📚 Multi-Agent Book Assistant

A production-ready AI assistant using LangGraph, AWS Bedrock Nova Lite, and LangSmith for intelligent book queries with full observability.

Three Implementations

1. Single-Agent (app.py)

Simple routing with LangGraph workflows - good for learning basics

2. Multi-Agent Custom (app_multiagent.py)

Custom supervisor pattern with agent memory - good for understanding multi-agent concepts

3. LangGraph Multi-Agent (app_langgraph.py) ⭐ RECOMMENDED

Full LangGraph StateGraph implementation with LangSmith integration:

• Proper State Management: TypedDict state across all nodes
• Supervisor Node: Intelligent routing with LLM
• Specialist Agent Nodes: BookDetails, Reviews, Recommendation, General
• LangSmith Tracing: Full observability and debugging
• Visual Workflow: See execution graph
• Production Ready: Scalable and maintainable

Features

• 📚 Query book details and specifications
• ⭐ Get book reviews and ratings
• 🎯 Personalized book recommendations
• 💡 Answer general questions about books
• 🤖 Multi-agent coordination for complex queries
• 🔍 LangSmith tracing for debugging
• 📊 Real-time analytics dashboard
• 🗺️ Visual workflow graphs

Quick Start

1. Install Dependencies

pip install -r requirements.txt

2. Configure Environment

cp .env.example .env
# Edit .env with your AWS credentials
# Optionally add LangSmith API key for tracing

3. Scrape Book Data

python3 scraper.py

4. Run the Assistant

# Recommended: LangGraph with LangSmith
python3 app_langgraph.py

# Or try other implementations:
python3 app.py                # Simple single-agent
python3 app_multiagent.py     # Custom multi-agent

See QUICKSTART.md for detailed instructions and LANGSMITH_SETUP.md for tracing setup.

Data Collection

The scraper uses:

• Google Books API for book details (title, author, pages, description, ratings)
• Mock reviews generated based on ratings (real review APIs require authentication)

Run python scraper.py to collect data for 10 categories with 10 books each.

Example Queries

• "What's the page count of Harry Potter?"
• "What do readers think about 1984?"
• "Recommend a good science fiction book"
• "Compare reviews of two fantasy books" (multi-agent)
• "Who wrote The Great Gatsby?"

Documentation

• 📖 QUICKSTART.md - Step-by-step setup guide
• 🏗️ ARCHITECTURE.md - Multi-agent design details
• 🔍 LANGSMITH_SETUP.md - LangSmith tracing configuration
• 📊 IMPLEMENTATIONS.md - Compare all three implementations
• 🗂️ PROJECT_STRUCTURE.md - File organization

LangSmith Integration

Enable tracing to debug and monitor your agents:

1. Sign up at https://smith.langchain.com
2. Get your API key
3. Add to .env:

LANGCHAIN_TRACING_V2=true
LANGCHAIN_API_KEY=your_key_here
LANGCHAIN_PROJECT=book-assistant-multiagent

See LANGSMITH_SETUP.md for detailed instructions.

Technology Stack

• LangGraph: Agent workflow orchestration
• LangSmith: Tracing and monitoring
• AWS Bedrock Nova Lite: LLM inference
• Gradio: Web interface
• Google Books API: Book data source