Inspiration The financial markets are incredibly complex, and everyday retail investors often lack access to the high-level quantitative analysis tools that Wall Street firms use. We wanted to bridge this gap. Our inspiration was to create an AI-powered "Guardian Angel" for retail investors—an autonomous agent that not only pulls real-time data but also reasons through it to provide clear, actionable investment theses.
What it does FinAgent AI is an autonomous financial analyst. Users simply enter a stock ticker, and the agentic system instantly extracts real-time market data (Current Price, 5-Day SMA, Daily Returns). The AI then processes this quantitative data to generate a structured, professional investment thesis (BUY/HOLD/SELL) explaining the market context and technical indications.
How we built it We built the backend API using FastAPI to ensure high-performance data extraction via the yfinance library. For the frontend, we used Streamlit, designing a sleek, dark-themed, enterprise-grade dashboard using custom CSS and HTML. The core intelligence is powered by OpenAI GPT-5.6, utilizing precise prompt engineering to force the model to act as a strict financial analyst.
Challenges we ran into One major challenge was dealing with rate limits and sudden API blocks from real-time data providers (like Yahoo Finance) during intensive testing. We solved this by implementing a robust "Mock Data Fallback" mechanism in our backend. If the primary data source fails, the system automatically falls back to generating realistic mock data, ensuring the AI agent always has information to process and the UI never crashes.
Accomplishments that we're proud of We are incredibly proud of the UI/UX. Financial tools are often cluttered and intimidating, but we successfully built an interface that is beautiful, modern, and highly intuitive. We're also proud of integrating a fully functional API backend with a separate frontend, mirroring a true enterprise architecture.
What we learned We learned how to orchestrate full-stack applications by running separate backend (FastAPI/Uvicorn) and frontend (Streamlit) servers. We also deeply improved our prompt engineering skills, learning how to constraint an LLM to output structured financial advice instead of generic conversational text.
What's next for FinAgent AI In the future, we plan to integrate Multi-Agent architectures where one agent specializes in Technical Analysis, another in Fundamental Analysis (reading news/earnings reports), and a final "Portfolio Manager" agent makes the final decision. We also want to add real-time WebSocket streaming for live stock tickers.
Log in or sign up for Devpost to join the conversation.