AI trading tools apply machine learning to the buying and selling of assets — and the category now stretches well beyond stock tickers. NeuronFeed tracks 21 companies with a combined $879M in funding, spanning three distinct markets: crypto, traditional finance, and, increasingly, electricity.
In crypto, Token Metrics provides AI-driven investment research and analytics, FereAI runs autonomous agents for real-time research and trading, and Donut Labs ($22M raised) builds an agentic browser that executes on-chain trades. Almanak offers an agent framework for building DeFi strategies. In traditional finance, Kensho ($123M) — an early mover later acquired by S&P Global — set the template for AI-powered financial intelligence and analytics. The most distinctive wing is energy: GridBeyond ($90M) and Sympower ($83M) use AI to trade flexibility and optimize distributed energy resources on power markets, while Shatterdome Energy applies AI to power trading and dispatch for renewables.
The practical mechanics are similar across all three: models ingest market, on-chain, or grid data; generate signals or forecasts; and either surface them to a human or hand them to an execution agent with defined risk limits. What separates leaders is auditable performance — verifiable backtests, transparent drawdowns, and hard-coded risk controls — because in this category marketing claims are cheap and survivorship bias is everywhere.
Buyers should demand out-of-sample performance data, understand exactly when the system can act without human approval, confirm custody and API-key permissions are read-only where possible, and check the regulatory status of automated trading in their jurisdiction and asset class before connecting real capital.