Gemini Launches Agentic Trading for AI-Powered Crypto Execution

Crypto exchange Gemini has launched Agentic Trading, marking the first time a regulated U.S. exchange has offered direct AI agent integration for automated trading execution. The platform operates through the Model Context Protocol (MCP), an open standard that provides AI agents with direct API access to execute actions on behalf of users.

According to the company's official blog post dated April 27, 2026, Gemini integrated its entire trading API with MCP, enabling AI models to access all exchange features without rebuilding existing setups. The system includes modular Trading Skills—pre-built functions that AI agents call to perform specific tasks. At launch, three modules are available: Get Market Data for real-time price information, Find the Spread for bid-ask analysis, and Retrieve Candles for historical price data.

Traders can connect any MCP-compatible AI model, including Anthropic's Claude and OpenAI's ChatGPT, to execute strategies ranging from basic buy and sell orders to complex multi-leg positions. The exchange positioned the launch as part of a broader transformation in financial market interaction. "We believe we're at the beginning of a fundamental shift in how people interact with financial markets," Gemini wrote in a blog post. "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."

Independent reporting from Yahoo Finance corroborates the timeline and scope of the changes. The article notes that Gemini's stock rose about 0.25% on the day of the announcement, trading at $4.40, though shares have fallen more than 55% since the start of the year.

Accessing the feature requires navigating to the Gemini Developer's Platform at developer.gemini.com and selecting "Agentic" under the dropdown menu. From there, users configure their AI agent connection through MCP. The physical experience involves clicking through developer documentation, pasting API keys, and waiting for authentication tokens to generate (a process that takes about 30 seconds if your credentials are already cached). Once connected, the agent can issue commands like "sell BTC when price reaches $100k" or "buy when bid-ask spread reaches 0.01%."

This launch comes shortly after Gemini announced significant operational changes. In February, the exchange said it would slash 25% of its workforce as it streamlined operations, abandoning its businesses in the European Union, United Kingdom, and Australia to sharpen its focus on the United States. The exchange said it would boost its use of AI to become more efficient with a smaller team. The timing suggests agentic trading isn't just a product feature—it's part of a cost-cutting strategy that replaces human oversight with automated systems.

Other protocols are building similar bridges between AI and crypto infrastructure. The x402 protocol, incubated by Coinbase and now under the Linux Foundation, provides AI bots access to crypto wallets and tools. Meanwhile, the Machine Payments Protocol—developed by the Stripe-backed Tempo network—enables automated machine-to-machine payments. Gemini's approach differs by integrating directly into a regulated exchange rather than building a separate protocol layer.

Experienced quant traders can build fully custom agents, chain together Skills, and implement sophisticated multi-leg strategies with fine-grained control. Traders who are newer to automation can start by connecting their preferred AI tool via MCP, use pre-built Skills to handle data retrieval and risk management, and scale up complexity as they get comfortable. The floor is low. The ceiling is high.

Regulatory implications remain unclear. While Gemini operates as a regulated U.S. exchange, the question of liability when an AI agent executes a flawed strategy hasn't been addressed in the documentation. If an agent misreads market data and executes a losing trade, who bears responsibility—the user, the AI model provider, or the exchange? (This is the kind of question that keeps compliance teams awake at night.)

The official documentation from Gemini's blog states the company will expand the Skills library to enable more diverse trading capabilities. However, no timeline was provided for additional modules beyond the initial three.

Whether users actually pay for this capability remains the real question. The feature is free to access through the developer platform, but trading fees still apply to executed orders. For retail traders, the value proposition depends on whether AI agents can consistently outperform manual trading after accounting for those costs. For institutions, the appeal lies in automation at scale—running multiple strategies simultaneously without human intervention.

As of April 26, 2026, Gemini will close all customer accounts in the European Economic Area. This regional exit coincides with the agentic trading launch, suggesting the company is consolidating resources around its U.S. operations and AI-driven automation. The broader industry impact depends on whether other regulated exchanges follow suit or if this remains a Gemini-specific differentiator.

Artūras Malašauskas is an AI Systems Integrator with 20+ years of production-grade web engineering experience. He has designed, shipped, and scaled enterprise Python/PHP systems for logistics, SaaS, and public-sector clients. For the past year, he has focused exclusively on AI integrations: deploying open-source LLMs, building generative media pipelines (image, audio, video), and engineering multi-agent workflows for real production environments. His standard: reproducibility, security, cost-efficient inference—no vaporware. He documents and evaluates emerging AI tooling, separating verified capabilities from marketing noise. Technical editor at: muza-ai.eu, ai-verslas.lt, ai-naujinos.lt Connect on LinkedIn