Artificial intelligence models can now independently execute financial transactions on the blockchain, a significant step made possible by Coinbase’s new “Payments MCP” system.
The leading U.S.-based cryptocurrency exchange announced the launch of the tool, designed to allow advanced language models, such as Anthropic’s Claude and Google’s Gemini, to directly access crypto wallets and conduct transactions using digital assets.
This development signals a major integration between AI and decentralized finance (DeFi), laying the groundwork for what the company terms “agent commerce,” where AI acts without direct human intervention.
Payments MCP, developed by the Coinbase Developer Platform, aims to provide AI models with the same on-chain financial tools available to humans. These include digital wallets, on-ramps, and stablecoin payment capabilities, all accessible via natural language.
Dan Kim, Coinbase’s Vice President of Business Development and Listings, and Erik Reppel, Head of Engineering for the company’s developer platform, stated that crypto payment infrastructures, particularly stablecoins, serve as the “ideal vehicle for autonomous agent commerce.”
The new system builds on ongoing efforts to standardize AI-Blockchain payment methods, including the x402 Foundation, an initiative backed by Coinbase and Cloudflare.
MCP stands for “Model Context Protocol,” which connects AI models to external services like APIs, wallets, and payment networks. It is distinct from Multi-Party Computation (MPC) in cryptography.
Coinbase executives noted that crypto payments operate at “code speed,” allowing for seamless programmatic integration and significantly reducing human friction in transaction processes. They anticipate this will enable language models to natively connect with the crypto economy and payment protocols like x402 for the first time.
The launch aligns with a broader technological trend to grant AI agents financial autonomy. This could allow AI to manage subscriptions, make purchases, or contract digital services on behalf of users, all through cryptocurrencies or stablecoins without intermediaries.
Coinbase highlighted that the system can run locally on a desktop and be customized for individual agents, enhancing user control and reducing reliance on external servers.
While the company did not specify commercial implementation dates, the move is presented as a crucial step toward an ecosystem where AI interacts freely with the decentralized economy. This progression is expected to create both new opportunities and challenges for digital security and regulation worldwide.