On May 27, Robinhood announced the launch of a beta version of ‘Agentic Trading’, allowing users to connect third-party AI agents such as Claude and ChatGPT to the platform via its MCP (Model Context Protocol) server, granting them permission to buy and sell stocks autonomously. Users must set up a separate, isolated account solely for the AI; this agent can only access funds specifically deposited into that account and has no access to assets in the main account. Notifications are sent for every transaction, enabling users to monitor profits and losses in real time, and they can revoke the AI’s trading rights at any moment with a single click. Currently, the beta version supports stocks only; support for options, cryptocurrencies, futures, and other assets is planned for future releases. Concurrently, Robinhood also rolled out an ‘Agentic Credit Card’ feature, which allows AI to make purchases automatically within predefined spending limits set by users. CEO Vlad Tenev stated, “Our mission to democratize finance now extends to AI agents.”
This marks the first time a major retail brokerage has granted ordinary individual investors autonomous trading rights. Previously, quantitative models and AI-driven automated decision-making systems were primarily used by institutional players like hedge funds, with technical barriers making them largely inaccessible to retail investors. In its announcement, Robinhood explicitly clarified that it does not supervise, control, or guarantee the performance of any AI agents; users bear full responsibility for all trading outcomes, and there remains a risk of agents misinterpreting instructions or basing decisions on outdated data. In a subsequent research report, Goldman Sachs characterized this launch as “an early attempt to integrate AI agents into retail brokerages and consumer finance,” noting that both the market size and likelihood of success remain difficult to assess at present.