Robinhood Rolls Out AI Trading Agents to All 29M Customers

Robinhood opened access to AI-powered trading agents for its roughly 29 million customers this week, becoming the first major U.S. brokerage to ship a nontechnical agentic investing tool at consumer scale.

The rollout, unveiled at Robinhood's annual HOOD summit, lets users pick between OpenAI's GPT-6 Luna, OpenAI's GPT-6 Sol, and Anthropic's Opus 4.8, then direct them in plain English. A demonstration viewed by Fortune showed agents executing straightforward orders such as "Buy $200 of Ford stock" and running recurring strategies the company calls "Loops."

"Ownership doesn't work without markets, and markets don't work without traders," CEO Vlad Tenev said in a statement. "We're making Robinhood the best place in the world for active traders by delivering tools once reserved for hedge funds, big banks, and quant firms."

What guardrails does Robinhood apply?

Robinhood built several controls around the product:

• Each agent runs from a separate, dedicated trading account
• Users can cap per-transaction trade size
• Users can require explicit approval before any order executes
• Customers receive temporary free access to data feeds including Unusual Whales and the crypto-focused Token Terminal

How are customers already responding?

The product builds on a Model Context Protocol tool Robinhood released in May for technical users. More than 150,000 customers had opened agentic accounts through that channel before the consumer launch. Agents already handle roughly 30 million transactions a day on Robinhood's platform as of late September, company figures show.

What could go wrong?

Mass deployment introduces new risks. Coordinated agents moving in and out of the same position could amplify volatility. Malicious actors could exploit the agent layer. Robinhood maintains that hosting the models does not amount to financial advice, likening agent output to "customers asking the internet or a friend." Where liability lands if an agent-driven trade produces a loss remains unresolved, with regulators yet to formalize rules on agentic brokerage activity.

How much will it cost users?

Robinhood plans to keep GPT-6 Luna free through year-end and charges the standard token rate for OpenAI's Sol and Anthropic's Opus 4.8. Executives say per-trade compute costs will be negligible for most users. That math shifts under three conditions:

• Token prices spike
• Retail users run research-heavy models around the clock
• Sustained "Loop" sessions drive heavy API consumption

Who else could follow?

Robinhood is alone among large U.S. brokerages offering nontechnical agents today. eToro, Public, and Coinbase allow users to connect agents via MCP but have not shipped native tools. Conventional brokerages Charles Schwab and Fidelity are widely expected to launch competing offerings within 18 to 24 months if adoption holds.

What does this mean for markets?

Active retail trading volume will probably climb as agents execute around the clock. New strategy archetypes—systematic factor trades, sentiment-driven crypto pivots—will surface. The interaction between thousands of unsupervised agents and market microstructure is untested. Robinhood's bet is that broadening access to execution will deepen participation; the open question is whether agents clustering on identical signals will introduce feedback loops that destabilize the very assets they target.