World Introduces AgentKit to Link AI Agents With Verified Human Identity

A new layer of trust for AI-driven interactions

World has introduced AgentKit in beta, a developer toolkit designed to connect AI agents with verified human identities. Built as an extension of the x402 protocol, an open standard initiated by Coinbase and Cloudflare, the toolkit enables AI agents to interact with websites, APIs, and online services while proving that a real person stands behind them. The verification layer relies on World ID, part of the Worldcoin ecosystem, which allows users to confirm their uniqueness without sharing personal data.

The launch comes as AI agents increasingly take on tasks traditionally handled by humans, such as booking reservations or comparing prices across platforms. Industry forecasts suggest that agent-driven commerce could grow to between $3 trillion and $5 trillion globally by 2030, with AI agents potentially accounting for up to a quarter of U.S. e-commerce activity.

As these agents become more active participants in digital economies, platforms face a key issue: distinguishing legitimate users from automated or coordinated bot activity. AgentKit attempts to address this by introducing a cryptographic “proof of human” layer through World ID.

Developers can use the toolkit to build what the company calls human-backed agents—systems that verify a unique individual is behind an agent without exposing personal data. This approach relies on privacy-preserving cryptographic methods rather than traditional identity checks.

The system complements existing mechanisms like micropayments. The x402 protocol already allows agents to pay for access to services, with more than 100 million transactions processed since its launch in 2025. However, payments alone do not confirm whether activity comes from unique individuals or coordinated networks.

AgentKit adds this missing layer by enabling platforms to request proof of a unique human alongside—or instead of—payment. Users with a verified World ID can assign that identity to one or more agents, while platforms can still recognize that those agents originate from the same person.

Managing access while preserving privacy

The model introduces practical use cases for platforms dealing with automated traffic. Reservation systems, for example, could allow AI agents to book tables while limiting large-scale abuse from automated scalping tools. Similarly, ticketing services could prioritize purchases tied to verified individuals rather than anonymous bots.

Other applications include managing free trials or API access based on the number of unique users rather than the number of accounts or transactions. This could help platforms allocate resources more effectively while reducing misuse.

AgentKit also introduces support for additional identity signals, such as age or location, using zero-knowledge proofs. These methods allow platforms to verify specific attributes without accessing or storing sensitive personal information.

World reports that its identity network includes nearly 18 million verified individuals across more than 160 countries. The current AgentKit beta builds on this infrastructure, with plans for a more advanced version tied to the next iteration of the World ID protocol.

The bottom line

AgentKit reflects a broader shift toward making AI agents more accountable participants in online ecosystems. By combining payment infrastructure with privacy-focused identity verification, the toolkit offers a framework for distinguishing between automated activity and real user representation. As agent-driven interactions continue to grow, systems that balance trust, scalability, and privacy are likely to play a central role in how digital services evolve.