Coinbase has rolled out new artificial intelligence tools that can manage trading and asset operations on behalf of users. Its newly announced platform, Coinbase for Agents, enables AI models like ChatGPT and Claude to connect directly with users’ Coinbase accounts for hands-free portfolio activity and trade execution.
Direct account control with AI tools
AI agents linked via the platform can execute trades, manage users’ portfolios, and apply pre-set investment strategies as instructed. The system is designed to function autonomously, operating without ongoing human input. Coinbase has made this structure available both through the secure Model Context Protocol, ensuring safe interactions, and via a command-line interface tailored for developers building autonomous applications.
Coinbase for Agents was created to allow AI models to connect with user accounts, handle trades, manage portfolios, and implement defined investment approaches.
Coinbase, known as one of the largest US-based cryptocurrency exchanges, stands out for serving both individual investors and institutional clients with a broad array of services.
AI-powered payment protocol integrated
In addition, the company has now integrated an AI payment protocol called x402 into the platform. With x402, AI agents can independently make payments for services such as data access and information gathering. This empowers the agents to autonomously purchase resources required for market data, research tools, or trading strategies, without user supervision.
Mini glossary: The Model Context Protocol is a framework designed to let AI systems communicate with external applications and data sources more securely and consistently. Meanwhile, x402 is a protocol focused on enabling AI agents to make automatic payments for digital services.
Supporters of this technology believe autonomous agents could prove highly effective, especially for high-frequency micro trades and around-the-clock investment strategies. As AI models continue to advance, some investors seem increasingly open to entrusting parts of their portfolios to such systems.
Coinbase Advisor unveiled
Coinbase has also introduced a new feature named Coinbase Advisor, now integrated into its app. According to the company, this tool operates as a registered financial advisor with both the US Securities and Exchange Commission and Commodity Futures Trading Commission. The service is designed to offer investment recommendations and automate portfolio management tasks for users.
In a demo shared by Coinbase, an AI agent was tasked with running a systematic purchase strategy for Ethereum. In this scenario, the agent analyzes historical price data to identify times during the day when Ethereum trades at relatively lower prices, then sets up an automated buying plan accordingly.
Studies highlight caution on AI in crypto
Despite strong interest in AI agents, there is still skepticism about their effectiveness and the extent of their autonomy. Recent research conducted by Pantera Capital, Stanford University, Ava Labs, and the Initiative for Cryptocurrencies and Contracts suggests that many AI-driven crypto projects might not yet meet expectations.
The study, which analyzed data from over 925,000 token holders, found that while AI agent treasuries showed roughly $30 million in paper gains, investors as a whole lost nearly $192 million.
Researchers noted limited evidence that most projects engage in truly autonomous operations. Findings indicate that many initiatives may still rely on basic API integrations rather than fully-fledged decision-making systems. This points to a crypto sector where AI agents are gaining traction but where the underlying technology remains at a relatively early stage.
