The End of the Brokerage Moat: Trading Inside the Chatbot

The interface for global finance is shifting from proprietary apps to conversational agents. Liquid has launched Co-Invest, a ChatGPT and Claude app that allows users to fund accounts, analyze markets, and execute live trades without leaving an AI assistant, according to new reports from The Block.

This launch represents a fundamental change in transaction architecture. While recent entries like MoonPay’s ChatGPT app function primarily as a bridge—generating checkout links that redirect users to external sites for KYC and payment—Co-Invest processes funding, analysis, and execution entirely within the conversation. The app supports exposure to crypto, equities, foreign exchange, Polymarket prediction market positions, and pre-IPO secondaries across more than 500 markets.

The significance lies in the removal of friction. Users can deposit via card, on-chain transfer, or external wallet, and set stop-loss and take-profit levels directly inside the chat interface.

This movement toward agentic finance is gaining momentum. The launch follows OpenAI's rollout on May 15 of personal finance tools for ChatGPT Pro users via Plaid, and Gemini's introduction of Agentic Trading, which enables AI models to connect to exchange accounts via the Model Context and Context Protocol (MCP).

The real battle is not just about technology, but about incentives. Established brokerages face structural disadvantages in this new era. Platforms that rely on payment for order flow or net interest margin often benefit from information asymmetry or ecosystem lock-in. As Liquid founder and CEO Franklyn Wang notes, these institutions are not incentivized to build high-quality experiences inside ChatGPT because their business models depend on keeping users within their own proprietary ecosystems.

If the next generation of traders uses AI to decide what to buy, the value will migrate to the platforms that allow for immediate execution. The brokerage of the future may not be an app you open, but a capability you call upon within an existing intelligence layer.

Watch the adoption of the Model Context Protocol (MCP). The ability for any LLM to securely interface with a trading account will determine which platforms survive the transition from walled gardens to open, agentic ecosystems.