FinAgent V2 — Autonomous Multi-Agent Investment AI

FinAgent V2 is an autonomous multi-agent investment analysis system built for VC/PE firms, fund managers, and investment analysts who need structured, deep analysis without a dedicated research team. Unlike single-model approaches, V2 deploys five specialized AI agents coordinated by an LLM-driven Orchestrator that uses function calling to dynamically plan which agents to activate based on user input and natural language intent. Three Analysis Modes: Mode A — Market Analysis: Enter a stock ticker. The Orchestrator dispatches QuantAgent, PeerAgent, and CIOAgent in parallel to deliver complete market analysis including valuation models (PE/PB/PEG/EV-EBITDA/DCF), peer comparison, technical trends, and investment verdict. Mode B — Report + Market Analysis: Upload a financial report PDF and enter a ticker. FundamentalAgent and QuantAgent run simultaneously — one reads the document using Gemini multimodal AI, one pulls live market data via yfinance. CIOAgent then cross-analyzes both outputs, surfacing divergence signals between reported fundamentals and current market performance. Mode C — AI Dialogue: Talk to the Orchestrator in natural language. It introduces itself, asks clarifying questions, confirms the analysis plan, then dispatches the right combination of agents. True conversational orchestration that adapts to user needs. Key technical features: streaming dashboard where results appear section by section as each agent completes, intelligent peer identification using sector and geography context, reflection loop where the Orchestrator synthesizes from LLM knowledge when agent data is incomplete, and optional Featherless open-source model inference. Built on: Gemini API, OpenAI function calling, React/TypeScript, Express, Vultr (Milan region), yfinance, Featherless.