Revolut X now lets traders analyze portfolios, backtest strategies, and execute orders via AI prompts. The 30-minute prototype using Anthropic's MCP protocol cuts friction but relies on human approval. Proof will come from user forums.
Revolut turned its crypto exchange into a conversational interface. The fintech company said Revolut X, its dedicated crypto trading platform, now integrates with AI assistants including Anthropic's Claude, Google's Gemini, and Cursor. Users can trade, analyze, and backtest strategies through natural language prompts.
The integration started as a 30-minute internal experiment. Engineers connected Claude to the Revolut X API and produced a functional market-making workflow covering inventory management, quoting, execution, monitoring, and alerts, the company said. The technical foundation is Anthropic's Model Context Protocol (MCP), released in November 2024, layered on top of the existing trading API.
Practical applications go beyond the usual AI-finance hype. A user can set up Telegram notifications triggered when ETH hits a specific price target, or backtest a grid trading strategy on Bitcoin over a 30-day window, all through conversational prompts. Revolut also published a dedicated GitHub repository with universal installation support across more than 50 AI applications.
Two safety measures matter. Revolut requires user review and approval for all orders before execution. No trade goes through without a human confirming it. The company also explicitly disclaims liability for AI-driven errors.
The feature follows Revolut's rollout of AIR, its in-app AI assistant for general financial tasks, in April 2026. That earlier product handled broader banking queries. The new integration targets the crypto trading audience at a deeper technical level, treating Revolut X's API as a core execution layer rather than just an information source.
For a trader looking at this, the real value is not the chatbot itself. It is the ability to string together alerts, backtests, and executions without leaving a chat window. The MCP protocol gives the AI read-write access to the exchange. That cuts friction for setting up a recurring grid strategy or a price-triggered order. The human-in-the-loop rule means the bot cannot runaway execute. Every trade still needs a finger on the button.
The risk is less about rogue AI and more about the complexity of the prompts. A poorly phrased backtest request could produce a strategy that looks good in a 30-day window but breaks in a volatility spike. Revolut's liability disclaimer covers that ground. The trader owns the output.
Confirming the setup works requires two things. If users report that the conversational interface reduces the time to set up a grid strategy from minutes to seconds, and if the error rate on executed orders stays low, the integration proves useful. A wave of complaints about misaligned backtest parameters or order types that the AI misinterprets would break the thesis. The proof will be in the user forums, not the press release.
Prepared with AlphaScala editorial tooling from the source reporting linked above. Indexable analysis may include a cited Alpha Score value. Publishing checks screen each story before release. Educational coverage, not personalized advice.