Scope
Useful signals identify who made a claim, what it covers, when it applies and how it can be checked or revoked. Structured data can make these questions easier to process; structure alone cannot make a claim true. A receiving agent still needs an appropriate trust policy, current evidence and a way to handle missing or contradictory information.
Illustrative example
A booking service exposes cancellation conditions, the source of an availability claim and the time at which that availability was checked. An agent can compare those details with a traveller’s constraints before proposing a purchase. If the evidence is stale or a condition is ambiguous, it asks for confirmation rather than treating a polished offer as proof.
Terms and contribution
The phrase is used here as a design framing, not the name of a single technical standard. Established mechanisms can support parts of it: the W3C Verifiable Credentials Data Model 2.0, for example, distinguishes issuers, holders and verifiers. Verifying a credential’s provenance does not by itself settle whether its issuer or claims should be trusted.
Read in this order
- Inverting the interface — Start with the argument that services must express value and conditions to user-side agents.
- From Interface Craft to Governable Delegation — Connect machine-readable meaning to authority, evidence and the ability to recover.
- Designing for agency — Examine how evidence and policy fit into the supervision of agentic work.