Tools for Humanity (TFH), a startup co-founded by Sam Altman, has recently unveiled an initial version of AgentKit, designed to assist commercial websites in confirming that actual humans authorize purchasing decisions made by AI agents.
This development aligns with the increasing prevalence of agentic commerce, where artificial intelligence programmes autonomously browse and execute purchases on platforms. This expansion has sparked significant concerns over fraudulent activities, spam, and large-scale automated abuse.
How AgentKit Functions
AgentKit combines World ID, TFH’s biometric identity credential, with the x402 protocol—a blockchain-based open standard created by Coinbase and Cloudflare. This protocol enables automatic transactions online without constant human interaction.
Users can register their AI agents using their World ID, enabling participating websites to recognize a verified individual’s authorization through the x402 system.
The most secure form of a World ID is created via an iris scan using TFH’s proprietary Orb device. This scan generates an encrypted digital code accessible through the World app. AgentKit necessitates this Orb-derived credential for its highest level of verification.
TFH positions AgentKit as a supplementary feature to x402 v2, developed in collaboration with Coinbase. The company noted that any website already using x402 can now enable proof-of-human verification alongside or instead of micropayments. According to TFH’s chief product officer, the system operates much like granting power of attorney: it confirms a genuine individual is behind the agent’s decisions while allowing websites to still exclude users they deem in bad faith.
Industry Background
The launch of AgentKit comes at a time when agentic infrastructure is gaining momentum. In 2024, both Amazon and Mastercard introduced automated buying capabilities on their platforms, followed by Google’s own protocol to support such transactions. As the ecosystem matures, ensuring human accountability in AI-driven commerce has become central to industry discussions.
TFH envisions World ID as a potential solution for this accountability layer. By linking agent behavior to verified human identities, AgentKit aims to address a critical gap in existing payment and commerce frameworks, which were not originally designed for autonomous AI actors. Additionally, the x402 protocol itself lacks built-in mechanisms to confirm human intent.
Currently in beta, AgentKit is accessible to developers. Feedback from this phase will influence future iterations of the tool. The success of AgentKit hinges partly on the adoption rate of the x402 protocol among commercial platforms and how the industry balances biometric verification against privacy concerns at scale.
