RAG-based Support Agent

A RAG (Retrieval-Augmented Generation) based support agent that learns from resolved Zendesk tickets to provide accurate responses to new support queries.

Features

• Fetches resolved tickets from Zendesk
• Builds a knowledge base using RAG
• Processes open tickets and suggests responses
• Interactive mode for testing responses
• Environment variable configuration for sensitive data

Setup

1. Clone the repository:

   git clone https://github.com/djpapzin/rag-based-support-agent.git
   cd rag-based-support-agent

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

4. Set up environment variables:

   cp .env.example .env

   Edit .env with your configuration values.

Usage

1. Process resolved tickets to build knowledge base:

   python -m src.main

2. Enter interactive mode to test responses:

   python -m src.main --interactive

Configuration

The following environment variables can be configured in .env:

• ZENDESK_API_URL: Your Zendesk API URL
• ZENDESK_API_KEY: Your Zendesk API key
• OPENAI_API_KEY: Your OpenAI API key
• HUGGINGFACE_API_KEY: Your Hugging Face API key
• VECTOR_STORE_PATH: Path to store vector embeddings (default: "data/vector_store")

License

MIT License