Robinhood Agents Bring AI Trading to 27 Million Customers

Robinhood has launched Robinhood Agents, an AI-native trading platform built directly into its app that can research markets, build strategies and place trades on a customer's behalf at any hour. The company unveiled Robinhood Agents on September 29, 2026 at its annual HOOD Summit in Houston, describing it as part of its most ambitious expansion for active traders. CoinDesk reported the details of the announcement in its September 30 coverage.

The launch of Robinhood Agents extends a soft release from May, when Robinhood let technically sophisticated customers connect their own AI agents to the platform. Since then, more than 150,000 customers have opened agentic trading accounts, according to the company. Those agents now tap Robinhood's tools almost 30 million times per day.

How Robinhood Agents work

Customers start by giving an agent a name, then open a dedicated agentic trading account that is kept separate from their main brokerage balance. The account structure limits the agent's ability to act beyond its intended scope, since Robinhood Agents can operate only with the funds held inside the dedicated account.

Next, the customer selects the model that powers the agent. Options include several AI providers. In a demonstration viewed by Fortune, a user chose between OpenAI's GPT-6 Luna, GPT-6 Sol and Anthropic's Opus 4.8. According to Crypto Briefing's report on the summit, the agent is designed to handle automatic trading assistance and delegate certain execution tasks to an AI co-pilot.

Once configured, the customer gives plain-English instructions to the Robinhood Agents setup. The agent can carry out tasks ranging from simple orders, such as buying a set amount of stock, to research assignments and the construction of full investment strategies. Pulse2 noted that usage with OpenAI GPT-Luna is free through the end of 2026.

A trade-approval setting is enabled by default. With approvals on, the agent cannot place an order until the customer approves it. The setting can be turned off, after which Robinhood Agents can execute orders without asking for confirmation on each one.

Robinhood also introduced Agent Apps, a layer of third-party intelligence that plugs into an agent to extend its analytical capabilities with institutional-grade data and specialized tools. Chief Executive Vlad Tenev described the broader goal as delivering tools once reserved for hedge funds, big banks and quant firms to everyday active traders.

Loops and the always-on trader

The next feature on the Robinhood Agents roadmap is Loops, which turns a strategy into a standing instruction the agent carries out on repeat, day and night. A Loop could check the market every morning and trade when certain conditions are met, or run an overnight strategy while the customer sleeps.

The Loop concept marks the shift from AI agents that act when asked to agents that act continuously. The May launch proved that Robinhood's customers will connect agents to real money; the Robinhood Agents release removes the need to configure external infrastructure to do it.

The same summit brought several companion announcements for active traders. Robinhood plans weekend equity trading covering a selection of US stocks and ETFs, expected in early 2027, along with crypto perpetual futures offering up to 10x leverage on bitcoin and ether for eligible US customers. Earnings contracts for key company metrics are also on the way.

Shares rose about 2.6 percent on the day of the announcement, signaling investor approval of the expansion.

The risks when agents move money

Robinhood's own disclosures make clear who carries the risk. Customers assume all risk for trades executed by AI agents and for any use of their data by third-party model providers. The company said it does not control, supervise, monitor, recommend or audit agents.

The warning sharpens for Loops. Once switched on, a Loop may place, modify or cancel trades automatically, without prompting for approval on each transaction, including while the customer is asleep or away. It follows the customer's rules exactly as configured, even during periods of market volatility.

Trades already placed by a Loop are not automatically reversed when the Loop is turned off. Robinhood also says it does not guarantee how Loops will perform in any given market condition.

Regulators and researchers are watching what happens when many trading agents operate at once. Bank of England Deputy Governor Sarah Breeden warned in June that autonomous AI agents could amplify volatility in stress and trigger a market meltdown, adding that existing financial regulation was not built for agentic systems. Her main concern is herding, in which many agents react to the same news in the same way at the same time.

A study by Wharton and the Hong Kong University of Science and Technology found that AI-powered trading agents in a simulated environment colluded with one another, fixing prices to make a collective profit even without an explicit communication channel. The findings complicate the regulatory picture further.

The launch fits a broader trend of AI agents that move money. Meta's Muse assistant can already see users' bank balances and investments, and x402, a payments protocol developed by Coinbase, lets agents pay for services in stablecoins. Agentic trading, long the domain of quant funds, is now reaching everyday investors one app update at a time.

The economics of the models behind these agents are also shifting fast. OpenAI and Anthropic recently matched prices on their agent-friendly model tiers, a cut that makes continuous agent operation far cheaper to run. At the same time, enterprises are bringing governed AI agents into their own systems, signaling that the agent economy is expanding on both the consumer and corporate fronts.