Skip to main content
Billions links an AI agent to the verified human behind it, and gives services a way to check that link before they trust the agent.

​
Pick a task

Pair an agent

Give your agent a DID and pair it with you. Two chat messages.

Verify an agent in my app

Check that a caller controls an agent DID and that a verified human is paired with it.

Protect an x402 endpoint

Offer paired agents a different price or limit on a paid API.

Inspect attestations

Read a real pairing in the Explorer and the API, field by field.
Each guide has its prerequisites, the exact commands, the response you should see, and the page to read next.

​
What’s involved

Three parts work together. You don’t need to know them to start a task above, but the guides refer to them.

​
What a pairing proves

A pairing proves that a person with a Proof of Uniqueness credential paired with a specific agent DID, and the ownership attestation records it on-chain. It does not prove:
  • That the caller is that agent. DIDs and attestations are public, so anyone can quote one. Your application has to check a signature from the agent, which is called proof of control.
  • That the agent behaves well. A paired agent can still send spam. Pairing lets you set limits per person.
  • Which agent did something, unless that request was signed with the agent’s key.

​
Production and testnet

Production holds real pairings and is what a live application should query. Testnet is for practice: creating and revoking your own records. Records never move between them. See Environments for every URL and address.