← Back to Projects
📈

Finance AI Agent

GitHub ↗
The Business Problem

Investment research is usually a single model's opinion with no adversarial pressure-testing, and no human checkpoint before a recommendation ships.

The Solution

A multi-agent investment committee — Bull, Bear, and Risk Auditor agents debate a stock adversarially before a CIO agent delivers a verdict, gated by human approval.

The Result
  • Verdicts scored against real 30-day price outcomes, not just backtested on paper
  • Hard human-in-the-loop gate before any report ships — no fully autonomous financial advice
  • Automatic Groq -> GPT-4o fallback keeps the pipeline reliable under provider outages
  • Professional PDF report generated automatically from the final debate
Architecture

Ticker Input -> Data Pipeline (yfinance/SEC/Tavily) -> Bull vs Bear Debate -> Risk Audit -> CIO Verdict -> Human Approval -> PDF Report

⚙️ Technical Details For engineers & technical reviewers — click to expand

Verdicts are scored against real 30-day price outcomes, not just backtested on paper — Bull, Bear, and Risk Auditor agents debate stocks adversarially, cross-examining each other's strongest objections before a CIO judge delivers a final call. Orchestrated by a LangGraph state machine with a hard human-in-the-loop gate before any PDF report ships, plus automatic Groq → GPT-4o fallback for reliability.

LangGraph Multi-agent Groq·Llama 3.3 OpenAI GPT-4o yfinance Tavily Human-in-the-loop State machine Streamlit

Need something like this for your own business?

Build Something Similar →

Have questions about this project? Ask my AI assistant for details.

Ask AI about this →