Odaily reported that Coinbase has launched “Coinbase for Agents,” a product designed to create dedicated accounts and sub-accounts for AI agents. With these accounts, AI agents can act on behalf of users to conduct trades, manage funds and make payments. The product places AI agents inside an account framework, allowing them to perform selected crypto-related operations after users provide instructions and define the relevant conditions.
Natural-language instructions for trading and payments
According to Coinbase, users can control AI agents through natural-language instructions. The examples given include regularly adjusting an investment portfolio, executing a trading strategy, researching market information, or automatically buying assets within a preset budget. In this setup, users do not need to manually transfer funds between different wallets for every task, because the AI agent can follow the user’s instructions and handle parts of the trading and payment process.
The central design of Coinbase for Agents is the creation of accounts and sub-accounts specifically for AI agents. Rather than requiring users to manage every wallet movement by hand, the structure gives the agent an operational space inside the account system. Based on the user’s instructions, the agent can carry out asset purchases, payments and fund-management tasks. The input describes the product as infrastructure that enables an agent to execute financial actions for a user.
Built on x402 and AgentKit development
Coinbase said the product is built on its AI-related development work over the past year. That includes the previously launched x402 open standard for AI payments and AgentKit. The release of Coinbase for Agents extends Coinbase’s product direction in AI payments and agent tooling, moving those capabilities further into account infrastructure and trade execution.
Gemini, OKX and MetaMask are also exploring financial-account and wallet infrastructure for AI agents. The activity across these platforms shows that the crypto industry is accelerating its layout for “non-human user” scenarios, with AI agents being incorporated into infrastructure for trading, asset management and payments.