow

UI of Feynman Flow

Inspiration

The problem most students run into while learning anything is that they don't practice enough. They are not solidifying their knowledge by testing enough. Even if they do they are not doing it in repeat so that they don't forget a lot of it. This forgetting happens in a curve gradually. Also even if they try to use the regular flashcard apps generating question is very time consuming . SIMPLY no app has all of these features that utilizes active recall, space repetition, learning in public, learning via teaching (Feynman Technique) or visually tracking progress altogether. I wanted to create an ultimate study tools using these main principles that has everything altogether.

Feynman Flow was inspired by the famous Feynman Technique — the idea that the best way to learn is to teach the concept in simple language to enforce repeated active testing in public. Using the Socratic Sidekick feature students can answer Gemini's curiosity driven questions and solidify conceptual understanding better. Most existing study tools focus on passive recall or static flashcards. We wanted something more interactive and conversational, where learners are guided to explain, revise, and refine knowledge in their own words.
This project aims to bring together flashcards, spaced repetition, and reflective explanation into a single smooth workflow.

What it does

Feynman Flow helps users:

Create decks and flashcards for active recall
Review cards based on spaced-repetition scheduling
Powered by Gemini I we use questions driven learning
Upload PDFs to auto-generate Q&A
Use a Socratic Sidekick that asks topic-driven questions
Auto-format user responses into structured notes
Store notes blog-style for future review

The goal is to make learning more like teaching — conversational, iterative, and public.

How we built it

Frontend:
- HTML, CSS, vanilla JavaScript
- Simple UI components for flashcards, Q-gen view, and Sidekick pane
- API routing configured via netlify.toml
Backend:
- Python + FastAPI
- SQLite + SQLModel for lightweight storage
- Endpoints for decks, cards, spaced repetition, PDF ingestion, and blog posts
- Socratic Sidekick workflow + Gemini API integration
Infra / Deployment:
- Backend → Railway
- Frontend → Netlify
- Redirect proxy for REST routes (/api/* → Railway)

The spaced-repetition logic is implemented using simple scheduling with: [ \text{nextDue} = \begin{cases} \text{now} + 1\text{ min} & \text{if reviewed as "again"} \ \text{now} + 10\text{ min} & \text{if "good"} \ \text{now} + 1\text{ day} & \text{if "easy"} \end{cases} ]

Challenges we ran into

Deploying Frontend + Backend separately and getting API routing correct
Proxying /api/* routes from Netlify → Railway
Handling PDF text extraction consistently across files
Keeping the Socratic workflow simple while still user-driven
Ensuring session state persisted during conversations
Debugging Gemini integration when providing contextual Q&A

Accomplishments that we're proud of

Built a full end-to-end learning tool in a short timeframe
Seamlessly integrated conversational generative AI with spaced-repetition
Designed an intuitive UI that makes reflective learning frictionless
Created Q-generation from PDFs that helps bootstrap study decks
Built a system that supports both private flashcards and public blog-style posts