NEXTAI has published its agentic identity policy, a public agent registry available in JSON and the technical documentation for the credentials it issues. All three resources are accessible without signing up and describe what data each agent's identity includes, who answers for it and how to check it.

What has been published

There are three complementary pieces. The agentic identity policy sets out the commitments: what is declared about each agent, who the responsible party is, how long a credential lasts and in which cases it is cancelled. The public agent registry returns the list of agents with their current status in JSON. The technical documentation details formats, fields and endpoints, for anyone who wants to verify on their own or reproduce the approach.

The identity is issued by the CORTEX brand under the identifier did:web:digitalcortex.tech. Both CORTEX and NEXTAI are brands of Cosmolabs OÜ, an Estonian company with registry code 17106582.

The minimum data for each agent

Each credential declares a bounded set of information:

• The agent's own identifier, in the form did:web:digitalcortex.tech:agents:<slug>.
• The issuing and responsible company, with its registry code.
• Declared scope: what the agent can do.
• Express prohibitions: what it cannot do under any circumstances.
• Autonomy level, on a scale from A0 (observe) to A4 (execute with valid human approval).
• SHA-256 hash of the registered manifest; scope and prohibitions are signed separately at issuance.
• Validity dates, with a 90-day expiry and renewal.

The credential contains no personal data about individuals. Accountability sits with the company, with an accountable role and a legal contact address, not with named employees.

There is a class of actions the policy places out of reach for any agent, whatever its level: moving money autonomously, deleting audit trails and escalating privileges.

Standards and verification

The credentials follow W3C Verifiable Credentials 2.0 in vc+jwt format, signed with ES256 over the P-256 curve, with the private key held in Google Cloud KMS. Revocation uses W3C Bitstring Status List v1.0.

The public check runs eight live verifications: credential format, the key declared by the issuer, signature, issuer consistency, validity, revocation, unchanged manifest and published DID document.

The standards are open and the profile specification will be published under an open licence, so that any company can issue credentials for its own agents without depending on NEXTAI. As far as the company has been able to check, it is not aware of another AI consultancy for SMEs in its market that publishes verifiable credentials for its agents.

How it relates to Article 50

Article 50 of the AI Act sets transparency obligations according to the role of the provider or deployer and the specific use. It does not require a cryptographic credential and it is not automatically satisfied by a seal. A verifiable identity complements the notices where those apply. See the official text and our guide for companies.

The declaration is signed, dated and bound to the registered manifest. A clear notice in each interaction and channel keeps its own function: it should not be replaced by a link the recipient has to go and find.

The concept, in plain language, is the digital ID for AI agents; the cryptographic mechanics are explained in W3C verifiable credentials for agents, and how it fits a company's internal inventory in the agent registry.

Verification links

• Agentic identity policy: digitalcortex.tech/identidad-agentica
• Public agent registry in JSON: digitalcortex.tech/functions/agentRegistry
• Technical documentation: digitalcortex.tech/docs/identidad
• Public verifier for the agent Argos: digitalcortex.tech/verify/argos

Editorial update of 17 September 2026, produced with the support of AI tools and cross-checked against public documentation. It is not presented as an external audit.

Identity and execution: the scope of this explainer

The public checks describe the credential and its relationship with the registered manifest. They do not prove the authentication of whoever presents it, the state of each external permission, or that actions are actually blocked. See the published evidence and limits.