Gemini, a U.S.-regulated cryptocurrency exchange, recently unveiled Gemini Agentic Trading—a pioneering agentic trading tool now available through the platform.

Accessible via Gemini’s Developer Platform, this innovative solution allows traders and developers to harness AI agents like Claude, ChatGPT, and others by integrating them with Gemini’s exchange through the Model Context Protocol (MCP). These agents can autonomously place trades, monitor markets, and manage risk on behalf of their users.

The tool leverages the full Gemini API capabilities via MCP. Thus, any action currently accessible through the API is available to a connected AI agent. Users have the flexibility to connect existing AI tools without needing to reconstruct their setup and can define trading strategies using natural language, which are then executed by the AI agent.

Trading Skills and User Accessibility

Gemini has launched Trading Skills—modular, pre-built functions that AI agents can invoke to perform specific tasks. Currently available Skills include fetching real-time prices and order book depth, querying bid-ask spreads for trading pairs, and accessing historical OHLCV candlestick data for pattern recognition and backtesting.

Users can instruct their AI agent using natural language commands, such as placing a sell order when Bitcoin reaches a specified price or when the bid-ask spread hits a defined threshold. The appropriate Skill is then utilized to execute these instructions.

This platform aims to support both experienced quantitative traders who are building complex, multi-leg strategies and novice users interested in starting with pre-built Skills and gradually increasing their complexity over time. Gemini plans to expand the Skills library to encompass a broader range of functions for more comprehensive agentic trading.