RAG Chatbot Project

This project implements a Retrieval-Augmented Generation (RAG) chatbot using:

• Chroma for vector storage
• OpenAI for LLM responses
• n8n as the orchestration layer
• Streamlit for a user-friendly chat UI

🏗️ How It Works

1. Document Ingestion & Embedding: PDF documents are loaded, chunked, and embedded using OpenAI embeddings, then stored in ChromaDB.
2. Query Workflow (n8n): When a user asks a question, the workflow:
   • Retrieves relevant context from ChromaDB (via a Python script)
   • Calls OpenAI's API to generate an answer using the retrieved context
   • Returns the answer via an HTTP webhook
3. Streamlit UI: Provides a chat interface that sends user queries to the n8n webhook and displays the bot's answers.

---

📦 Prerequisites

• Python 3.10+
• Node.js + n8n
• OpenAI API Key
• pip install langchain openai chromadb pymupdf streamlit requests

---

📁 Project Structure

ragbasedchatbot/
├── workflows/                  # n8n workflow export
│   └── rag-chatbot-workflow.json
├── scripts/                    # Setup and query code
│   ├── setup.py                # Document loader + vector indexer
│   └── query_chroma.py         # Used in n8n to perform query
├── data/
│   └── sample-documents/       # FAQ PDFs
├── docs/                       # Documentation
│   ├── README.md
│   └── technical-approach.md
├── tests/
│   └── sample-queries.json     # For testing chatbot
├── streamlit_app.py            # Streamlit chat UI

---

⚙️ Setup & Run

1. Setup Embeddings

python3 scripts/setup.py

2. Start n8n

n8n start

3. Import Workflow

• Go to n8n UI → Import workflows/rag-chatbot-workflow.json
• Activate the workflow (ensure the webhook is listening at http://localhost:5678/webhook/faq-query)

4. Test via HTTP Webhook (Optional)

curl -X POST http://localhost:5678/webhook/faq-query \
     -H "Content-Type: application/json" \
     -d '{"query": "What are the values of Altibbe?"}'

5. Run the Streamlit Chat UI

streamlit run streamlit_app.py

• Open the provided local URL (usually http://localhost:8501) in your browser.
• Type your question and chat with the bot!

---

🔑 Environment Variables

• OPENAI_API_KEY: Your OpenAI key (required for both embedding and LLM generation)
• CHROMA_DB_PATH: Local path to Chroma persistence (default: chroma_db)

---

🧪 Testing

• Run all sample queries in tests/sample-queries.json and verify expected keywords are returned.
• Add your own PDFs to data/sample-documents/ and re-run the embedding setup as needed.

---

📝 Project Workflow Summary

1. Ingest & Embed: Load and embed documents with setup.py.
2. n8n Orchestration: Handles query routing, retrieval, and LLM response.
3. Streamlit UI: User-friendly chat interface, interacts with n8n via HTTP.

---

🙌 Credits & Further Improvements

• Built with LangChain, Chroma, OpenAI, n8n, and Streamlit.
• See docs/technical-approach.md for more details and future improvement ideas.