IAB Tech Lab's product chief this week published a point-by-point rebuttal of the comparisons circulating between its Agentic Advertising Management Protocols and the Ad Context Protocol, listing 13 business functions that both frameworks address and rejecting the idea that the two occupy separate, complementary layers.

The argument matters because the layer story has become the industry's default shorthand. Under that account, AdCP handles campaign planning and AdCP-mediated media buys, while AAMP handles what happens at the moment an impression clears. If that were accurate, a buyer could adopt both without choosing. The document published by IAB Tech Lab says it is not accurate, and it names the specific places where it breaks.

The post was written by Shailley Singh, chief operating officer and executive vice president of product at the standards body. It appears at the top of the organisation's news listing, above the finalised release of the Agentic Real Time Framework version 1.0, which IAB Tech Lab published earlier this month. The copy reviewed for this article carries no date stamp.

Singh opens by describing the source of the problem. Over recent months, articles, social media posts and event discussions have set out what AAMP does, and in his assessment they have done so without reading the published material. According to the document, the comparisons tend to begin from a hypothetical architecture rather than from the specification, and the result is a straw man built around presumed roles.

Seven claims, assessed

The centrepiece of the post is a table that takes seven statements found in published comparisons and rules on each. Five are labelled incorrect or an incorrect generalisation. One is called too absolute. One is called hypothetical and not standardised. One is called a weak comparison.

The first claim under review holds that AAMP extends the impression layer while AdCP defines the campaign layer. Singh marks it incorrect, writing that AAMP explicitly includes a management layer covering audience and media discovery, planning, negotiation, orders, deals, signals and pre-execution setup.

The second holds that the two are not substitutes. That assessment is softened rather than reversed. They are not complete one-for-one substitutes, the document states, but major portions are directly substitutable or overlapping.

The third holds that AdCP runs before and after the auction while AAMP runs inside the platform where the auction clears. Marked incorrect. The document says AAMP supports direct transactions, Programmatic Guaranteed, private marketplaces, deal creation, bookings and campaign planning.

The fourth holds that one AdCP media buy produces thousands of AAMP impression events. The response is that no normative decomposition or interoperability contract between the two establishes that relationship, making the claim a hypothesis dressed as a mapping.

The fifth compares the Agentic Real-Time Framework to AdCP's Trusted Match. The document calls this a weak comparison: ARTF is described as a low-latency, host-executed agent-service framework supporting several use cases, while Trusted Match is characterised as one targeting plane.

The sixth holds that IAB Tech Lab standards carry no view of the connected television pod. That is the claim the document treats most bluntly, and it is addressed at length below.

The seventh holds that AAMP primitives assume auction dynamics. The document calls this an incorrect generalisation, drawing a line between ARTF and OpenRTB, which do address auctions, and the AAMP Buyer and Seller Agents, which it says support media buying and selling processes with no connection to auctions.

One pillar mistaken for a framework

Singh identifies what he calls the central category error: treating one component of AAMP as the whole of it. AAMP is described as three pillars. Agentic Foundations, of which ARTF is one element. Agentic Protocols and Standards, codified in the Buyer and Seller Agent SDKs. And an Agent Registry for trust and transparency.

That structure matches the architecture PPC Land documented when IAB Tech Lab formally named the initiative on February 26, 2026, setting out seven protocol components each mapped to an existing standard and a three-level agent hierarchy running from portfolio orchestration down to functional agents. The naming exercise was itself an attempt to stop observers from treating ARTF and the audience framework as the sum of the organisation's agentic work.

The post argues that the Agentic Protocols layer is a management layer by IAB Tech Lab's own description, covering how agents discover buyers, sellers and inventory, discover audiences and ad products, understand campaign requirements, negotiate pricing, manage orders and deals, exchange audience signals, and complete campaign setup before execution. Those are campaign and transaction lifecycle functions, not impression-layer auction activity.

The Buyer Agent is offered as the sharpest evidence. According to the document, it accepts a campaign brief, allocates budgets across channels, researches multiple sellers, negotiates prices, obtains human approval, books deals and produces an audit trail. Its documentation covers linear television buying, change requests, pacing, creative management, reporting and multi-seller orchestration.

Singh writes that AAMP "is not an 'impression-layer protocol.'" The framework, he continues, includes campaign planning, media and audience discovery, transaction management and impression-time execution.

The overlap list

The most consequential section is the shortest. Where most comparisons open and close on a layer distinction, the document states that the opposite is closer to the truth and lists 13 areas both frameworks address:

Buyer and seller agents. Natural-language campaign briefs. Inventory and product discovery. Seller discovery. Media planning. Pricing and negotiation. Deal or order creation. Human approval before spend. Audience discovery and activation. Agent identity and registry functions. MCP and A2A communication. Campaign lifecycle and reporting. Real-time or serve-time decisioning.

The object names and abstractions differ, the document concedes, but the business functions overlap. From that it draws a conclusion aimed squarely at the prevailing narrative: comparisons cannot assign those functions exclusively to AdCP and then use that assignment to prove complementarity.

Shared plumbing does not close the gap either. On the treatment of the Model Context Protocol, which both frameworks use, the document is direct. Support for MCP does not make the frameworks complementary, Singh writes, because "MCP is a communication substrate; it does not reconcile their advertising semantics."

That distinction has practical weight. MCP itself has been moving underneath both. The base standard shipped its largest specification rewrite since 2024, removing sessions and forcing advertising vendors that built on it to rework server implementations.

The pod objection

The claim that IAB Tech Lab standards carry no view of the CTV pod draws the most technical response in the document, and the shortest patience.

OpenRTB 2.6, the post notes, explicitly introduced ad-pod support for connected television and audio through a set of named fields: poddur, podid, podseq, slotinpod, maxseq and mincpmpersec, together with exact-duration support and pod deduplication signalling. Those fields have been in production use for years. PPC Land's reference material records that the vocabulary arrived with OpenRTB 2.6 in spring 2022, giving sellers a standard way to group impression opportunities inside a break and price them by position rather than treating every slot as identical.

The commercial consequences are documented. A+E Global Media cut inbound ad server requests by 84% and lifted impressions 39% after moving from individual slot auctions to pod-level bidding on an OpenRTB 2.6 integration with Index Exchange and FreeWheel. The pod is not a gap in the standard. It is a field set with a revenue record attached.

Singh also takes issue with the assertion that CTV is not genuinely impression-level. He calls that "rhetoric rather than a useful technical proposition," and lists five grains at which connected television can be traded: campaign or deal, show or audience package, ad break or pod, slot, and delivered impressions. A system may need several simultaneously. That does not make the impression layer unreal.

Channels without auctions

A related argument holds that AAMP is structurally disadvantaged for broadcast, CTV, radio and digital out-of-home, because those channels do not always clear through impression auctions. The document rejects the premise.

AAMP does not require every transaction to clear through OpenRTB, according to the post. Its architecture includes OpenDirect-based direct transactions, Deals API integration, private marketplaces, programmatic guaranteed transactions, buyer and seller booking workflows, and linear television scatter buying in the current Buyer Agent. The Buyer Agent is described as supporting DMA targeting, cost-per-point and CPM pricing, dayparts, bookings, order state machines and post-deal changes for linear television.

Market activity has been running in the same direction. VIOOH's seller agent automated more than 100 curated DOOH deal package requests during the first half of 2026, with planned registration under both AdCP and AAMP rather than one of them.

What the document says AAMP is

The closing section sets out design priorities: reuse of established advertising semantics, compatibility with deployed infrastructure, buyer and seller reference implementations, direct and deal-based and programmatic workflows, deterministic execution at advertising-system latency, registries and diligence, post-campaign reconciliation and billing support, and incremental industry adoption.

The stated advantage is continuity. AAMP builds on standards and operational concepts that platforms already use to transact advertising spend today, with integrations into existing systems.

Why the framing matters to buyers

The layer story is convenient for anyone who does not want to choose. If AdCP owns the campaign layer and AAMP owns the impression layer, an agency can implement both and treat the question of which standard prevails as somebody else's problem. Singh's rebuttal removes that comfort. If the two frameworks genuinely overlap across planning, discovery, negotiation, ordering and audience activation, then a buyer implementing both is duplicating work, and a publisher exposing inventory to both is maintaining two descriptions of the same supply.

Adoption is where the practical stakes sit. AdCP launched on October 15, 2025 with six founding members and 23 additional launch participants, defining nine core tasks across the advertising lifecycle. It divided the industry within days, and Ari Paparo's November 3, 2025 analysis supported parts of the specification while questioning the media buying applications. Google, Amazon, The Trade Desk and Microsoft were absent from the founding roster.

On the other side, the IAB Tech Lab Agent Registry reached 10 entries on March 11, 2026, every one classified as an MCP server and none as agent-to-agent. Neither framework has anything resembling universal coverage, which is what makes the semantic dispute more than academic. A protocol war fought over definitions still ends with implementation costs landing on the same media buyers and publishers.

The frameworks have also kept shipping. AAMP 2.3 arrived on July 30, 2026 in a release authored by Singh, adding a pricing provenance field designed to stop AI buying agents fabricating bid prices when real market data is unavailable, alongside a vendor approval gate and audience embeddings. Days later, PubMatic launched guardrails that run as an independent check on AdCP traffic, enforcing at the platform where the protocol only describes.

That sequence illustrates the point Singh is making from the other end. The two frameworks are not dividing labour along a clean seam. They are converging on the same problems, at the same time, from different vocabularies. Whether the industry treats that convergence as duplication or as healthy competition will determine how much integration work the next two years require.

Timeline

Summary

Who: Shailley Singh, chief operating officer and executive vice president of product at IAB Tech Lab, the global digital advertising technical standards body established in 2014. The post responds to unnamed authors of published comparisons between AAMP and the Ad Context Protocol, an open standard governed by AgenticAdvertising.org.

What: A point-by-point rebuttal arguing that AAMP is an umbrella framework spanning campaign planning, media and audience discovery, transaction management and impression-time execution, rather than an impression-layer protocol. The document assesses seven claims found in published comparisons, marking five incorrect, and lists 13 business functions that AAMP and AdCP both address. It rejects the argument that shared MCP support makes the two complementary, and disputes the assertion that IAB Tech Lab standards lack a view of the connected television pod by naming the OpenRTB 2.6 fields poddur, podid, podseq, slotinpod, maxseq and mincpmpersec.

When: IAB Tech Lab published the post this month. It sits at the top of the organisation's news listing, above the finalised release of the Agentic Real Time Framework version 1.0. The copy reviewed for this article carries no date stamp. AAMP was formally named on February 26, 2026, and version 2.3 shipped on July 30, 2026.

Where: On IAB Tech Lab's website, addressing a debate that has run across trade articles, social media posts and conference discussions since AdCP launched on October 15, 2025.

Why: The layer distinction has become the industry's working assumption about how the two agentic advertising frameworks relate. If the functions overlap as extensively as the document argues, buyers and publishers implementing both are duplicating integration work rather than assembling complementary pieces, and the question of which standard prevails remains open rather than settled.