• InvestGPT
  • Quick Start
  • Implementation Details
    • Session-based Storage
    • Deciding When to Call API
  • Credit

A COMP4531 Project. Use GPT-3.5 to map user risk profile and use our own API and database to suggest investment portfolio.

Make sure you have OpenAI API key stored in .env file (OPENAI_API_KEY)

# Proxy
pip install -r requirements.txt
python app.py # :5000

# Client
cd client
npm install
npm start # :3000

The simple demo app does not have an account system. Instead, we use session-based (flask-session) storage to store user's conversation history in a single session. When you refresh the frontend, you will start a new conversation.

We've instructed GPT to signal to us when the user data is sufficient to make an API call by let it including a structured data (i.e., JSON) in the response. We then parse the response and make the corresponding API call.

React client is adapted from react-openai-chat.