Finance Transcripts Bot

A Streamlit-based chat application that uses Retrieval Augmented Generation (RAG) to analyze financial transcripts and provide intelligent responses. The application includes tools for stock trading simulation and observability through Galileo.

Demo URL: https://build-better-agents-with-galileo.streamlit.app/

Key Features

• RAG-powered question answering over financial transcripts
• Stock trading simulation with purchase/sell functionality
• Galileo observability integration
• Configurable AI models (GPT-4 or GPT-3.5 Turbo)
• Experiment runner for batch processing
• Streamlit web interface with real-time chat

Project Structure

.
├── app.py                 # Main Streamlit application
├── experiment_runner.py   # (Unused) Script for experiment runner 
├── generate_env.py        # (Unused) Script to convert secrets.toml to .env
├── galileo_api_helper.py  # Helper functions for Galileo API
├── pages/                 # Streamlit page components
│   └── 2_run_experiment.py
├── tools/                 # Financial tools
│   ├── purchase_stocks.py
│   ├── sell_stocks.py
│   └── get_stock_price.py
│   └── get_ticker_symbol.py
└── log_hallucination.py  # Hallucination logging utility

Installation

1. Clone the repository
2. Install dependencies:

pip install -r requirements.txt

3. Set up secrets in .streamlit/secrets.toml

openai_api_key = "your_openai_api_key"
pinecone_api_key = "your_pinecone_api_key"
galileo_console_url = "your_galileo_console_url"
pinecone_index_name = "galileo-demo"
pinecone_namespace = "sp500-qa-demo"
galileo_api_key = "your_galileo_api_key"
galileo_project = "your_galileo_project"
galileo_log_stream = "your_galileo_log_stream"
ALPHA_VANTAGE_API_KEY= "your_alpha_vantage_key"

Running the Application

Streamlit UI

streamlit run app.py

The application will be available at http://localhost:8501

Usage

The application provides:

1. Interactive chat interface for financial queries
2. Stock trading simulation tools
3. Batch experiment processing
4. Galileo observability for monitoring and debugging

Tools

• Stock Purchase Simulation: Simulates buying stocks with specified ticker, quantity, and price
• Stock Sale Simulation: Simulates selling stocks with specified ticker, quantity, and price
• Ticker symbol lookup: Using the company name (uses AlphaVantage API key or mocks the data if over API limit)
• Get Stock price: Using the ticker symbol (uses AlphaVantage API key or mocks the data if over API limit)
• RAG System: Uses Pinecone vector store for semantic search over financial transcripts

Observability

The application uses Galileo for:

• Logging chat interactions
• Tracking tool usage
• Monitoring RAG performance
• Debugging and error tracking