AI-powered document Q&A with decentralized storage on Filecoin
Track: Fresh Code | Bounty: Filecoin + AI Integration
FileMind AI lets users upload any PDF or text document, stores it permanently on Filecoin via IPFS (web3.storage), and uses LangChain + OpenAI GPT to answer any questions about the document.
Problem it solves: Documents stored on centralized servers can be lost, censored, or tampered with. FileMind combines the permanence of decentralized storage with the power of AI to make any document permanently queryable.
| Layer | Technology |
|---|---|
| Decentralized Storage | Filecoin via web3.storage |
| AI / LLM | OpenAI GPT-3.5-turbo |
| RAG Framework | LangChain + FAISS vector store |
| Backend | Python FastAPI |
| Frontend | React (Vite) |
cd backend
pip install -r requirements.txt
export OPENAI_API_KEY=your_key_here
export WEB3_STORAGE_TOKEN=your_token_here
uvicorn main:app --reloadcd frontend
npm install
npm run dev- User uploads a PDF or TXT file
- File is uploaded to Filecoin via web3.storage → gets a permanent CID/IPFS URL
- File is split into chunks and embedded using OpenAI Embeddings
- Vectors stored in FAISS for fast similarity search
- User asks a question → LangChain retrieves relevant chunks → GPT answers
Every uploaded document is stored on Filecoin via web3.storage's API. The returned CID is a permanent, decentralized address for the file — no central server needed.
Upload any research paper, contract, or report and ask:
- "What are the key findings?"
- "Summarize section 3"
- "What are the risks mentioned?"