Ask the registry whether an agent is real, and it has an answer: a matched company domain, a validated privacy identifier, a live endpoint that responded to a health check. Ask it whether that agent is entitled to sell a specific publisher's homepage inventory, and it has no file to check, because nobody has built one yet. That gap between what the IAB Tech Lab Agent Registry verifies and what publishers actually want to know is the whole story of the tool so far.
What the Agent Registry actually is
The Agent Registry is a free directory that IAB Tech Lab operates for AI agents built on the Model Context Protocol (MCP) and Agent-to-Agent (A2A) standards. Companies submit an agent for listing, choose a category such as demand-side platform, identity resolution, measurement and attribution, customer data platform or consent management, and the registry publishes it for others to find. It works two ways: a person can browse it through a web interface with filtering by category, and an agent can query it programmatically, because the registry is itself built as an MCP server. IAB Tech Lab chief executive Anthony Katsur framed the purpose plainly at launch: the registry "is here to bring trust and transparency to the Agentic Advertising ecosystem."
That framing matters, because "trust and transparency" is doing a specific and narrower job than it sounds. The registry answers the question of whether a listed agent belongs to a real, identifiable company. It does not answer the question of whether that agent has been given permission to act on any particular publisher's behalf.
What gets checked before an agent is approved
A submission is not simply published on request. IAB Tech Lab runs three checks before an agent appears in the registry: domain verification, so the company's stated web domain matches what it operates; validation against the company's IAB Tech Lab Global Privacy Platform (GPP) and Transparency and Consent Framework (TCF) identifiers, so the registrant is a known, privacy-compliant entity rather than an anonymous submission; and an endpoint health check, so the agent's technical address is live and responding rather than a placeholder.
That is a reasonable bar for company identity. It confirms that Equativ is Equativ and that Amazon's listed agent is operated by Amazon. It says nothing about scope. Passing those three checks tells a buyer that an agent exists and belongs to who it claims to belong to. It does not tell a publisher that the agent is allowed to transact against their content, their inventory, or their brand.
Who has actually registered
The registry launched on 4 March 2026 with Equativ and PubMatic among the first entries. A week later, on 11 March, IAB Tech Lab's own reporting put the total at ten registered agents, all built on MCP, with five new names added in that first week: Amazon's Amazon Ads MCP, Burt Intelligence (running separate EU and US servers), Optable, Dstillery and HyperMindZ. That same update introduced a three-tier deployment classification - remote, for hosted or cloud-based agents reachable over the internet; local, for software installed and run on a user's own machine; and private, for agents deployed inside a customer's own network or virtual private cloud - and reported the split as nine agents remote, one private (Optable) and none local.
No later published count of registered agents was found in this research. The registry's own live listing page did not return a readable agent count or list at the time of writing, which means the most recent hard figure on the public record is still the ten-agent snapshot from the launch week. Whatever the true figure is six months on, it has not been reported anywhere with the specificity of that first update, which is itself informative: a fast-growing, widely adopted registry tends to generate its own press.
The gap the registry does not close
IAB Tech Lab has been candid about this limitation from the start, in terms that are worth quoting exactly. The registry's own published roadmap lists, as a future feature rather than a current one, "creating the ability to verify that seller agent actually represents a property (like ads.txt for agents)." That single line concedes the entire structural problem that this site has already set out at length: the open web's authorisation model runs on domain-anchored files such as ads.txt and sellers.json, a generated AI answer has no domain the publisher controls, and nothing yet fills that hole for agent-mediated advertising either. The Agent Registry was the most plausible existing candidate to fill it, since IAB Tech Lab itself named the gap and the registry is the nearest available piece of infrastructure. Six months after launch, the roadmap item is unchanged: still a stated intention, not a shipped capability.
In practice this means a publisher cannot use the registry to answer the question that actually matters commercially: has this specific agent been authorised by anyone to sell, buy against, or represent my property. The registry can confirm the agent's operator is who it says it is. It cannot confirm the operator has any relationship with the publisher at all. Those are different questions, and only the first one currently has an answer.
Where the registry sits inside IAB Tech Lab's wider push
The Agent Registry is one piece of IAB Tech Lab's broader Agentic Advertising Management Protocols (AAMP) work, which has continued to add capability through 2026 on a separate track. A July 2026 update to the AAMP framework, for instance, added vendor approval gating through IAB Tech Lab's Diligence Platform and SafeGuard Privacy, so that an agentic workflow can be configured to reject unapproved partners automatically, and added pricing provenance fields aimed at stopping an agent from fabricating a floor price it was never given. Those are genuine trust mechanisms, and they sit closer to solving the authorisation problem than the registry does, because they are about gating a transaction rather than listing a participant. They are also a separate development track from the registry itself, described in a different document with no stated integration between the two as of the most recent material available. Publishers evaluating AAMP as a whole should treat the registry as one input among several, not as the finished trust layer.
Should a publisher do anything with this today
For most publishers, the registry is worth a bookmark rather than an action. It is a reasonable place to check whether a vendor pitching an agentic integration is a real, privacy-registered company, which is a low but non-trivial bar worth clearing before any conversation goes further. It is not yet a place to check whether a specific agent has any standing to act on a specific publisher's behalf, because that capability has not shipped. A publisher who wants that answer today still has to get it the way every publisher gets it now: through a direct commercial agreement, not a public directory.
blankspace's own position does not depend on which registry an agent has joined or how it identifies itself upstream of a request. Detection happens at the CDN edge, against the retrieval itself, recorded on the publisher's own logs, which is a narrower and more immediately verifiable signal than any third-party directory can currently offer. That addresses a different question from the one the registry is trying to answer, though, and it is not a substitute for an industry-wide authorisation standard when one eventually exists.
Frequently asked questions
Is registering with the IAB Tech Lab Agent Registry mandatory for AI agents?
No. It is a voluntary, free directory. An agent can operate in the market without ever appearing in it, and appearing in it does not currently unlock any privileged access or transaction right.
Does an agent's listing in the registry mean it is authorised to sell a specific publisher's inventory?
No. The registry verifies that the agent's operator is a real, privacy-compliant company with a working technical endpoint. It does not verify any relationship between that agent and a specific publisher's property, and IAB Tech Lab's own roadmap lists that capability as a future feature it has not yet built.
How many agents are currently registered?
The last figure reported with any specificity was ten registered agents, as of 11 March 2026, a week after the registry's launch. No more recent published count was found in researching this article, which is itself a sign of how early-stage the directory still is.
What is the difference between the Agent Registry and AAMP?
AAMP, the Agentic Advertising Management Protocols, is IAB Tech Lab's broader set of standards and tools for agentic advertising, covering areas such as deployment infrastructure, vendor approval and pricing integrity. The Agent Registry is a single tool within that wider effort: a directory for discovering and identifying agents, rather than the mechanism that governs how a transaction is approved or priced.
Should a publisher check the registry before working with an agentic advertising vendor?
It is a sensible first step but not a sufficient one. Confirming a vendor's agent is listed and its operating company is verified is a reasonable baseline check. It does not replace the commercial and contractual work of confirming that vendor has any actual authority to transact against the publisher's own inventory.