The interaction method is primarily chat-driven: the AI agent suggests an operation, then opens the Base wallet window for the user to confirm or cancel. Importantly, the agent can never access private keys, and all suggested operations must receive explicit approval from the user. Base emphasizes that each transaction follows the same review process as any Base account request, simulating asset changes before confirmation. This balance between automation and user control is a core design principle for AI agents handling more on-chain tasks.
This vision faces a market that is still in its infancy. According to data tracked by x402scan, the x402 protocol processed approximately $1.1 million in transaction volume over the past 30 days, reminding us that the theory of “agent payments” is still in its early stages, despite attracting the attention of investors and developers. The broader ecosystem already includes integrations with DeFi News protocols such as Morpho, Moonwell, Uniswap, Aerodrome, Avantis, Bankr, and Virtuals, all of which can be accessed through AI-driven prompts.
Key Points
The MCP workflow adapts to the evolving AI-on-Chain space.
Essentially, the Base MCP acts as an intermediary layer, transforming natural language prompts into specific blockchain operations while maintaining human oversight. Users initiate operations in the AI chat, and the AI subsequently proposes a suggested action. Users see a clear confirmation step in the Base wallet window, with asset changes simulated before any real fund transfers occur. This design ensures security while making interactions with DeFi News protocols smoother through the AI agent.
Base's approach also emphasizes continuity in user experience. Lincoln Murr points out that the MCP and similar integrations ensure that users' Base accounts remain portable and synchronized across various interfaces, whether interacting directly within the Base app or through an AI assistant. This continuity aims to eliminate friction that could hinder the adoption of agent payments, especially when users need to switch between different tools to manage their portfolios.
The MCP initiative progresses in parallel with Coinbase's broader x402 standard, a framework designed to enable AI-assisted payments securely and at scale. As the MCP matures, developers and users may see more seamless cross-protocol transactions, leveraging AI agents to navigate liquidity pools, governance actions, and asset transfers through a single chat workflow. However, even with such promise, the road to widespread adoption remains gradual, as the $1.1 million transaction volume over 30 days for x402 indicates that the market is still in its early stages.
Risks, Unresolved Issues, and Next Steps
Security remains a major theme in discussions about AI agents in the crypto space. Research consensus suggests that AI agents can be viewed as “untrusted system components,” highlighting the need for a clear separation between commands and data, as well as robust validation checks before executing any on-chain operations. As AI models gain more capabilities,