Why banks need to know the agent as well as the customer
A bank authenticates the account holder. When an agent acts for that holder, the bank also needs to know which agent it is, which company operates it and whether that is the same agent it dealt with last week. Without that, every agent channel is either closed or an open door.
What AgenID contributes and what it leaves alone
The finance agent presents its AgenID. The bank resolves it, re-verifies the operator's signature, and reads which claims about the operator have been checked. The illustration shows L3_ORGANIZATION_VERIFIED, meaning an authority reviewed evidence of the operating organization, and L4_DEPLOYMENT_VERIFIED, meaning the deployment itself was checked.
Everything else stays with the bank: customer authentication, account ownership, transaction authorization, fraud controls and its regulatory obligations. AgenID does not hold accounts, move money or approve transfers.
Why the boundary matters
An identity layer that started approving transactions would have a stake in saying yes. Keeping AgenID to identity, and keeping every financial decision with the institution, is what lets a bank use it as one input among its existing controls.
Today the reference registry at agenid.com issues DECLARED and L1_REGISTERED. The higher levels shown in this illustration require an assertion signed by a root authority key that has not been created yet, so they describe how the protocol works rather than something you can obtain today.
Frequently asked questions
Can AI agents make bank transfers?
Some banks are opening agent channels. In every case the bank still authenticates the customer and authorizes the transfer. AgenID adds a checkable answer to which agent is making the request.
What is "know your agent"?
It is the agent-side counterpart to know your customer: establishing which AI agent is acting, who operates it and which facts about that operator have been independently verified.
Does AgenID authorize payments?
No. AgenID answers which agent is acting and who declared it. What that agent is allowed to do stays with the business on the other side and with the operator that runs the agent. Protocol v1.1.1 defines no signed authorization object.
Is L4 deployment verification available?
Not yet. The L4 sampling methodology has not been finalised, and no level above L1_REGISTERED is issuable until the root authority key exists. The page shows L4 to explain the model.