Let’s be clear: AdCP isn’t a finished product — it’s a prototype in its infancy, a draft idea being written and rewritten by a handful of engineers, founders, and visionaries who believe that advertising’s next great leap won’t be about data, cookies, or auctions — it will be about language.

The Ad Context Protocol (AdCP) aims to give artificial intelligence a shared grammar for how media is bought, sold, and optimized.

If it succeeds, it could become the connective tissue between AI systems and the advertising economy, enabling machines to negotiate and execute campaigns with the same nuance, logic, and accountability that humans claim — and often fail — to achieve.

But even as the excitement builds, the people closest to it are the first to warn against overstatement. Jeff Hirsch, CEO of QuantumPath, believes the concept has potential, but also complexity that’s being glossed over by the “next big thing” crowd.

Hirsch said that agentic advertising isn’t just about agent-to-agent communication — it can also be used to make current processes more efficient. He cautioned, “Let’s make sure we don’t classify AdCP as the only approach to agentic.”

It’s a nuanced point — and a necessary one. The industry loves a savior acronym. But as Hirsch and others remind us, AdCP is just one experiment in a much larger shift toward automation, interoperability, and agentic infrastructure.

The founding consortium behind AdCP reads like a who’s who of companies that have quietly shaped ad tech’s back-end for years: Triton Digital, Yahoo!, PubMatic, Scope3, Optable, and Swivel. Each has committed engineers and resources to design a protocol that would allow intelligent agents — buyer- and seller-side — to speak a common language.

At the center is Brian O’Kelley, who has spent his career building, breaking, and rebuilding ad tech’s core plumbing. The AppNexus founder and Prebid co-creator calls AdCP “the universal ads API” — not another marketplace or buying platform, but a communication layer that sits between systems, enabling transparent, structured conversations between machines. He noted that every major social platform already has its own ads API, and AdCP extends that logic to the open web — so that publishers, streamers, and retailers can communicate in the same structured way.

In practice, that could look like an advertiser’s agent sending a prompt — “Find women who rock climb in the U.K. and care about sustainable outdoor gear” — and a publisher’s agent responding, “Here’s our available inventory, pricing model, and expected ROAS.”

No dashboards. No spreadsheets. No middlemen toggling between DSP and SSP interfaces. Just machine-to-machine negotiation, backed by shared logic and verifiable metadata.

Anne Coghlan, COO and Co-Founder of Scope3, described AdCP as “a universal handshake” — the first open attempt to standardize communication between agents so that intent, inventory, and outcomes are expressed in a consistent language.

An industry executive who asked to remain on background said the same forces driving this protocol are the same ones that have broken the current system. “Before we call this the future, we need to fix what’s already collapsed — fraud, attribution, reporting,” the executive said. “Otherwise, AdCP risks becoming a faster, shinier way to hide the same problems.”

PubMatic’s Kyle Dozeman agrees that AdCP is “a Prebid-type layer for agentic advertising.” But even among its early supporters, there’s recognition that this is still conceptual. Hirsch added that AdCP is the first step in solving interoperability between buyer and seller agents, but there’s a lot to figure out before it can scale — formats, walled gardens, publisher controls, and the future of direct sales. All solvable, but slow.

Scott Ryan, CEO of TVIQ, comes at this from a different vantage point — one grounded in streaming, connected TV, and the rising economics of FAST channels. To him, AdCP could be a long-overdue bridge between content owners, distributors, and ad buyers stuck in disconnected systems. He said that CTV and streaming have been waiting for an intelligent negotiation layer, and if AdCP matures the way it’s envisioned, publishers like TVIQ will be able to dynamically package audiences and inventory for AI buyers — without giving up control to opaque intermediaries.

Ryan believes that what makes AdCP intriguing isn’t just its technical ambition, but its political one. For the first time, AI could give publishers leverage — a chance to meet advertisers as equals, with agentic systems representing both sides. “Publishers have been reduced to passive inventory suppliers,” he said. “AdCP could help flip that script by giving them an active, automated role in the negotiation.”

That’s also why the governance question looms so large. As Ryan pointed out, protocols like this don’t just evolve technically — they evolve politically. The same companies that help build them often try to own them later. The challenge will be keeping AdCP open as the biggest ecosystems start integrating it.

For all the optimism, the core challenges remain unsolved. Who verifies the agents? How will billing, provenance, and fraud be standardized? What happens when the machines make mistakes — or worse, optimize themselves into opacity?

Hirsch is both hopeful and blunt about what’s broken. “A lot of programmatic buying is broken,” he said. “If our industry can look at AdCP and agent-to-agent buying and selling as a way to restart — to focus on integrity and transparency — then there’s a there there.” But he’s also realistic: standardizing agent communication could just create a more efficient black box. “We can’t automate our way out of the trust problem unless the protocol itself enforces transparency,” he warned.

True agentic advertising — where AI agents operate autonomously and at scale — won’t happen overnight. Hirsch predicts it will take five years before AdCP and related frameworks like MCP (Model Context Protocol) are stable enough to be woven into daily operations. “Agentic buying and selling is going to take some time to get right and to scale,” he said. “At the very least, I can predict that AdCP, agentic advertising, and MCP will be the topics du jour at tentpole events such as POSSIBLE and Cannes. By the way, if anyone needs help, I’m a PAC — Professional Acronym Creator.”

It’s that kind of self-aware humor that grounds the hype. For all the talk of revolution, AdCP’s early supporters understand that this is a marathon, not a launch event. The technical ambition is real — but the commercial and political work will decide whether it becomes the backbone of the open internet or another protocol captured by incumbents.

Still, something important is happening. AdCP might just be the first credible attempt to teach machines the language of media buying — and make them accountable for what they say.

And for the first time in a long time, the industry seems ready to listen.

Robert Webster isn’t one for overhyping shiny objects—but this time, he’s cautiously lighting a flare. His first take on the Ad Context Protocol (AdCP) reads less like a press release and more like a blueprint for the first real reset button the open ad ecosystem has seen in a decade.

He describes AdCP as “agentic curation at scale”—a framework that finally lets advertisers and publishers speak the same machine language without the middlemen garbling the message. Built atop Anthropic’s Model Context Protocol, AdCP turns every campaign into a structured, auditable conversation between buyer and seller agents.

Webster’s optimism isn’t blind. He breaks it down with pragmatic precision:

Webster sums it up as a “bloody good way” to rebuild trust and value across the supply chain—if, and only if, the industry doesn’t smother it with bureaucracy before it takes root.

Bottom line: AdCP isn’t just another acronym—it’s a potential operating system for an honest ad economy.

AdCP is being pitched as programmatic’s next evolution—not its replacement, but its grown-up successor. It’s the bridge between the brittle plumbing of OpenRTB and a future where AI agents plan, negotiate, and execute media without waiting for a human to find the “insert creative” button.

The idea sounds clean. The reality, as always, is messy. Every “open standard” starts as a manifesto and ends as a moat. That’s why “cynical optimism” is the only responsible posture here—cheering for innovation, but keeping one hand on your wallet.

OpenRTB gave us the language for automated auctions—impressions, bids, responses. It standardized the transaction, not the thinking. It’s the protocol equivalent of a fax machine: efficient, but dumb.

AdCP, by contrast, wants to standardize the intelligence layer—how agents discover, plan, negotiate, and report. It’s not just “RTB 4.0.” It’s the first step toward machine-to-machine media strategy.

Where OpenRTB focused on the “what” (inventory, bid, price), AdCP focuses on the “why.” Its primitives—“Get Products,” “Create Media Buy,” “Sync Creatives”—aren’t just technical endpoints. They’re a declaration that buying media should finally reflect intent, context, and fluidity.

As PubMatic’s leadership explains it, this isn’t a replacement—yet. It’s “both.” A supplement for now, a succession plan later. The two protocols will coexist, like parallel nervous systems: one running the heartbeat of auctions, the other teaching the brain to think.

That duality isn’t a hedge. It’s a timeline. OpenRTB is the old pipes. AdCP is the new water.

“Agentic” is the latest buzzword, but here it actually means something. It means the difference between a static spreadsheet and a living system.

Imagine an advertiser prompt: “Find adventure travelers in the U.K. and Germany, target them with dynamic video, and stay under an $80 CPA.”

Under AdCP, an agent doesn’t just bid—it builds a campaign on the fly, pulling from publisher APIs, contextual data, and creative templates. It assembles, negotiates, executes, and learns—all while feeding real-time reporting back into the system.

Media planning stops being a quarterly ritual and becomes a continuous dialogue. IO execution isn’t paperwork—it’s protocol. And reporting becomes conversation, not postmortem.

It’s the same dream programmatic once sold—automation, efficiency, transparency—but this time, it might actually deliver, because the intelligence lives in the protocol itself, not in some black-box bidder.

AdCP’s primitives aren’t the boring syntax of engineers—they’re the DNA of the next media marketplace.

Get Products: Forget static deal IDs. Agents can dynamically query what inventory exists, what formats are supported, what currencies are accepted, and what the pricing models look like. Discovery becomes negotiation.

Create Media Buy: Instead of bid requests and responses, agents co-design flexible deals—bundles, packages, or adaptive buys that evolve as performance data changes. Planning becomes execution.

Sync Creatives: This is the creative handshake. Assets, metadata, and brand guidelines update continuously. A campaign doesn’t “launch”—it mutates. Optimization becomes culture.

By encoding intelligence at the protocol layer, AdCP pushes strategy closer to the wire. The logic that used to live in dashboards or demand platforms now lives in the network itself.

That’s liberating—for now. Until, inevitably, someone adds “extensions.”

PubMatic is playing this exactly right. They’re calling AdCP both a supplement and a replacement, depending on who’s listening. It’s a diplomatic way of saying: we know OpenRTB isn’t going away overnight, but we’re not betting our future on it either.

The pragmatic reality is hybridization. For years, we’ll see OpenRTB handling the transactional volume—the constant hum of bids and impressions—while AdCP orchestrates the strategic flow above it.

The early adopters—Scope3, Optable, Triton—will run AdCP for agent-driven planning and reporting. Everyone else will pretend to, while quietly waiting to see if the bots break anything.

But make no mistake: this is a power shift. The companies building AdCP aren’t just writing a protocol—they’re defining how automation will express value. Whoever controls that syntax controls the next ad economy.

If history has taught ad tech anything, it’s that every “open” standard eventually becomes someone’s gated garden.

First, there’s the collaboration stage—press releases, GitHub repos, kumbaya panels. Then come the extensions, the proprietary hooks, the “enhancements” that only one vendor can run. Before long, openness becomes optional, and the new open protocol starts looking suspiciously like the old walled garden.

That’s why AdCP inspires cynical optimism. The optimism: this could finally give the open internet a fair, intelligent protocol to compete with the walled gardens’ data advantage. The cynicism: we’ve seen this movie before, and it ends with one company calling itself “the ecosystem.”

AdCP’s potential is undeniable. It could:

But its pitfalls are just as clear. Agentic advertising could birth new opacity as AI agents optimize in ways humans can’t trace. It could reintroduce data hoarding under the guise of automation. It could create a new hierarchy—those who train their agents faster versus those who get trained by them.

This is the paradox of AdCP: it’s the most promising thing to happen to programmatic since Prebid—and the most dangerous if left to evolve unchecked.

So yes, AdCP is the heir. But heirs inherit both the power and the dysfunction of their lineage.

The machines are ready. The standards are coming. The only question is whether the humans will resist the temptation to close the door once they’ve built the new house.

Not everyone is buying the optimism around AdCP and its backers.
In a recent LinkedIn post that quickly circulated across ad tech circles, Dr. Augustine Fou, founder of FouAnalytics and one of the in dustry’s most persistent critics, issued a blunt warning to advertisers: “Don’t let adtech bros sell you the next black box of snake oil, wrapped as a shiny object. It will just continue to waste your ad budgets and accelerate the flow of money into their pockets.”

Fou didn’t mention specific companies in that post, but across multiple threads, he has repeatedly referenced Brian O’Kelley, Scope3, Swivel, and others tied to the emerging AdCP consortium. His tone was consistent — deep skepticism toward any “new” transparency solution not directly owned or operated by advertisers or publishers.

Fou’s core argument is simple, if familiar: if advertisers and publishers don’t control it, it’s just another toll booth on the supply chain.
In his own words, “As long as it is NOT controlled by the advertisers or publishers, it’s still a middleman toll-taking mechanism, no matter what acronym you give it.”

That perspective taps into a long-running distrust of ad tech’s “openness.” To Fou, protocols like AdCP risk becoming just another black box wrapped in open-source branding—a system that promises transparency but ends up serving the same entrenched intermediaries.

Executives close to the AdCP initiative quietly reject Fou’s characterization, calling it premature and misleading. They note that the protocol is still in early-stage design and collaboration, with no commercial model, governance board, or monetization framework in place.

You just read the free version — but the real story starts inside ADOTAT+, where the filters come off, the sources go on record (or off, when it matters), and the analysis cuts through the PR varnish.

You’re missing the internal memos, the Slack screenshots, and the whispered power plays behind AdCP — who’s quietly rewriting the governance charter, which holding company is already testing agentic pipelines, and how “open” this protocol actually is when money and machine intelligence meet.

If you care about where advertising power really moves next — the companies building the bots, the protocols that will run them, and the people trying to control the flow — you can’t afford to sit out.

ADOTAT+ subscribers get the unfiltered versions: the politics, the financial incentives, and the documents the press releases don’t mention. Everyone else? They’ll be reading about it three months later, on someone else’s timeline.

Subscribe to ADOTAT+ — because by the time Kokai decides what “quality” means, it’ll already be scoring you.