Customer Support Agent

AI-powered customer support system with LangGraph workflow orchestration, RAG (Retrieval-Augmented Generation), and intelligent routing.

Features

• 🤖 AI-Powered Responses: Uses OpenAI GPT-4o-mini for intelligent customer support
• 🔍 RAG with ChromaDB: Vector database for knowledge base retrieval
• 🎯 Smart Routing: Automatic categorization (Technical, Billing, General)
• 😊 Sentiment Analysis: Detects customer sentiment (Positive, Neutral, Negative)
• 🔄 LangGraph Workflows: State-based conversation management
• 🚀 Production Ready: FastAPI backend with NGINX reverse proxy
• 📊 Real-time & REST APIs: WebSocket and HTTP endpoints

Quick Start

Prerequisites

• AWS EC2 instance (Ubuntu 24.04 LTS, t2.medium or larger)
• OpenAI API key
• Domain name (optional, for HTTPS)

Deployment on EC2

1. Clone the repository:

git clone https://github.com/YOUR_USERNAME/YOUR_REPO.git ~/customer-support-agent
cd ~/customer-support-agent

2. Run the setup script:

sudo chmod +x deployment/ec2/setup.sh
sudo deployment/ec2/setup.sh

3. Configure your OpenAI API key:

sudo nano ~/customer-support-agent/backend/.env
# Add: OPENAI_API_KEY=sk-your-actual-key-here

4. Restart the service:

sudo systemctl restart support-agent

5. Access your application:

• Frontend: http://YOUR-EC2-IP/
• API Docs: http://YOUR-EC2-IP/docs
• Health Check: http://YOUR-EC2-IP/health

Project Structure

.
├── backend/
│   ├── app/
│   │   ├── agents/          # AI agent modules
│   │   ├── config/          # Configuration
│   │   ├── database/        # ChromaDB integration
│   │   ├── models/          # Pydantic schemas
│   │   ├── workflows/       # LangGraph workflows
│   │   └── main.py          # FastAPI application
│   ├── data/                # Knowledge base JSON
│   ├── requirements.txt     # Python dependencies
│   └── .env.example         # Environment variables template
├── frontend/
│   ├── index.html           # Web UI
│   └── static/              # CSS, JS, images
└── deployment/
    ├── ec2/                 # EC2 deployment files
    │   ├── setup.sh         # Automated setup script
    │   ├── nginx/           # NGINX configuration
    │   └── systemd/         # Systemd service
    └── docker/              # Docker deployment (optional)

Configuration

Edit backend/.env:

# Required
OPENAI_API_KEY=sk-your-key-here

# Optional (defaults provided)
LLM_MODEL=gpt-4o-mini
EMBEDDING_MODEL=text-embedding-3-small
API_PORT=8000

API Endpoints

• POST /api/chat - Send a customer query
• GET /health - Health check
• GET /api/status - API status
• WS /ws/chat - WebSocket chat endpoint

Managing the Service

# Check status
sudo systemctl status support-agent

# View logs
sudo journalctl -u support-agent -f

# Restart
sudo systemctl restart support-agent

# Stop
sudo systemctl stop support-agent

Tech Stack

• Backend: FastAPI, Python 3.12
• AI: OpenAI GPT-4o-mini, LangChain, LangGraph
• Database: ChromaDB (vector database)
• Web Server: NGINX
• Process Manager: Systemd
• Frontend: Vanilla HTML/CSS/JavaScript

License

MIT License

Support

For issues and questions, please open a GitHub issue.