Coinbase is introducing a new tool designed to bridge artificial intelligence and blockchain finance, unveiling what it calls “Agentic Wallets” — wallet infrastructure built specifically for autonomous AI agents.
The product, announced Wednesday by the Coinbase Developer Platform team, aims to give AI-powered systems the ability to hold and move funds independently. In practical terms, that means AI agents can send payments, trade tokens, earn yield, and interact directly with blockchain networks, all within defined security controls.
Giving AI Agents Financial Capability
AI agents have quickly become part of everyday digital workflows. They summarize documents, respond to emails, and assist with research. But according to Coinbase developers Erik Reppel and Josh Nickerson, these systems face a limitation when real-world payments are required.
“AI agents are everywhere,” the developers wrote in a blog post. “But today’s agents hit a wall when they need to actually do something that requires money.”
Agentic Wallets are intended to remove that barrier. The infrastructure allows developers to equip AI agents with crypto wallets that can transact on behalf of users, operating within programmable guardrails. For example, if an agent identifies a better yield opportunity in the middle of the night, it could automatically rebalance funds, provided the user has already approved the relevant permissions.
Built on the x402 Protocol
The new wallet system builds on Coinbase’s x402 protocol, an open-source standard named after the HTTP 402 “Payment Required” status code that was proposed but never implemented. The protocol is designed to support crypto-native payments over the internet without human intervention.
Since launching last year, x402 has processed 50 million transactions, according to Coinbase. A 2.0 version released in December expanded compatibility, including support for more traditional payment rails. In September, Coinbase and Cloudflare also formed the x402 Foundation to encourage broader adoption.
Agentic Wallets also extend Coinbase’s earlier AgentKit toolkit, which allowed developers to embed wallets during the creation of AI agents. Unlike AgentKit, which focused on integration at build time, the new system is positioned as a plug-and-play solution that can “give any agent a wallet.”
Multi-Chain Support and Gasless Transactions
At launch, Agentic Wallets will support Ethereum Virtual Machine (EVM) chains and Solana. The system can also enable gasless transactions on Base, Coinbase’s Layer 2 network, which may reduce friction for developers and users.
A command-line interface lets users monitor agent activity, fund wallets, and deploy additional capabilities through simple commands. Coinbase has also released an “agent-wallet-skills” repository that provides basic transaction functions agents can execute.
coinbaseSecurity Guardrails and Key Protection
Given the financial authority these agents would hold, Coinbase has emphasized built-in safeguards. The wallets include programmable spending limits, session caps, and transaction controls to reduce misuse.
Private keys remain secured within Coinbase’s infrastructure through what the company calls enclave isolation. This setup ensures sensitive credentials are never exposed directly to the AI model or its prompts.
Moving From Advice to Action
The launch reflects a broader shift in how AI systems are evolving. Rather than simply offering suggestions, developers are increasingly building agents capable of executing tasks independently.
“We’re moving from AI agents that advise to agents that act,” Reppel and Nickerson wrote. “From assistants that suggest to helpers that execute.”
Coinbase has also introduced Payments MCP, a protocol that enables large language models such as Anthropic’s Claude and Google’s Gemini to access blockchain wallets, further integrating AI systems with digital assets.
A New Phase for AI and Crypto Integration
By combining AI autonomy with programmable blockchain payments, Coinbase is positioning Agentic Wallets as infrastructure for the next phase of internet services. While adoption and practical use cases will take time to develop, the launch highlights a growing intersection between artificial intelligence and decentralized finance.
As AI tools continue to mature, the ability to securely manage and move funds could mark a significant step toward more autonomous digital systems — ones that don’t just recommend actions, but carry them out.
