The frustration is familiar to anyone who uses AI to research investments. ChatGPT or Claude builds the thesis, weighs the risks and defines the position. Then the tab closes and the mediation application opens. Analysis and execution have always lived in separate places – until now.
The liquid is new Participate in investing The app is embedded directly within ChatGPT and Cloud, allowing users to fund accounts, analyze markets, and make live trades without leaving the chat. By Tuesday (May 26) press releaseThe app covers over 500 markets including cryptocurrencies, stocks, foreign exchange, Polymarket prediction sites and pre-IPO secondaries. Users can set stop loss and take profit levels within chat and confirm each order with one click. Liquid operates in a non-custodial manner, keeping users’ funds in their private wallets.
Millions of people are already listening to AI to know what to invest in, Franklin Wangsaid Liquid’s CEO Cluster. Putting implementation into the same interface is the next logical step, he said.
Consumer habit is already forming
The market that Co-Invest enters is already large. PYMNTS INTELLIGENCE Found As of December, 37% of powerful AI users reported using native AI platforms as their primary tool for managing finance and banking. Among the main users, the share doubled in one month. PYMNTS Intelligence too I mentioned Only 14% of light users felt comfortable using AI for financial tasks, a gap that indicates how concentrated financial AI adoption is among more engaged users.
Co-Invest arrives alongside a host of similar moves. Robinhood Fired Proxy Trading and Proxy Credit Card, allowing AI agents to execute trades and card payments on a customer’s behalf through a personalized virtual card with a user-defined spending limit. MoonPay recently Fired Its ChatGPT integration for cryptocurrency purchases. OpenAI Rolled Personal finance tools for ChatGPT Pro users through Plaid integration in May. Gemini introduced proxy trading that allows AI models to connect directly to stock exchange accounts. The pattern across all of them is the same: AI assistants move from the information layer to the transaction layer.
Risks that come with implementation
The argument for convenience is straightforward. It is difficult to reject the risk argument. FINRA2026 Annual Regulatory Oversight Report It has been marked Hallucinations as a specific compliance concern for brokers and traders, warning companies to develop procedures for AI agents that may behave outside the user’s intended scope. Organizer Specific hallucinations Such as cases where the model generates inaccurate or misleading information but presents it as factual.
Advertisement: Scroll to continue
In financial contexts, this definition has direct consequences. luck I mentioned In April, large language models are stochastic in nature, meaning that even well-tuned agents can produce errors with the same confidence as correct output. When a dealer sits at the helm of a brokerage account, an obsessive price, a wrong recording, or a misinterpreted order parameter can move money before the user even notices.
Both Liquid and Robinhood have built confirmation steps into their products. Liquid requires a user click to confirm each trade. Robinhood lets customers choose whether they want manual approvals and defaults agents to a dedicated virtual card without access to basic account information. Robinhood Vice President of Product Management Abhishek Fatehpuria He said The design reflects what customers said they wanted: giving customers the power of Robinhood, but in a secure way.
Whether these guardrails are sufficient for users watching AI-generated analyzes that may almost certainly be wrong is a question that has yet to be answered.
