Early access

Trust infrastructure
for AI agents

Organisations and developers register on Agentis and issue cryptographic identities to their agents. Any counterparty can verify who owns an agent and whether to trust it, before interacting with it.

The problem with agent credentials today

No attribution

Agents call APIs with shared tokens. There's no standard way to know which agent made a request or who is accountable for it.

No proof

Bearer tokens can be stolen and replayed. There's no cryptographic proof that a request actually came from a specific agent.

No audit trail

When something goes wrong, you can't reconstruct what your agents did, under what authority, or whether the record has been tampered with.

How Agentis works

01

Verify your organisation

Register your company or individual identity. UK companies are verified against Companies House. Domain ownership is confirmed via DNS TXT record.

02

Issue agent identities

Each agent gets an Ed25519 keypair and a W3C DID (did:web:agentis.dev:agents:…). The private key is returned once and never stored. Use it with proveyouragent to sign every request.

03

Verify any agent, anywhere

Any counterparty calls GET /api/agents/[did] to verify an agent's identity, capabilities, and the organisation behind it, before interacting with it.

Who uses Agentis

Building multi-agent systems

Your orchestrator calls sub-agent APIs. Those servers have no way to verify the request is legitimate or know which agent is calling. Agentis gives every agent a verified identity your servers can check.

Deploying third-party agents

You are running agents built by a developer you have never met. Your security team is asking questions. Call the public DID endpoint before deployment and see exactly who built the agent and what it is allowed to do.

Operating in a regulated environment

Your agents touch sensitive data and take consequential actions. When your compliance team or an auditor asks what happened, the Agentis audit chain gives you a tamper-evident record you can prove has not been altered.

One endpoint to verify any agent

Public, unauthenticated, designed to be called by counterparties at runtime.

Request
$ curl https://agentis.dev/api/agents/did:web:agentis.dev:agents:28e83dff641920ad
Response
{  "did": "did:web:agentis.dev:agents:28e83dff641920ad",  "name": "billing-agent",  "capabilities": ["invoices:read", "payments:write"],  "status": "active",  "verifiedName": "Acme Ltd",  "verificationTier": 3,  "verificationMethod": "company",  "verifiedDomain": "acme.com"}

Verification tiers

TierMethodWhoWhat you get
1
Email onlyAnyoneAgents marked unverified
2
IndividualSolo developersReal person confirmed
3
OrganisationUK registered companiesCompanies House + domain verified
4
Compliance readyRegulated industriesComing soonComing soon
Layer 5: Request signing

Built to work with proveyouragent

Agentis handles identity registration and verification. proveyouragent handles request-level signing with Ed25519 and DPoP. Together they give your agents cryptographic identity from registration to every single request.

Register for the beta

Agentis is in early access. Register your organisation and start issuing agent identities today.

Tier 2Individual

Solo developer

For individual developers building agents. Verified via email.

Tier 3Organisation

Registered company

For UK companies. Verified via Companies House and DNS domain ownership.