Customer Service Starter

An intelligent customer service AI tool that helps customers with flight information, bookings, and travel-related queries. The bot uses Claude 3 Sonnet for natural language understanding and Azure OpenAI for embeddings. Langgraph framwork is used for state and agent chaining & routing. This starter is based on official Langgraph documentation.

Features

• Flight information lookup
• Ticket management (view, update, cancel)
• Car rental services
• Hotel bookings
• Trip recommendations
• Policy information retrieval
• Natural language conversation

Prerequisites

• Python 3.8+
• Anthropic API key (for Claude 3)
• Azure OpenAI API key and endpoint (for embeddings)
• SQLite database with travel information

Environment Variables

Create a .env file in the project root with the following variables:

ANTHROPIC_API_KEY=your_anthropic_api_key
AZURE_OPENAI_API_KEY=your_azure_openai_api_key
AZURE_OPENAI_ENDPOINT=your_azure_endpoint
AZURE_OPENAI_API_VERSION=your_api_version
DB_URL=your_database_url
FAQ_URL=your_faq_url
MODEL_NAME=claude-3-sonnet-20240229  # Optional, defaults to this value

Installation

1. Clone the repository:

git clone https://github.com:evuori/customer-service-starter.git
cd swiss-airlines-bot

2. Create and activate a virtual environment:

python -m venv venv
source venv/bin/activate  # On Windows: venv\Scripts\activate

3. Install dependencies:

pip install -r requirements.txt

Usage

1. Run the bot:

python customer_support_bot.py

2. The bot will automatically:

   • Download and set up the SQLite database
   • Load FAQ information
   • Initialize the conversation system
   • Start processing customer queries

3. Example interaction:

config = {"configurable": {"passenger_id": "0123456789"}}
messages = [("user", "Hi there, what time is my flight?")]
result = part_1_graph.invoke({"messages": messages}, config=config)

Project Structure

• customer_support_bot.py: Main bot implementation
• travel2.sqlite: SQLite database (downloaded automatically)
• .env: Environment variables (not included in repository)
• requirements.txt: Project dependencies

Logging

The bot includes comprehensive logging that shows:

• Conversation flow
• Tool execution
• State transitions
• Error messages
• Agent and tool names

Logs are written to the console with timestamps and log levels.

Contributing

1. Fork the repository
2. Create a feature branch
3. Commit your changes
4. Push to the branch
5. Create a Pull Request

License

This project is licensed under the MIT License - see the LICENSE file for details.

Acknowledgments

• Anthropic for Claude 3
• Microsoft Azure for OpenAI services
• LangChain for the framework