Agentic identity
Identity for AI agents on the agentic web.
Agentic identity is the verifiable identity an AI agent uses to prove who it is, who operates it, and what it is authorized to do, before it acts, buys, or shares data on your behalf. Every interaction turns on three questions: is the agent real and who runs it, did the person agree, and is it allowed to act.
What does an AI agent need to prove?
When an agent acts on your behalf across systems that have never met it, three things have to be established before any data moves.
The agent is cryptographically bound to the person or organization it works for, so any counterparty can confirm who stands behind it.
The person’s approval travels with the request as signed, checkable proof, verified on arrival rather than taken on trust after the fact.
The agent carries scoped, revocable permission that traces back to a human mandate and can be narrowed at each step or withdrawn.
Answering these three questions before any data moves is what agentic identity provides.
Why can’t AI agents use API keys or tokens?
Today, most agents authenticate with API keys or bearer tokens. A token proves only that whoever holds it possesses a secret string. It says nothing about who the holder is or what they are allowed to do, and any copy of it works as well as the original. The moment an agent crosses from one organization to another, there is no shared login or directory to fall back on, and a token cannot establish identity, consent, or authority.
A bearer / access token
Proves possession of a string
- Says nothing about who holds it
- Any stolen copy works the same
- Breaks across organizational boundaries
A verifiable credential
Proves who issued it, and what it says
- Cryptographically bound to a real identity
- Tamper-evident and can’t be replayed
- Portable across organizations with no prior integration
A token proves possession of a secret. A credential proves identity.
How do verifiable credentials give AI agents an identity?
A verifiable credential is a signed, portable digital identity the agent presents and anyone can verify, without calling back to whoever issued it. It proves two things anywhere it is presented: who issued it, and that its contents have not been altered. Two agents that have never met, in organizations with no prior relationship, can establish trust on first contact, with no shared login and no integration arranged in advance. Verification runs in both directions: each party verifies the other before any data moves.
What a credential does that a token cannot:
- Carry authenticated, tamper-evident data signed by its issuer.
- Carry consent you can audit, checked cryptographically on arrival.
- Carry delegated, revocable authority that traces to a human mandate.
- Reveal only what is needed, proving a fact without exposing the data behind it.
- Establish cross-organization trust with no bespoke integration.
Who is building the standards for AI agent identity?
The standards bodies, and now regulators, are converging on verifiable credentials for agent identity. This is the emerging consensus for how AI agents will prove who they are across organizations.
Updated quarterly as the standards evolve.
How does Indicio approach agentic identity?
Most of what the market calls agentic governance watches agents at runtime, inside an organization that already trusts them. Indicio provides the layer beneath that: verifiable identity, delegated authority, and auditable consent that let an agent prove who it is, who runs it, and what it may do, to a party that has never met it, before any data moves.
This is not theory. Indicio has already shown an AI agent verify a real person’s credential before acting, building on Indicio Proven, our award-winning, analyst-recognized platform already in production for human and organizational identity. The roadmap extends the same model to authority held at the edge, on the user’s device, and to the mediation that lets an agent reach a person whose device is offline.
Agentic identity, answered
What is agentic identity?
Agentic identity is the verifiable identity an AI agent uses to prove who it is, who operates it, and what it is authorized to do, before it acts, buys, or shares data on someone’s behalf. It answers three questions at every step: identity (is the agent real and who runs it), consent (did the person agree), and authority (is it allowed to act).
Why can’t AI agents use API keys or tokens?
A token or API key is a shared secret. It proves only that whoever holds it possesses a string, not who they are or what they may do, and any stolen copy works as well as the original. It also breaks down the moment an agent crosses from one organization to another, where there is no shared login or directory to fall back on.
What is a verifiable credential for an AI agent?
A verifiable credential is a signed, portable digital identity the agent presents and anyone can verify without contacting the issuer. It proves who issued it and that its contents have not been altered, so two agents that have never met can establish trust on first contact.
What is the difference between authentication and delegated authority for agents?
Authentication proves what an agent is. Delegated authority establishes what it may do, for whom, and for how long. Knowing an agent’s identity does not establish its permission to act, so both are required. A credential can carry scoped, revocable authority that traces back to a human mandate.
How do two AI agents that have never met establish trust?
Each agent presents a verifiable credential, and each verifies the issuer of the other’s credential against a machine-readable trust list before any data is shared. Trust comes from the issuer’s signature, so no shared login, shared directory, or prior integration is needed.
What is a mandate in agentic commerce?
A mandate is a one-time, cryptographically signed authorization describing a specific action, its amount, the counterparty, and the transaction it belongs to. Because the details are bound into the signature, a mandate works only for that action and cannot be replayed against another.
Who is standardizing AI agent identity?
W3C, the OpenID Foundation, the IETF, NIST, and the Cloud Security Alliance are converging on decentralized identifiers and verifiable credentials for agent identity. Google’s Agent Payments Protocol carries mandates as verifiable credentials, and Singapore’s IMDA published the first governance framework dedicated to agentic AI.
The full argument, in one paper
Identity, Commerce, and the Agentic Web lays out how verifiable credentials become the trust layer for AI agents: mutual verification, delegated authority, privacy-preserving proofs, authority at the edge, and the economics of portable trust.
About Indicio
Indicio makes proving identity simple. An award-winning, analyst-recognized company, Indicio builds verifiable credential and decentralized identity infrastructure that is in production at national scale across travel, banking, hiring, and government. Indicio Proven® and Proven AI give people, organizations, and AI agents a verifiable identity they can use anywhere, without a central gatekeeper.
