Exchange operator Coinbase launched a new tool that allows artificial intelligence agents to execute cryptocurrency trades and make payments directly through users' accounts, deepening its bet that autonomous software will become a major participant in digital asset markets.
The product allows users to connect AI assistants to their Coinbase accounts and authorize them to execute transactions within predefined limits. Coinbase said users can instruct agents to buy and sell cryptocurrencies, rebalance portfolios, place orders based on market conditions and make payments on their behalf.
While professional investors have long relied on automated trading strategies, Coinbase is betting that retail users will increasingly turn to AI assistants such as ChatGPT and Claude to help manage and execute financial transactions. At launch, trading is available across spot and derivatives cryptocurrency markets.
Not alone in AI
Coinbase is not alone among exchanges in exploring AI and automation in trading. Binance, OKX, Bybit and others offer trading bots that can automatically execute strategies such as grid trading, dollar-cost averaging and arbitrage based on user-defined parameters.
The launch also follows a similar move in stocks by trading platform Robinhood, which introduced in May tools allowing AI agents to trade through dedicated customer accounts and make purchases using virtual credit cards. Robinhood has said support for crypto trading may follow.
Infrastructure for agents
Coinbase said the product is part of a broader effort to build infrastructure for what it calls "agentic finance" – the idea that AI agents will eventually be able to carry out financial tasks on behalf of users. The company has previously launched AgentKit, which allows developers to give AI agents control of crypto wallets, and x402, a payments protocol designed to let software and AI systems transact with one another online.
Jesse Pollak, creator of Base, Coinbase's Ethereum Layer 2 network, said during the Consensus conference in May that the industry reached an "inflection point" for AI agents late last year as advances in large language models made it possible for software to handle increasingly complex tasks.
Pollak said the rapid improvement in AI models, combined with the growth of stablecoins, is creating new opportunities for agents to transact and interact with financial systems. "Those things [are] combining with the overall AI trend to make it so that you're starting to see agents be able to do new things with money," he said.
According to Pollak, most activity on x402 currently revolves around trading, access to market data and payments for AI inference services, suggesting early demand is emerging from financial and data-intensive applications.
