World, co-founded by Sam Altman, focuses on developing what it refers to as “proof of human” technology — identification verification solutions for an internet increasingly filled with AI-generated content of questionable quality. It’s notable that Altman’s other enterprise, OpenAI, has been frequently criticized for generating a significant amount of that content (though it’s arguable that he anticipated these issues when he established World).

This week, Tools for Humanity (TFH), the organization behind World, launched the beta version of a new verification tool — tailored to enhance the growth of agentic commerce, the rapidly expanding practice of utilizing AI programs to navigate the internet and make purchases on behalf of users.

An increasing number of consumers are opting for AI agents to browse websites and purchase items for them. While this trend offers a degree of automated convenience, it has also brought to light new potential for fraud, spam, and various forms of widespread internet misuse.

On Tuesday, World unveiled its claimed solution: AgentKit, a software development tool aimed at commercial websites that facilitates the integration of a new verification method that enables those platforms to confirm a genuine human is behind an agent’s purchasing actions.

AgentKit is based on World ID, which serves as the cornerstone of TFH’s verification framework. The most secure iteration of the ID is created from a scan of a user’s eyes through World’s Orb device. The Orb transforms an iris into a distinct and encrypted digital identifier — the authenticated World ID — which can subsequently be utilized to access TFH’s suite of services through the company’s World app.

AgentKit permits a user’s World ID to be incorporated into a newly introduced payment framework known as the x402 protocol. Developed by Coinbase and Cloudflare, x402 is a blockchain-driven open standard that enables automated computer systems to interact with one another directly online — without needing human oversight at each stage. To utilize AgentKit, users only need to register their AI agents with their World ID, which then informs websites — via the x402 system — that a specific verified human approves the agent’s buying choices.

“AgentKit is designed as a supplementary extension to the x402 v2 protocol, in collaboration with Coinbase,” Tools for Humanity stated in a press release. “The integration is intended so that any site already utilizing x402 can facilitate proof of unique human verification alongside (or in lieu of) micropayments.”

In a discussion with TechCrunch, TFH Chief Product Officer Tiago Sada likened the new feature to assigning “power of attorney” to an agent. By confirming that the AI program represents a specific user, a website can determine whether to trust the transactions initiated by those agents, Sada explained. “What the World ID credential indicates is that an individual is a genuine and distinct human,” he remarked, noting that websites still retain the option to block certain users they suspect of acting in bad faith.

AgentKit is currently being made available in beta to developers, with the expectation that feedback will enhance it over time. Sada also emphasized that consumers will need to have a validated World ID, obtained from an Orb scan, to be eligible for this level of verification.

This is a timely initiative. Leading e-commerce platforms and financial services have already begun to adopt agentic commerce. Last year, firms such as Amazon and Mastercard rolled out automated purchasing functionalities on their platforms, while Google recently introduced its own protocol aimed at supporting this trend. As the sector expands, the industry will undoubtedly seek mechanisms that guarantee it remains trustworthy and stable. World is evidently striving to establish itself as the primary provider of that stability.