Coinbase Unveils Tool Enabling AI Agents to Trade Crypto and Make Payments

Coinbase for Agents will enable exchange users to oversee their crypto portfolios without requiring continuous manual monitoring, while allowing a variety of tasks to be carried out autonomously on their behalf.

Crypto exchange Coinbase has introduced a new platform that enables artificial intelligence agents to execute payments and conduct cryptocurrency trades for users, as firms across the digital asset sector seek to capitalize on growing enthusiasm surrounding AI technologies.

Coinbase announced on Thursday the launch of Coinbase for Agents, a service that enables AI systems such as ChatGPT and Claude to link with a user’s exchange account and receive instructions to place trades or carry out specific investment strategies.

AI agents can also use Coinbase’s x402 AI payments protocol to complete transactions for data services, enabling them to obtain market information and execute trading strategies independently without requiring direct human involvement.

Crypto firms have increasingly promoted the technology as a solution for handling the rapid, small-scale transactions commonly performed by AI agents. As more advanced AI models have emerged, these agents have gained wider adoption, with a growing number of traders relying on them to independently implement trading strategies.

Coinbase stated that the platform can be accessed through a model context protocol (MCP), which enables AI models to link directly with user accounts, as well as through a command-line interface designed for developers.

The company also unveiled Coinbase Advisor, an AI-powered assistant embedded within its application. According to Coinbase, the tool operates as a financial adviser registered with both the US Securities and Exchange Commission and the Commodity Futures Trading Commission, enabling it to provide guidance on trading decisions.

Coinbase said the feature can assist users in managing their digital asset portfolios without requiring continuous manual supervision. The tool can also handle activities such as directing funds into rewards programs and executing recurring cryptocurrency purchases.

“Imagine you want to dollar-cost average into ETH at the optimal time of day. Just tell your agent your target and timeframe. It can pull 30 days of hourly price data to identify when ETH historically trades lowest, set a recurring $20 market buy at that time, and schedule it to run daily for the next two weeks,” the company explained.

Although many firms are encouraging investors to adopt AI-driven tools, a study released last month showed that users of AI agents are experiencing financial losses, while the agents themselves may not be operating as independently as often claimed.

Researchers from Pantera Capital, Stanford University, Ava Labs, and the Initiative for Cryptocurrencies and Contracts analyzed more than 925,000 token holders and discovered that agent treasuries recorded paper gains of roughly $30 million, while token holders as a group suffered losses totaling $191.7 million.

The study also concluded that many of the projects reviewed still lack convincing proof of autonomous trade execution, with a significant portion of them relying primarily on basic API integrations rather than fully independent agent-driven systems.

Coinbase has become the latest company to wager that AI agents will engage with and conduct transactions across a wide range of connected services and platforms.

Last month, stablecoin issuer Circle introduced a suite of tools that allows AI agents to access wallets, locate services, and execute programmable payments using its stablecoin. Circle CEO Jeremy Allaire has forecast that billions of AI agents will adopt stablecoins for transactions over the next five years.

Earlier this month, Crossmint, a provider of stablecoin and wallet infrastructure, unveiled a new service that allows AI agents to complete payments through supported Visa credit and debit cards.

In a report released in May, crypto investment firm Keyrock stated that AI agents had rapidly built a mature ecosystem, processing $73 million in value through approximately 176 million transactions between May 2025 and April 2026.