duobrain is your personal "Second Brain"—a central hub to store, organize, and retrieve everything that matters. Whether it's a deep-dive YouTube video, a viral tweet, a random thought, or an essential document, duobrain keeps it all in one place.
Unlike a simple bookmarking app, duobrain uses AI-powered vector embeddings to help you find your content based on meaning, not just keywords.
- Multi-Format Support: Native embedding for YouTube, X (Twitter), Instagram, Pinterest, Documents, and Web Links.
- AI Search: Integrated with Hugging Face models to provide semantic search—find that "video about pasta" even if the title doesn't contain the word "pasta."
- Brain Sharing: Generate a unique link to share your entire collection or just a specific selection of items.
- Knowledge Merging: One-click import to merge a friend's shared brain into your own.
- Smart Filtering: Instantly toggle between content types using the sleek sidebar navigation.
- Secure Auth: OTP-based email verification and JWT-protected sessions.
| Component | Technology |
|---|---|
| Frontend | Next.js 15 (App Router), TypeScript, Tailwind CSS |
| Backend | Node.js, Express, TypeScript, Zod |
| Database | MongoDB (Atlas), Mongoose |
| AI/ML | Hugging Face Inference (all-MiniLM-L6-v2 for embeddings) |
| Mailing | Nodemailer (Gmail SMTP) |
- Ingestion: When you add a link or text, the backend sends the content to an embedding model.
- Vector Storage: The resulting high-dimensional vector is stored alongside your content in MongoDB.
- Retrieval: Searching triggers a vector search (
$vectorSearch), comparing your query's "meaning" against your stored data. - UI: A responsive dashboard built with Next.js renders the content using social media embed providers for a rich preview experience.
cd server
npm install
# Create a .env or modify config.ts with your:
# MONGO_URL, JWT_SECRET, HF_TOKEN, GMAIL_PASSWORD
npm run build
npm start
cd client
pnpm install
pnpm run dev
Open http://localhost:3000/duobrain to start building your brain.
client/app/: Next.js pages and AI search logic.client/app/_components/ui/: Modular UI components likeCard,SearchBar, andModal.server/src/index.ts: Main Express entry point with CRUD and Search routes.server/src/generate_embeddings.ts: The bridge to Hugging Face for semantic analysis.server/src/model.ts: MongoDB schemas for Users, Content, and Cache.