How AI agent identity verification works
An agent saying its AgenID is not the same thing as proving it. Seventeen scenarios, one visual language: the same identity object, the same verification handshake, the same point at which a counterparty stops and asks for proof. Every one of them also plays without a portable identity layer, because that comparison is the argument.
These scenarios illustrate the protocol's verification model. The reference deployment at agenid.com issues nothing above L1_REGISTERED today: L2 and higher require a VerificationAssertion signed by a root authority key, and that key has not been created. L5 is reserved by the spec and is not issuable at all.
The challenge–response exchange each scenario shows is a proposed interaction pattern. Protocol v1.1.1 defines the signed objects — manifest, proof, key document, assertion — not this exchange. What you can do today is register an identity and resolve one.
Same agent, same claim, two outcomes. Watch it declare — then watch it prove, and watch the proof get checked.
Today a person is on one end of every transaction. Next, an agent is — often on both ends. Which raises a question the old model never had to answer.
Human → agent → business. The practice resolves the caller’s AgenID before it opens a calendar.
One purchasing agent, three sellers. Each seller resolves the same identity independently before it prices anything.
One agent, four platforms, one persistent identity — and each platform resolves it independently.
One household agent, three plumbers. Each provider resolves the same identity on its own before it holds an arrival window.
A personal agent asks a provider’s support agent to cancel a subscription. Support resolves which agent is asking; the provider still authenticates the account holder.
“I represent ABC Logistics” is a claim. The carrier issues a challenge and makes it a proof.
Agent to agent, no human introduction. Each side resolves the other’s AgenID before a number is spoken.
A buyer’s agent asks a listing brokerage for a showing and the disclosure package. The brokerage resolves who is asking before it opens a calendar or a document.
A candidate’s agent applies and an employer’s screening agent answers. Each side resolves the other before a screen is booked — and the hiring decision stays with the employer.
An outbound voice agent calls a business. Before anyone engages, the business resolves who is calling, who operates the caller, and whether it declared itself an AI.
AgenID establishes identity and proof. The bank still decides, on its own controls, whether the money moves.
A claims agent files for a policyholder. The insurer’s intake agent resolves who is filing before it opens a claim — and still adjudicates the claim itself.
A client company’s agent asks its outside counsel for case documents. The firm resolves who is asking — then privilege, conflicts and release stay the firm’s decision.
A filing agent renews a license at an agency portal. The portal resolves which agent is filing; proofing the resident’s own identity stays a separate process the agency runs.