An AI-powered document notebook built with Next.js, LangChain, and Prisma. Upload documents, add sources, and chat with them using Retrieval-Augmented Generation (RAG).

• 📂 Multi-source ingestion – Upload PDF, TXT, CSV, or provide URLs for web crawling & indexing.
• 🔍 Smart retrieval – Embeddings-based semantic search across all or selected sources.
• 💬 Interactive chat – Ask questions and get contextual answers, citations, and suggested follow-ups.
• ⚡ Streaming responses – Real-time AI responses with typing indicators.
• 📑 Citations & references – Clickable citations for source traceability.
• 🛠 Modern stack – Next.js 15, Prisma + PostgreSQL (pgvector), LangChain, React Query, Tailwind.

         ┌───────────────┐
         │   Frontend    │
         │ (Next.js App) │
         └───────┬───────┘
                 │
     ┌───────────▼───────────┐
     │     API Layer (app/)   │
     │  - /api/session        │
     │  - /api/sources        │
     │  - /api/chats          │
     │  - /api/query          │
     └───────────┬───────────┘
                 │
     ┌───────────▼───────────┐
     │     Backend Logic      │
     │  - Ingestion (PDF/CSV/URL)
     │  - Embeddings (pgvector)  
     │  - LLM (Groq/Google GenAI)
     │  - Query Processing + RAG  
     └───────────┬───────────┘
                 │
         ┌───────▼────────┐
         │  Database      │
         │ (Postgres +    │
         │   Prisma ORM)  │
         └────────────────┘

• Frontend (React/Next.js) – Sidebar for sources, chat panel, streaming responses.
• APIs – REST endpoints for session mgmt, source ingestion, chat logs, and querying.
• Vector DB (pgvector) – Stores embeddings for semantic search.
• LLM Layer – Uses LangChain + Groq/Google APIs for query answering.

• Sidebar.tsx – Manages sources (select all, add new, toggle filters).
• ChatPanel.tsx – Displays messages, citations, follow-ups, and handles streaming queries.
• AddSourceModal.tsx – Handles uploads (PDF, TXT, CSV) and web URLs.
• useNotebook.ts (hooks) – React Query hooks for fetching chats, sessions, and sources.
• lib/ingest.ts – Parsing & chunking of PDF/CSV/TXT, embeddings creation.
• lib/llm.ts – LLM client (Groq, Google GenAI).
• lib/responseProcessor.ts – Enhances answers, formats citations, generates follow-ups.

• Session – Represents a notebook.
• Source – Uploaded file or URL.
• SourceChunk – Chunked embeddings.
• Chat – Stores user + assistant messages with citations.

• 👤 Personal Accounts – Add authentication & multi-user support.
• 🗂 User-owned Notebooks – Allow multiple notebooks per user (e.g., "Work", "Research").
• 📝 Custom prompts – Users define their own query strategies.
• ☁️ Cloud Sync – Store notebooks securely across devices.

# Install dependencies
npm install

# Setup env variables
DATABASE_URL = ""
DIRECT_URL = ""
GEMINI_API_KEY=""
GROQ_API_KEY = ""

# Setup database
npx prisma migrate dev --name init

# Run dev server
npm run dev

App will be live at https://notebook-lm-r86c.vercel.app/.