AAMP

Why Advertising Agents Need Verified Identity

An autonomous agent negotiating a media buy on your behalf is, from the other side of the transaction, an anonymous piece of software making claims about itself. It says who it represents, what authority it has, what terms it can agree to. Nothing about the negotiation itself confirms any of that is true.

Human-mediated advertising solved this problem informally — reputations, relationships, contracts, a phone call to confirm. Autonomous agents removed the human, and with it, the informal trust mechanism that made the rest of the system work without anyone noticing it was there.

AAMP, the Agentic Advertising Management Protocols, exists to rebuild that trust mechanism explicitly, in a form machines can check.

What AAMP Standardizes

AAMP provides a verified identity layer for agents operating in advertising. At its center is a registry: an industry directory where buying and selling agents register their organizational identity, legal declarations, operational scope, and public signing keys. When one agent receives a proposal from another, it can check the registry to confirm the counterparty is who it claims to be, and authorized to make the claims it's making.

Beyond identity, AAMP standardizes workflow across the campaign lifecycle — direct buying, audience signal exchange, creative handling — giving enterprise buy-side and sell-side systems a shared object model instead of each vendor inventing its own.

How AAMP Relates to AdCP and MCP

AAMP operates as a governance and management layer above the transaction-specific work AdCP does. Where AdCP standardizes how a buyer agent and seller agent negotiate a specific deal — discovery, planning, buying, measurement — AAMP standardizes who is allowed to be in that negotiation in the first place, and provides the enterprise workflow scaffolding around it.

AAMP also incorporates MCP directly, allowing generative agents to interact safely with lower-level delivery systems through the same tool-access standard MCP defines elsewhere in the stack. An agent verified through AAMP's registry can then use AdCP to actually conduct the negotiation it's now authorized to have.

Where Trust and Provenance Fit

Identity verification and provenance are related but distinct concerns. AAMP's registry answers is this agent who it says it is. Provenance, as used elsewhere in the stack — including in supply descriptors — answers is this claim about a piece of inventory reliable, and where did it originate. AAMP provides the identity foundation that provenance claims rest on: a signed claim from a verified agent carries more weight than the same claim from an unverified one, precisely because AAMP makes verification checkable.

What AAMP Does Not Do

AAMP does not conduct the negotiation itself — that remains AdCP's domain. It does not evaluate supply quality or content meaning — that is the work of supply intelligence systems operating upstream of any transaction. And AAMP is not a replacement for AdCP; it is closer to a prerequisite, establishing who is allowed to transact before defining how the transaction happens.

Why This Matters

Removing a human from a negotiation removes a trust mechanism nobody had to formalize, because a person on the other end of a phone call was trust enough. Autonomous agents don't get that shortcut. Every claim an agent makes — who it represents, what it's authorized to do — has to be checkable by another piece of software, or the whole system runs on faith rather than verification.

AAMP is the answer to that gap: not a negotiation protocol, but the identity layer that makes every other protocol's negotiations worth trusting.