Proof of Humanity network World has made its first foray into the agentic web by releasing AgentKit, a framework for AI agent deployment backed by real users. The beta release of the developer toolkit gives agents cryptographic proof that they’re being operated by a real human – and its use cases are manifold.
WorldKit has been created using the x402 standard devised by Coinbase and Cloudflare, an open protocol that has powerful applications in the context of agent deployment. With WorldKit now integrated with it, developers are free to create applications that provide a provable link between man (or woman) and machine. It’s a breakthrough in making the agentic web a working reality for millions of onchain users.
World ID meets AI
World is famed for its World ID which enables web users to anonymously prove their identity without giving up personal information. One of the reasons it was developed was to provide proof of humanity in a web swimming with bots and agents. But in releasing WorldKit, World is proving that it doesn’t have a problem with agents: merely their lack of verifiability, which makes it difficult to tell whether or not they’re being operated by a unique human.
It’s easy to see why this capability will appeal to the developers of apps running AI agents and the real users interacting with them. Now, app developers can determine that their users are unique, distinct entities, while individuals can participate in the agentic web without disclosing sensitive data. It’s a solution that effectively balances the need for privacy with the need for accountability and it’s poised to drive further adoption of AI agents by streamlining access.
AgentKit in action
The core problem AgentKit has been designed to address concerns the fact that payments alone can’t establish trust in an agent-driven internet. The x402 protocol enables AI agents to pay for access but doesn’t determine whether those agents represent real individuals or coordinated automation.
AgentKit introduces a second layer to this stack by binding identity to execution. This allows a verified individual to cryptographically delegate their World ID to an AI agent, which still operates independently but carries a verifiable proof that it’s acting on behalf of a unique human – without exposing who that person is.
Initially available in beta, AgentKit preserves the efficiency of autonomous agents while reintroducing scarcity and fairness at the human level. It’s effectively the first viable solution to align machine-scale execution with human-scale accountability and has the potential to foster a web where agents and individuals can co-exist, without forcing app developers to guess whether their users are real or farms being operated by a single entity.
There’s also a lot more that comes bundled with AgentKit, though, such as the ability to support age- or country-based services, using zero knowledge proofs that keep the user’s personal information private. Given that World now has more than 18 million verified humans on its network in over 160 countries, the potential for WorldKit is huge.
But first, developers will need to familiarize themselves with the toolkit and discover the ways in which it can be used to add an authenticity layer to the x402 protocol. Once that’s done, an array of applications that intelligently leverage the technology will inevitably follow.
Featured Image by World
