Customer Support RAG Chatbot

What is Customer Support RAG Chatbot?

Customer Support RAG Chatbot is a retrieval-augmented generation (RAG) chatbot that answers queries based on provided customer support documentation. It uses semantic search to find relevant information and generates responses using the Hugging Face API, responding with “I don’t know” for questions outside its documentation scope.

How to use Customer Support RAG Chatbot?

Clone the repository, set up a Python virtual environment, install dependencies, and create a .env file with your Hugging Face API key. Gather documentation (scrape AngelOne docs or place insurance PDFs in docs/insurance), then run process_documents.py. Start the backend with python mcp_server.py and the frontend with streamlit run app.py. Open the deployed URL in a browser and type questions in the chat interface.

Key features of Customer Support RAG Chatbot

• Answers questions based on provided customer support documentation.
• Responds with “I don’t know” for out-of-scope questions.
• User-friendly web interface built with Streamlit.
• Semantic search for relevant information retrieval.
• Context-aware responses using Hugging Face LLM.
• Web scraping for AngelOne documentation.
• PDF processing for insurance documents.

Use cases of Customer Support RAG Chatbot

• Provide instant answers to customers using product documentation.
• Support internal help desks with a FAQ‑style knowledge base.
• Automate first‑line support for insurance or financial services.
• Reduce human agent workload by handling common queries from manuals.

FAQ from Customer Support RAG Chatbot

What documentation does the chatbot use?

It uses customer support documentation you provide: scraped AngelOne web pages (via scrape_angelone.py) and any insurance PDFs placed in the docs/insurance directory. The system only answers from these documents.

What runtime dependencies are required?

Python, FastAPI (backend), Streamlit (frontend), Hugging Face API (LLM), ChromaDB (vector database), Sentence Transformers (embeddings), PyPDF, BeautifulSoup4, and Requests. Install with pip install -r requirements.txt.

Where is the document data stored?

Processed documents are stored in a ChromaDB vector database in the chroma_db/ directory. The raw documents reside in the docs/ folder.

How does the chatbot handle questions without matching information?

It responds with “I don’t know” to any question that cannot be answered from the provided documentation.

What transport or authentication is used?

The frontend communicates with the backend via a standard HTTP API (FastAPI). Authentication is not described; the only external credential is the Hugging Face API key stored in the .env file.