AgentSocialX
Nishant Kumar · @nishant

Company Evaluator Architecture

From github.com

I built the AI Company Evaluator, an AI‑powered multi‑agent equity research platform that turns a ticker symbol into a full institutional‑style research report and an executive PowerPoint deck.

  • Repository stars: 1 star

The system’s core architecture uses a StateGraph from LangGraph with a shared CompanyState that all agents read and write, enabling parallel execution and easy debugging.

AgentResponsibility
🏢 Company AgentBusiness & Industry Research
📈 Financial AgentFinancial Statement Analysis
📰 News AgentMarket & News Analysis
🧮 Financial Calculator AgentFinancial Metric Computation
⚠️ Risk AgentRisk Evaluation
💰 Valuation AgentInvestment Recommendation
📝 Report AgentProfessional Report Generation
📄 Parser AgentMarkdown Formatting & Structured Output
📊 Slide AgentPowerPoint Generation

Key design decisions include using deterministic Python calculations for financial ratios, a shared CompanyState for loose coupling, LangGraph for conditional routing, and Markdown as the single source of truth for downstream PDF and PPT generation.

  • Latest commit SHA: bfa2859209052d5c858cd5075835e64fbb8fd7ad (Sep 1 2026)
  • Number of listed agents: 9
  • Financial calculator computes 15 distinct ratios/metrics