Introduction
Norn is a portfolio analytics and AI agent for retail investors. You connect the brokerages you already use, and Norn pulls your real holdings, layers on quantitative research (performance, returns, risk, diversification), and gives you an AI assistant that can reason over both.
Most investing tools tell you what the market did. Norn tells you what your portfolio did, why, and what that means for the risk you're carrying.
What you get
- Your real portfolio, in one place. Connect brokerage accounts through our aggregation partner and see positions, performance, and transactions across every account you hold.
- Research tailored to your holdings. Every insight — catalysts, risks, correlations, screening — is scoped to the positions you actually own, not a generic watchlist.
- Quant metrics on every asset. 252-day Sharpe ratio, expected return, volatility, and price history for every equity, ETF, and (for supported orgs) crypto asset Norn tracks.
- Portfolio optimization. Build saved portfolios and get optimal weights using mean-variance (historical) or Black-Litterman (with your own forward views).
- An AI assistant that reads your data. Ask questions in plain language — Norn's AI queries your holdings, runs analytics, and cites its work. Available inside Norn, and via MCP from Claude Desktop or Claude Code.
What Norn does not do
Norn is a research and information tool. It reads your accounts with read-only permissions — it cannot place trades, transfer funds, or move your assets. Nothing Norn tells you constitutes investment, financial, legal, or tax advice.
AI-generated output may be inaccurate or incomplete. You are solely responsible for your investment decisions and should consult a licensed professional where appropriate.
Who it's for
Norn is built for two audiences in the same product:
- Everyday investors who want to understand what's happening in their portfolio without drowning in jargon or generic market commentary.
- Technical users who want the same data available programmatically — through the MCP server, from any AI client that speaks the Model Context Protocol.
If either sounds like you, start with the Quickstart.