Cryptocurrency exchange Gemini has announced the launch of Agentic Trading, a new feature that allows users to connect artificial intelligence models such as ChatGPT and Claude directly to their trading accounts. The tool enables automated market monitoring, trade execution, and risk management based on predefined strategies, representing a significant shift in how retail and institutional traders interact with digital asset markets.
The system is built on the MCP open standard, originally developed by Anthropic, which allows AI agents to interact with external tools and APIs. Gemini said it has integrated its full trading API into this framework, enabling AI systems to execute trades without manual intervention. The exchange described the launch as a first for regulated platforms, calling it “the first agentic trading tool to be available directly through a regulated US-based exchange.”
The product introduces modular functions known as Trading Skills, which allow AI systems to perform specific tasks including querying bid-ask spreads, retrieving historical price data, and supporting pattern recognition and backtesting. Users can delegate execution while retaining control over strategy inputs. The model separates decision-making from execution, with AI handling timing, order placement, and monitoring. “We believe we're at the beginning of a fundamental shift in how people interact with financial markets,” Gemini stated. “Agentic trading isn't just a feature. It's a new paradigm where AI handles the execution, patterns, and discipline, while you focus on strategy and goals.”
Unlike traditional algorithmic trading which requires custom coding, Agentic Trading uses natural language instructions. For example, a user might tell an AI: “Buy Bitcoin when its price drops 5% in an hour.” The AI then interprets the command, monitors the market, and executes the trade via the Gemini API. This lowers the barrier to entry for non-programmers.
Gemini emphasizes that the AI cannot withdraw funds or change account settings without user approval. The exchange implements multiple layers of authentication, including API keys and OAuth tokens. Each AI request is logged, providing an audit trail, and daily trading volumes for AI-driven accounts are capped to mitigate risk.
Broader industry initiatives are building similar infrastructure. Coinbase has supported the x402 protocol, an open payments standard that allows AI systems to access crypto wallets and services. Tempo is also developing the Machine Payments Protocol, which focuses on machine-to-machine financial transactions. Both systems rely on the same MCP standard used by Gemini, suggesting that AI-native financial infrastructure is forming around shared interoperability layers.
The launch comes as Gemini faces internal and market pressures, including recent departures of key executives and weak trading activity. The rollout of Agentic Trading points to a strategy focused on differentiation through technology rather than competing solely on fees or liquidity.
