AI & agents

AI agent identity and authorization

Give AI agents credentials that name their operator, their scopes and an expiry.

Use case: Agent identity
Wallet data: Agent credential
Authbound capability: Credential Issuance
Flow: Issue

Try the live demo Talk to Authbound

What changes

Before and after the wallet

Today

Shared API key
User agent string
Block or trust all
Access to your API

With the EUDI Wallet

Agent presents credential
Authbound verifies
Operator and scopes
Access to your API

The wallet changes how the evidence is collected. The step at the end stays yours.

The problem

Why this is worth changing

For the business: You either block all automation or trust any agent holding a leaked key.
For the user: Legitimate agents get blocked by bot rules written for scrapers.
For operations: API keys spread across repos and teams, and revoking one means rotating everywhere.

Why the wallet helps

Issue agents credentials the way you issue them to people. An agent credential names the operating organisation, the scopes it may use, and an expiry, signed by the issuer and checkable by any service it contacts. Access then rests on attested authority instead of a shared secret that leaked into a repo two years ago.

Agent traffic becomes attributable to a named operator
Scopes and expiry enforced by the credential, not by a table somebody forgot to update
Revoking a compromised agent is one revocation, not a key rotation everywhere

User journey

What actually happens

01
The operator is issued a credential for the agent
02
The agent calls your API
03
It presents the credential instead of a key
04
Authbound checks the issuer signature
05
Authbound checks revocation and expiry
06
Your backend receives operator and scopes
07
You grant exactly those scopes

How it works

Three steps with Authbound

Step 01
Issue the credential
The operating organisation issues a credential naming the agent, its owner, its scopes and its expiry.
Step 02
Present it per request
The agent presents the credential when calling your API instead of a bearer key.
Step 03
Authorize on attributes
You validate the signature, check revocation, and grant exactly the scopes the credential carries.

Technical flow

AI agent
Your API
Authbound
Signature, revocation, expiry
Your backend

Who does what

What Authbound handles, what you handle

Authbound handles

The wallet request, with QR on desktop and a deep link on mobile
Verifying the presentation the wallet returns
Checking the issuer against the EU trust lists
Revocation and device binding checks
A structured result delivered to your backend
An audit record of every verification

Your application handles

Choosing which attributes the flow requests
Your business logic and account creation
Access to your API
Risk and fraud decisions
How long you keep the result
A path for users without a wallet

What's involved

Credentials, standards and rules

Wallet presents

Agent credential

Terms to know

Regulatory context

EU AI Act transparency obligations, eIDAS 2.0 electronic attestations

The EU AI Act adds transparency duties for automated systems but does not mandate a specific agent identity method. Agent credentials use the same signed-attestation model as eIDAS 2.0 attestations. Deciding which operators and scopes to trust remains your call.

Live demo

See the complete flow in Sentinel

Authbound Agent Trust for letting verified AI agents into APIs and websites.

What it asks the wallet for

agent_id
owner_org
issuer_signature
allowed_scopes
expires_at

What the business gets back

Package: @authbound/agent-trust
Decision: allow
Replay audit: Stored

Open the Sentinel demo

Sentinel is a fictional company built by Authbound. The wallet integration in it uses the real Authbound SDK.

Questions

What buyers ask

How is this different from an API key?

A key is a shared secret with no meaning attached. Whoever holds it is the agent. A credential is signed by an issuer, names the responsible organisation, carries scopes and an expiry, and can be revoked centrally and checked by anyone.

Do we need a relationship with the agent operator?

No, and that is the useful part. Verification runs against the issuer signature and trust list, so you can admit an agent you have never seen on the basis of who vouched for it.

Does this relate to the EU AI Act?

The Act pushes toward transparency about when a system is automated and who is accountable. Attributable agent identity is a practical way to do that, though the Act does not require this specific approach.

Do agents use the EUDI Wallet?

No. Agents do not hold citizen wallets. They carry credentials in the same signed formats, such as SD-JWT VC, and are verified with the same trust and revocation model.

Can we keep our API keys?

Yes. Accept credentials alongside keys and move agent traffic over as operators adopt them.

What does our API receive?

The verified agent ID, the operating organisation, the allowed scopes and the expiry, after the signature and revocation checks pass.

What about agents without a credential?

Treat them as you do today, with keys, rate limits or blocks. Credentials let you give verified agents better access.

How do we revoke an agent?

The issuer revokes the credential and the next verification fails. You do not rotate keys across services.

Who issues agent credentials?

The organisation running the agent, or an issuer it works with. You decide which issuers you trust.

What do we build ourselves?

The check in front of your API and the mapping from scopes to permissions. Authbound handles credential verification.

Keep exploring

Related use cases

Banking & fintech

Next step

Have a different wallet use case in mind?

Tell us what you are trying to build. We will map the user journey, the credentials and the wallet interaction to an implementation.

Talk to Authbound Explore the SDK