If you’ve been involved in digital advertising for a couple of months, you’ll already have noticed something odd: adverts are increasingly being launched not by humans, but by a layer of AI that decides on budget, audience and creative almost in real time. Until now, every platform — Google, Meta, TikTok, Amazon, the major DSPs — spoke its own language. The problem is clear: if AI agents are going to buy media on our behalf, someone has to decide how they communicate with one another. That’s the challenge facing the Ad Context Protocol (AdCP), the open standard that was introduced in late 2025 and which, by 2026, was beginning to feature in discussions amongst serious media organisations.
What exactly is the Ad Context Protocol?
The Ad Context Protocol It is an open communication protocol that enables AI agents from different companies — advertisers, agencies, publishers and ad-tech firms — to communicate with one another using a common language to buy, sell and activate audiences. It is built on the Model Context Protocol from Anthropic, the same standard also used by productivity tools and LLM integrations, and is developed under open governance so that it does not depend on a single company. The full specification is published at adcontextprotocol.org and in the official GitHub repository.
The comparison most frequently cited in the sector is with OpenRTB. In the early 2010s, OpenRTB was what enabled RTB (real-time bidding) to scale: a common grammar that anyone could implement. Today, with agent-driven advertising on the scene — campaigns planned, launched and optimised by autonomous agents — another layer of common framework is needed. The Ad Context Protocol aims to be that layer, as described in the project’s own documentation and in analyses by specialist media such as Digiday and AdExchanger.
Who is behind the standard?
The coalition that launched the Ad Context Protocol is broad and brings together a mix of players who would normally be competing with one another. The founding members include Yahoo, PubMatic, Scope3, Optable, Swivel and Triton Digital. Among the launch members were Samba TV, Magnite, Kargo, LG Ad Solutions, Raptive, The Weather Company, Butler/Till, Classify, Newton Research and Accuweather, amongst others. The official press release from Samba TV provides a full list of participating companies at the launch.
The governance structure is not merely for show: there are four voting categories with equal weighting — brands, agencies, publishers and technology providers — and an open working group that maintains the specification. When a standard is controlled by a single company, it often ends up shaping the rules to favour its own technology stack. AdCP was designed from the outset to prevent precisely that, which in practice influences whether major advertisers decide to adopt it or hold off.
The four parts of the protocol
The Ad Context Protocol is not a single, monolithic block. It consists of several modules covering the key operations that any campaign needs to carry out:
- Signals Activation Protocol: how an agent describes and triggers audience signals (interests, context, first-party data) across different platforms without having to rewrite the integration each time.
- Media Buy Protocol: the negotiation and execution of media buying between the advertiser’s representative and the publisher’s representative or the SSP.
- Creative Protocol: how creative assets — images, videos, copy — are exchanged, validated and approved within the agency workflow.
- Curation Protocol: intended for a later phase, this regulates the reconciliation of stock and deal IDs in environments where agents negotiate complex terms.
An important technical point: the Ad Context Protocol works by asynchronous. A negotiation can last anywhere from seconds to days, allowing for human approvals to be incorporated into the process — something the sector is beginning to call human-in-the-loop— without interrupting the flow. It’s not just a fast bidding channel; it’s a layer of machine-to-machine communication that recognises that sometimes a human has the final say.
What problem does it solve for a brand or an agency?
The day-to-day work of any team managing Google Ads, Meta Ads, LinkedIn Ads and a couple of DSPs is a jumble of tabs. Each platform has its own API, its own bidding logic, its own way of defining audiences and its own method of measuring conversions. When OpenAI has made ChatGPT available to ad managers And as agents began to appear who promised to “run your campaign from start to finish”, the problem became more apparent: each agent only understands their own little world.
With a standard such as the Ad Context Protocol, in theory, a single agency can request quotes from different ad inventories, compare terms, activate its own audiences across various providers and approve creative without your team having to set up bespoke integrations with each one. The difference compared to the current situation is similar to that between writing an email and having to learn a different protocol for each email client. If it works, the operational cost of coordinating multi-platform campaigns should fall significantly.
Why the Ad Context Protocol matters right now
The context of 2026 helps to explain the urgency. In Google Marketing Live 2026 Ask Advisor and the Business Agent for Leads were unveiled – two tools that assume advertisers will increasingly delegate more nuanced decisions to agents. Meta, for its part, is redesigning its advertising stack around generative AI and, according to figures released in June 2026, is on track to overtake Google in global advertising revenue for the first time. OpenAI has opened up ChatGPT to advertisers of all budget sizes. Meanwhile, Hearst Magazines has launched Aura IQ, an agent-based platform that builds audiences from scratch and doubled the CTR in initial tests.
With so many ad networks emerging at once, the question is not whether ad network-based advertising will grow, but whether it will do so within closed ecosystems or on open standards. The Ad Context Protocol is the first serious attempt on the supply side to ensure that the answer is the latter.
Limitations and grey areas
Let’s be honest: the Ad Context Protocol is still at a very early stage. The protocol is publicly available on GitHub and there is a reference implementation, but actual adoption depends on major demand-side platforms supporting it. Google, Meta, Amazon and TikTok are not listed as founding members. Without their involvement, the standard may remain confined to premium publishers, SSPs and a handful of DSPs.
Another reasonable concern is measurement. The fact that two parties negotiate a purchase using a common language does not automatically mean that attribution, reports and KPIs become comparable. The measurement module was not available at the time of launch and, according to the official documentation, forms part of the work planned for subsequent phases. In the meantime, media teams will continue to combine the Ad Context Protocol with their own tracking layers.
Nor is the exact role of the brand safety in this new workflow. When an agent makes decisions at machine speed, the rules governing context, exclusions and independent verification must also be included in the protocol. The specification provides fields for this, but the approach has yet to be validated at scale.
What you can start looking at if you’re managing campaigns today
You don’t need to change your stack tomorrow, but it’s worth making sure you’re not left behind. Three sensible steps for the coming months, with the Ad Context Protocol on the horizon:
- Check the official documentation at adcontextprotocol.org and the public GitHub repository. Even if you’re not a technical person, reading the high-level specification will help you understand what the agents you start hiring will and won’t be able to do.
- Ask your suppliers: your DSP, your SSP, your data platform. If they are part of the coalition or plan to adopt the Ad Context Protocol, their future integrations will be simplified, which will reduce your operating costs. If they do not have a clear answer, this is a sign to bear in mind.
- Organise your own signals. An agent-driven world rewards brands that have clean, well-tagged first-party data. AdCP defines how these signals are triggered, but it does not create them for you. The groundwork — CRM, site events, modelled audiences — remains the advertiser’s responsibility.
At Vandelay, we’re keeping a close eye on this development because it aligns with how we already approach campaigns: prioritising clean data, automating what can be automated, and leaving room for the human decisions that really matter. If the Ad Context Protocol delivers on its promises, brands that have their house in order will benefit before the rest. If it falls short, it will at least have pushed the sector to speak the same language – which is quite an achievement after a decade of silos. We’ll be posting updates as integrations are confirmed and the first measurable real-world case studies emerge in Spanish-speaking markets.
Sources consulted: official documentation from adcontextprotocol.org, the project’s GitHub, Samba TV press release (15 October 2025), Digiday “WTF is Ad Context Protocol”, AdExchanger, MediaPost, Scope3, Equativ and coverage of Google Marketing Live 2026 published on blog.google.
English (UK) 
Español