Conversational Customer Support Agent (RAG)

Welcome to the Conversational Customer Support Agent project! This repository contains the code for a customer support chatbot built using Retrieval-Augmented Generation (RAG). The chatbot leverages FAISS vector search, OpenAI embeddings, and LangChain conversational chains to provide accurate, context-aware responses to user queries based on product documentation.

---

Features

• Retrieval-Augmented Generation (RAG): Combines retrieval of relevant documents with generative AI for effective query resolution.

• FAISS Vector Search: Efficient similarity search over product documentation and FAQs.

• OpenAI Embeddings: Uses OpenAI's text-embedding-ada-002 model to generate vector embeddings.

• Multi-turn Memory: Maintains conversation history for coherent multi-turn interactions.

• Escalation Logic: Automatically escalates complex queries to human support agents when necessary.

• Modular Design: Easily extendable and customizable for different use cases.

---

Installation

1. Clone the Repository

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

2. Set Up a Virtual Environment

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

3. Install Dependencies

Install all required Python libraries using pip:

pip install -r requirements.txt

4. Set Up OpenAI API Key

Since you are using OpenAI embeddings, ensure your API key is set in your .bash_profile or equivalent shell configuration file.

Add the following line to your .bash_profile (or .zshrc, .bashrc, etc.):

export OPENAI_API_KEY="your-openai-api-key-here"

Then reload your shell configuration:

source ~/.bash_profile # Or source ~/.zshrc, depending on your shell

---

Usage

1. Create the Knowledge Base

Run the knowledge_base.py script to generate the FAISS vector store from your product documentation:

python knowledge_base.py

This script reads text files from the product_docs/ directory, splits them into chunks, generates embeddings using OpenAI's embedding model, and saves the FAISS index locally.

2. Run the RAG Model

Run the rag_model.py script to interact with the chatbot:

python rag_model.py

You can test queries like:

• "What is your return policy?"
• "How do I reset my password?"

3. Optional: Streamlit Interface

If you want a web-based interface, run app.py using Streamlit:

streamlit run app.py

---

Customization

1. Product Documentation

Add or update text files in the product_docs/ directory to include your own FAQs or product details.

2. Embedding Model

To use a different OpenAI embedding model, update the model parameter in both knowledge_base.py and rag_model.py.

3. Escalation Logic

Modify the _should_escalate() method in rag_model.py to customize escalation triggers.