Alex

We built Alex to make investing concepts and market information easier to explore through conversation. We wanted to create a tool that explains financial topics in plain language and connects those explanations to market data, so learners can ask questions and understand the reasoning behind an answer. Our project combines a FastAPI backend, Google’s Gemini API, and a market-data MCP server that uses yfinance and pandas. The assistant can gather information with market tools and respond with educational context. We designed it as a learning aid to help us understand investing, not as a substitute for research or personal financial judgment. Building it taught us how to connect an AI model to external tools through MCP, and how much deployment details matter when an application runs multiple Python processes. One of our biggest challenges was debugging Vercel startup failures. First, the deployed app couldn’t find its Gemini API key. Then the market-data subprocess couldn’t import pandas, even though it was listed as a project dependency. Reading the logs helped us trace the errors from the HTTP 500 back to startup and the subprocess environment. We’re continuing to improve the assistant’s reliability, handle model limits gracefully, and make its financial explanations clear enough to help us learn rather than simply follow a recommendation.

Built With

• fastapi
• gemini
• javascript
• mcp
• pandas
• python
• vercel
• yfinance