🦙 Sandbox-NotebookLM - Next.js Edition

Enterprise-ready alternative to Google NotebookLM, powered by LlamaIndex and multiple AI models.

Built with: Next.js 15, React 19, FastAPI, PostgreSQL, LlamaCloud

Repository: https://bitbucket.org/zlalani/sandbox-notebookllamalm-nextjs

---

🚀 Quick Start

Prerequisites

• Node.js 18+ and npm
• Python 3.10+
• Docker and Docker Compose
• PostgreSQL (via Docker)

Installation

1. Clone the repository:

git clone git@bitbucket.org:zlalani/sandbox-notebookllamalm-nextjs.git
cd sandbox-notebookllamalm-nextjs

2. Start Docker services:

docker-compose up -d

This starts:

• PostgreSQL (port 5433)
• Redis (port 6379)
• Jaeger (port 17000)
• Adminer (port 9001)

3. Set up Backend:

cd backend

# Create .env file with your API keys
cat > .env << 'EOF'
OPENAI_API_KEY=your_openai_key
LLAMACLOUD_API_KEY=your_llamacloud_key
ANTHROPIC_API_KEY=your_anthropic_key
GOOGLE_API_KEY=your_google_key
GROQ_API_KEY=your_groq_key
DEEPSEEK_API_KEY=your_deepseek_key
ELEVENLABS_API_KEY=your_elevenlabs_key

pgql_user=postgres
pgql_psw=admin
pgql_db=postgres
EOF

# Install dependencies with uv
pip install uv
uv sync

# Initialize database
uv run python -c "import sys; sys.path.insert(0, 'src/notebookllama'); from database import Base, engine; Base.metadata.create_all(bind=engine); print('✓ Database initialized')"

# Start backend
uv run uvicorn src.api.main:app --host 0.0.0.0 --port 9000 --reload

4. Set up Frontend:

cd ../frontend

# Install dependencies
npm install

# Start frontend
npm run dev

Access Points

• Frontend: http://localhost:4000
• Backend API: http://localhost:9000
• API Docs: http://localhost:9000/docs
• Database UI (Adminer): http://localhost:9001
  • System: PostgreSQL
  • Server: postgres
  • Username: postgres
  • Password: admin
  • Database: postgres
• Jaeger Tracing: http://localhost:17000

---

📚 Features

Core Features

• ✅ User Authentication - Signup, login, protected routes
• ✅ Multi-Notebook Management - Create unlimited notebooks
• ✅ 6 AI Models - GPT-5, Claude 4.5, Gemini 2.5 Pro, GPT-4o, Gemini Flash, GPT-4
• ✅ Multi-Document Upload - Upload 1-20 PDFs per notebook
• ✅ Background Processing - Non-blocking document analysis
• ✅ Task Tracking - Real-time status updates

AI Analysis Features

• ✅ Document Summaries - AI-generated summaries per document
• ✅ Highlights Extraction - Key points from each document
• ✅ Q&A Generation - Auto-generated Q&A pairs per document
• ✅ Cross-Document Synthesis - Analyze all documents together
  • Overall summary
  • Common themes
  • Key insights
  • Comparative findings
• ✅ Podcast Generation - Create audio discussions from documents
  • Customizable length (5-30 minutes)
  • ElevenLabs voice selection
  • Custom themes/prompts
  • Background generation
  • Audio player

Collaboration Features

• ✅ Notebook Sharing - Share with other users
  • Permission levels (Read, Write, Share)
  • Email-based sharing
  • Manage access
• ✅ Chat Sessions - Multiple chat conversations
  • Private chats
  • Shared chats
  • Rename/delete sessions
  • Chat history

Chat Features

• ✅ Real-time WebSocket Chat - Instant responses
• ✅ Multi-Document Querying - Ask across all documents
• ✅ Source Citations - See which documents were used
• ✅ Session Management - Organize conversations
• ✅ Chat Sharing - Share conversations with collaborators

---

🏗️ Architecture

Tech Stack

Frontend:

• Next.js 15 (App Router)
• React 19
• TypeScript
• Tailwind CSS 4
• React Query (TanStack Query)
• Axios
• Zustand (state management)

Backend:

• FastAPI (Python)
• SQLAlchemy ORM
• PostgreSQL
• LlamaCloud/LlamaIndex
• WebSockets
• Background task queue

AI/ML:

• LlamaCloud for document indexing
• LlamaIndex for RAG
• OpenAI (GPT-4, GPT-5)
• Anthropic (Claude 3.5)
• Google (Gemini 2.0, 2.5)
• Groq, DeepSeek
• ElevenLabs (voice synthesis)

Infrastructure:

• Docker & Docker Compose
• Redis (caching)
• Jaeger (tracing)
• Adminer (database UI)

Database Schema (9 Tables)

1. users - User accounts
2. notebooks - Document collections
3. documents - Uploaded files
4. notebook_documents - Many-to-many relationship
5. document_summaries - AI analysis results
6. chat_sessions - Chat conversations
7. chat_messages - Individual messages
8. document_shares - Sharing permissions
9. background_tasks - Async job queue

---

📖 Usage Guide

1. Create Account

• Navigate to http://localhost:4000
• Click "Sign up"
• Enter email, username, password (8+ characters)

2. Create Notebook

• Click "View Notebooks"
• Click "New Notebook"
• Enter name and description
• Select AI model (see pricing)
• Click "Create Notebook"

3. Upload Documents

• Open your notebook
• Click "Select PDF Files"
• Choose 1-20 PDFs
• Click "Upload X File(s)"
• Wait ~1 minute per document for processing

4. View Analysis

• Click on any document to expand
• View AI-generated summary
• See highlights
• Review Q&A pairs

5. Generate Cross-Document Analysis

• Click "Analyze" button
• Wait for AI to analyze all documents together
• View themes, insights, and findings

6. Generate Podcast

• Click "Podcast" button
• Select length (5-30 minutes)
• Add custom theme (optional)
• Click "Generate in Background"
• Wait 3-5 minutes
• Listen to audio discussion

7. Chat with Documents

• Click "Chat" button
• Create new chat session or select existing
• Ask questions about your documents
• Get AI responses with source citations
• Rename, share, or delete chat sessions

8. Share Notebook

• Click "Share" button
• Enter collaborator's email
• Select permission level:
  • Read: View only
  • Write: View + add documents
  • Share: Write + can share with others
• Click "Share"

9. Access Shared Notebooks

• Click "Shared" in navigation
• View notebooks shared with you
• See owner, permission, document count
• Click "View" or "Chat"

---

🔑 API Keys Required

Add these to backend/.env:

# Required for all features
OPENAI_API_KEY=sk-...
LLAMACLOUD_API_KEY=llx-...

# Required for specific models
ANTHROPIC_API_KEY=sk-ant-...
GOOGLE_API_KEY=AI...
GROQ_API_KEY=gsk_...
DEEPSEEK_API_KEY=sk-...

# Required for podcast generation
ELEVENLABS_API_KEY=...

# Database (already configured)
pgql_user=postgres
pgql_psw=admin
pgql_db=postgres

Where to get keys:

• OpenAI: https://platform.openai.com/api-keys
• LlamaCloud: https://cloud.llamaindex.ai/
• Anthropic: https://console.anthropic.com/
• Google AI: https://makersuite.google.com/app/apikey
• Groq: https://console.groq.com/
• DeepSeek: https://platform.deepseek.com/
• ElevenLabs: https://elevenlabs.io/

---

🎯 AI Models & Pricing

Model	Provider	Speed	Cost (per 1M tokens)	Best For
GPT-5	OpenAI	Fast	$1.25/$10	Latest, balanced
Claude 4.5	Anthropic	Fast	$3/$15	Complex reasoning
Gemini 2.5 Pro	Google	Medium	$1.25/$5	Cost-effective
GPT-4o	OpenAI	Very Fast	$5/$15	Speed + quality
Gemini Flash	Google	Ultra Fast	$0.075/$0.30	Budget-friendly
GPT-4	OpenAI	Medium	$30/$60	Legacy support

---

🔌 API Endpoints

Full API documentation available at http://localhost:9000/docs

Key endpoints:

• /api/auth/* - Authentication
• /api/notebooks/* - Notebook management
• /api/documents/* - Document upload & summaries
• /api/chat/* - WebSocket chat & sessions
• /api/admin/* - Admin dashboard

---

🐛 Troubleshooting

Backend won't start

# Check if port 9000 is in use
lsof -i :9000

# Kill existing process
pkill -f "uvicorn.*9000"

# Restart backend
cd backend && uv run uvicorn src.api.main:app --host 0.0.0.0 --port 9000 --reload

Database connection errors

# Check PostgreSQL is running
docker ps | grep postgres

# Restart Docker services
docker-compose restart

# Reinitialize database
cd backend && uv run python -c "import sys; sys.path.insert(0, 'src/notebookllama'); from database import Base, engine; Base.metadata.create_all(bind=engine)"

Frontend won't build

cd frontend
rm -rf .next node_modules package-lock.json
npm install
npm run dev

---

📝 License

Open source - based on NotebookLlama project

---

Version: 2.0.0 Last Updated: October 2025