APRA Information Chatbot 🏛️

An AI-powered chatbot that provides information about Australian Prudential Regulation Authority (APRA) standards and guidelines. Built using RAG (Retrieval-Augmented Generation) technology with Chainlit, LangChain, Pinecone, and OpenAI.

📋 About

This chatbot uses publicly available APRA documentation to answer questions about:

• APRA Connect Guide (May 2025)
• Prudential Standard CPS 230 (Operational Risk Management)
• Prudential Practice Guide CPG 230 (Operational Risk Management)
• Other APRA regulatory guidance

Note: This chatbot uses only public APRA documents and is intended for informational purposes.

✨ Features

• 🤖 Conversational AI Interface - Beautiful Chainlit UI for natural conversations
• 📚 Vector Database Search - Pinecone-powered document retrieval
• 🔍 Source Citations - Shows exact document sources for each answer
• 💬 Memory-Enabled Chat - Maintains conversation context
• 🏛️ APRA-Specific Knowledge - Focused on Australian prudential regulation
• 📊 Quality Evaluation - Integrated RAGAS and LLM-as-Judge evaluation metrics
• 📈 Observability - Langfuse integration for conversation monitoring and analytics

🚀 Live Demo

Deployed on Railway: [Your App URL]

🛠️ Technology Stack

• Frontend: Chainlit (Python-based chat interface)
• Backend: LangChain (LLM orchestration)
• Vector Database: Pinecone (document embeddings storage)
• LLM: OpenAI GPT-3.5-turbo
• PDF Processing: PyMuPDF
• Deployment: Railway (Docker container)

⚙️ Setup & Installation

Prerequisites

• Python 3.11+
• OpenAI API key
• Pinecone API key
• Docker (for deployment)

Local Development

1. Clone the repository:

   git clone https://github.com/xichaow/knowledgebase_chatbot.git
   cd knowledgebase_chatbot

2. Install dependencies:

   pip install -r requirements.txt

3. Set up environment variables:

   cp .env.example .env

   Edit .env and add your API keys:

   OPENAI_API_KEY=your_openai_api_key_here
   PINECONE_API_KEY=your_pinecone_api_key_here
   LANGFUSE_PUBLIC_KEY=your_langfuse_public_key_here
   LANGFUSE_SECRET_KEY=your_langfuse_secret_key_here
   

4. Process PDFs and build vector store:

   python vector_store.py

   This will process the APRA PDFs in the pdf/ directory and create embeddings in Pinecone.

5. Run the chatbot:

   chainlit run app.py

6. Open your browser to http://localhost:8000

🐳 Deployment

Railway Deployment (Recommended)

This app is configured for easy deployment on Railway:

1. Fork this repository
2. Sign up at Railway
3. Create new project → Deploy from GitHub repo
4. Add environment variables in Railway dashboard:
   • OPENAI_API_KEY
   • PINECONE_API_KEY
   • LANGFUSE_PUBLIC_KEY (optional, for evaluation)
   • LANGFUSE_SECRET_KEY (optional, for evaluation)
5. Deploy automatically - Railway will build and deploy using the Dockerfile

Cost: ~$5/month on Railway

Other Deployment Options

• Google Cloud Run: Use the included Dockerfile
• AWS ECS/Fargate: Container-ready deployment
• DigitalOcean App Platform: Simple container deployment
• Render: Alternative to Railway

📁 Project Structure

knowledgebase_chatbot/
├── app.py                      # Main Chainlit application
├── vector_store.py             # PDF processing and Pinecone setup
├── requirements.txt            # Python dependencies
├── Dockerfile                  # Container configuration
├── .env.example               # Environment variables template
├── .gitignore                 # Security exclusions
├── pdf/                       # APRA PDF documents
│   ├── APRA Connect Guide May 2025_0.pdf
│   ├── Prudential Practice Guide CPG 230.pdf
│   └── Prudential Standard CPS 230.pdf
└── README.md                  # This file

🔒 Security & Privacy

• Environment Variables: API keys stored securely, not in code
• Public Documents Only: Uses only publicly available APRA documents
• No Sensitive Data: PDF directory excluded from git repository
• Secure Deployment: Environment variables injected at runtime

💡 Usage Examples

Ask the chatbot questions like:

• "What is CPS 230?"
• "How do I access APRA Connect?"
• "What are the operational risk management requirements?"
• "Explain prudential standards for operational risk"
• "How do I create returns in APRA Connect?"

🤝 Contributing

1. Fork the repository
2. Create a feature branch
3. Make your changes
4. Test locally
5. Submit a pull request

📄 License

This project is open source. The APRA documents used are publicly available from the APRA website.

⚠️ Disclaimer

This chatbot provides information based on publicly available APRA documents. It is not official APRA guidance and should not be used as a substitute for professional advice or official APRA communications. Always refer to the official APRA website for the most current information.

---

Built with ❤️ for the Australian financial services community