Agent

OpenAI's Agents API turns the Codex loop into a hosted API: sessions, compaction, tools, sandboxes, plugins. That layer is real. It is still not a computer with identity, durable state, and an owner who can walk away with it. ARC attaches at the seam they already documented.

Many companies are building compute for agents: sandboxes, VMs, even digital workers. They answer where untrusted code should run. What agents still lack is a machine that can last: identity, state, constrained authority, verifiable history, and the right to leave. We call that machine ARC.

AI services already use credits to meter storage, compute, and API calls. As Agents begin to act across systems, the important work is not giving those units a fashionable name. It is making the rights, authority, and usage they represent possible to verify and reconcile.

Payment is the most visible connection between AI Agents and blockchains, but the more fundamental questions are: who is the Agent, who authorized it, what can it do, and how do we reconcile what happened when something goes wrong?

An agent cannot exist only as a platform account and an API key. DID can provide a portable, long-lived identity anchor, but authorization, runtime proof, revocation, and audit still have to do their own jobs.

AFS has a design principle we call Small World: every agent, every app, every process sees not the global world, but a small world projected specifically for it. Anything outside that world isn't "access denied". It simply doesn't exist. Here's where the idea comes from, why it's more than a security mechanism, and how to try it yourself.

The runtime for Blocklets. It gives a developer a place to run an application described as a Blocklet, together with the resources that Blocklet declares it needs.

A DID is a digital identifier. It helps its holder and a verifier establish who controls an identity and who a claim relates to. It is not a wallet, and it does not replace application accounts, sign-in methods, or permission rules.

A verifiable credential is a signed statement from an issuer about a subject. It gives a verifier something more specific to inspect than a screenshot or an assertion passed along without its source.