Paperflow

Paperflow is an AI-powered workflow for scanning, organizing, and searching paper notes.

Check it out live: https://paperflowapp.vercel.app

Running locally

Frontend

1. Install dependencies

cd client
npm install

2. Create a .env.local file inside the client directory. You will need to make an account on Clerk to get an API key.

NEXT_PUBLIC_API_URL=http://localhost:8000
NEXT_PUBLIC_CLERK_PUBLISHABLE_KEY=...

3. Run in development mode

npm run dev

Backend

1. Create a virtual environment + install dependencies

python3 -m venv venv
source venv/bin/activate
pip install -r requirements.txt

2. Create a .env file inside the server directory. You will need to create API keys for Gemini, OpenAI, Pinecone, and Supabase.

CLERK_ISSUER=
CLERK_JWKS_ENDPOINT=
CLERK_SECRET_KEY=
SQLALCHEMY_DATABASE_URL=
GOOGLE_API_KEY=
GEMINI_API_KEY=
OPENAI_API_KEY=
PINECONE_API_KEY=
PINECONE_ENVIRONMENT=
SUPABASE_URL=
SUPABASE_KEY=
SUPABASE_BUCKET=
FRONTEND_URL=http://localhost:3000

3. Run server

uvicorn main:app

Tech Stack

The frontend is built with Next.js, with a FastAPI backend for handling core logic and OpenCV for image processing. The RAG pipeline is built with OpenAI embeddings, Pinecone, and Gemini. The database is PostgreSQL, storing images in Supabase S3.