The Ad Context Protocol (AdCP) was recently launched as an open standard to establish a shared technical foundation for the emerging era of agentic advertising. Developed by a consortium of over 20 companies, including Yahoo, PubMatic, and Magnite, AdCP enables AI agents to communicate with advertising platforms and, eventually, directly with each other. This protocol is designed to streamline complex programmatic workflows, allow for more flexible transaction models , and facilitate direct access to inventory, ultimately aiming to reduce complexity, increase transparency, and improve efficiency in media buying and selling. NDA spoke to the industry to get their reaction.
Mateusz Rumiński, VP of Product, PrimeAudience, comments:
“AdCP, similarly to other standards, requires high adoption from market participants to succeed. It is promising that some influential figures in the advertising ecosystem are involved. Still, it will take much more than big names to make it work. Even if the industry aligns around this, it will likely take years for advertisers and agencies, in particular, to adjust their processes and ensure this approach satisfies all required use cases.
“The standard is one to watch, however, as there is no doubt it will require tech platforms to rebuild themselves. While the demo showcased at the launch could be implemented relatively easily, it may be some time before all the nuances of programmatic are covered.
“It is also worth noting that the approach of AdCP requires a lot of trust from those involved from all sides of the ecosystem. This is because it involves providing natural language input and trusting that the agent will interpret it exactly as intended by the advertiser. While some are optimistic, this cannot be guaranteed at this early stage.”
Maor Sadra, CEO and co-founder of INCRMNTAL
The swift attempt to create a protocol and control the standards so the industry can have an open and competitive landscape of buyers, bidders, and sellers, is super positive. But, if we learned anything from the playbook of the most successful ad platforms in our space – an open industry doesn’t necessarily work to the advantage of the strongest players, and allow them to hold the fort around their walled gardens.
Some of us remember the wild west era of the industry trying to make claim for an RTB standard. Before openRTB, there were multiple proprietary protocols: AppNexus, RightMedia, Nexage, RTBKit. Most SSPs will tell you that OpenRTB was standard but featured many customisations.
That said, OpenRTB was a clear signal that the industry was maturing. It defined a framework for how inventory would be traded. The Ad Context Protocol feels like that same inflection point – this time, for AI-driven advertising. It’s a sign that advertising in LLM environments isn’t just a thought experiment. Once inventory becomes addressable, the floodgates of supply will open, and the race for demand will begin all over again.
And here’s where the measurement challenge kicks in: when agents start buying and optimising in real-time across a mix of human and machine-generated placements, AI attribution will be the only source of truth. Standards can define how ads are bought – but not how they’re valued. That’s the next frontier.
Thomas Ives, Co-founder RAAS LAB, comments:
“The launch of the Ad Context Protocol (AdCP) is a positive move. We’ve seen what happens in our industry when we disagree on a standard: a fragmented mess of closed systems and black-box decisioning.
“Given the seismic shift AI will create, it’s important that we build on strong foundations and create a healthy ecosystem from the start, rather than rely on a new sticking plaster every week. If the AdCP gets the fundamentals right – acting as a common framework that supports freedom of innovation at scale – it will enable both collaboration and healthy competition. The responsibility now lies with the industry to embrace it and move forward together.”
Andy Oakes , Publisher New Digital Age and CEO of Bluestripe
As a publisher, obviously the sell-side implications are the ones I’m thinking about.
To me it looks like the most significant promise for publishers is the potential for cutting out intermediary layers (ad exchanges, multiple SSPs etc). By allowing the buyers agents to communicate directly with a “Seller Agent” running on the publisher’s ad server, more of the advertiser’s pound is intended to reach the publisher. Yes, we’ve heard this before but this looks promising as a way of getting to the promised land of higher yields.
Unlike the anonymous dynamics of open real-time bidding, AdCP should allow publishers to set explicit, human-vetted business rules. For example, a publisher’s agent could be programmed to auto-approve small campaigns but require human approval for large deals or ads in sensitive content categories, which looks to be a sensible option. ‘Never entirely take humans out of the mix’ should always be our market motto.
The protocol looks to provide a structured way for publishers to expose their unique, first-party audience data and premium inventory packages to buyer agents in a universally understood format. The aim is to make it easier for them to sell high-value inventory at scale without extensive manual work. Hopefully.
All this being said, we’re not ripping up the current way of doing things. As I see it, AdCP is generally viewed as an additional layer on top of existing programmatic infrastructure rather than a replacement, allowing publishers to adopt agent-based buying gradually and a rate which makes sense to them
Kelly McMahon, EVP of Global Operations at LG Ad Solutions
“We see AdCP as an important step toward an advertising ecosystem built on connected intelligence. The next wave of media buying will depend on interoperability, transparency, and trust — and AdCP represents a shared effort to make that future real. By engaging early, we’re helping shape how AI agents communicate, transact, and collaborate across the industry, ensuring that innovation evolves openly and responsibly for everyone involved.”
