To address the trust gap in the current AI agent economy, Coinbase and World ID have jointly launched a test version of AgentKit, aimed at linking AI agents with verified real human identities. This innovative initiative marks a significant step towards greater accountability and credibility in the AI agent economy.
Building the Trust Foundation for the AI Agent Economy
In an increasingly digital business and service landscape, AI agents are playing a more prominent role. However, platforms still face challenges in distinguishing beneficial AI agents from malicious bots. Traditional security measures often treat all automated traffic as potential threats, which not only stifles innovation but also hinders many compliant AI application scenarios.
AgentKit introduces a new mechanism called “proof of unique human.” Through World ID, developers can bind AI agents to their verified human identities. This way, platforms can be assured that the automated actions are backed by real individuals. More importantly, the system effectively protects user privacy while achieving identity verification, preventing the leakage of sensitive personal data.
Industry data shows the immense potential in this field. According to McKinsey, the scale of business transactions driven by AI agents is expected to reach $3 trillion to $5 trillion by 2030. Bain & Company estimates that during the same period, AI agents will account for 25% of U.S. e-commerce.
Integration of AgentKit with the x402 Protocol
AgentKit operates in conjunction with the x402 protocol developed by Coinbase and Cloudflare. This protocol has processed over 100 million microtransactions in APIs and in-app payments in just six months, providing a solid payment foundation for AI interactions. However, mere payment functionality does not solve the identity verification challenge. A user can easily deploy thousands of agents, circumventing existing security measures.
AgentKit was created to address this pain point by binding multiple AI agents to a single verified real human identity. This enables platforms to effectively identify collusion attacks and implement appropriate restrictions.
Erik Reppel from Coinbase clearly articulated this: “In AI agent-driven business activities, payment is the ‘how’ of transactions, while identity is the ‘who’ conducting the transaction. By integrating World ID with the x402 protocol, developers now have a complete trust solution: AI agents can pay as needed, while the platform can verify that the wallet behind it indeed belongs to a real human. This is an important step towards a new era of the web, where AI agents are no longer just seen as automated traffic but as legitimate economic participants.”
Expanding Real-World Application Scenarios
AgentKit brings new possibilities across various industries. For instance, in booking platforms, it can prevent malicious hoarding of reservations by limiting the number of accesses per verified human. Ticketing services can effectively combat scalping through this mechanism without raising ticket prices.
Moreover, platforms can track usage based on each individual rather than each wallet, providing fair free trial opportunities while reducing abuse and ensuring balanced resource distribution. Telecom operators can also assign a phone number to each verified user, even if that user communicates through multiple agents.
Currently, World ID has nearly 18 million verified users across more than 160 countries. Such a vast user base further solidifies its position as a global identity verification layer. With the accelerated adoption of AI technology, identity verification systems like AgentKit are expected to become a key component of future online trust mechanisms.