Agentic AI is the new advertising battleground. Does it have to be?

The hype around agentic AI is reaching a fever pitch in 2026, and it might be driving some ad tech aficionados a bit loony. But with hype comes competing narratives, loud debates and plenty of questions.

A big pain point across the industry right now is the emergence of agentic standards that seem to be at odds with one another, intensifying confusion around agentic AI at a time when clarity — not vague press releases claiming “firsts” — is what marketers actually need.

Let’s ask ourselves: What is agentic actually trying to solve?

Many agree that standards are a crucial first step as agentic technology plays a bigger role in digital advertising.

“Marketers must bridge inevitable AI fragmentation using neutral collaboration partners,” said Matthew Hogg, senior vice president of North American connectivity and ecosystem at LiveRamp.

“We think that a critical step here is providing the ecosystem with agentic standards that help provide safety and certainty, which will then help provide the necessary foundation to unlock the value that agentic AI promises.”

The big question is what exactly that looks like in practice. Should the industry adopt Ad Context Protocol (AdCP), which aims to standardize automated buying from the ground up? Or is the better path the Agentic RTB Framework (ARTF), IAB Tech Lab’s new agentic framework that builds on the existing programmatic infrastructure?

Execution vs. allocation

As one exec at a DSP stressed, this debate doesn’t simply boil down to “agentic vs. programmatic.”

“I don’t see agentic as a replacement for programmatic, but as an additional layer that sits above programmatic execution, where strategy and allocation are reasoned about, while auctions continue to execute deterministically underneath,” the exec said, who wished to remain anonymous.

“The real risk is not the existence of AdCP, but its misapplication: treating agentic reasoning as an execution system rather than an allocation layer that hands decisions to deterministic infrastructure.”

William Cichowski, a machine learning engineer formerly with Chalice, raised similar concerns in a LinkedIn article, questioning the value of agentic protocols that position agents as transacting real-time ad buys.

“An agent that reasons step by step over large aggregates reduces precision, creates higher costs and introduces unnecessary risk with no upside when the objectives are clear,” he wrote.

How AdCP differs from ARTF

Supporters of AdCP have insisted recently that it doesn’t compete with real-time bidding rather it seeks to complement it by introducing a new strategic layer.

Ashwin Navin, CEO of Samba TV described AdCP as sitting above RTB, making the ecosystem “flatter, smarter and less labor-intensive.”

He also emphasized allocation, saying that AdCP can introduce agentic reasoning to enable “intelligent allocation across ad platforms and marketing channels.”

Benjamin Masse, chief product officer at Triton Digital, called AdCP an “expansion beyond the execution layer.” Scope3 CEO Brian O’Kelly called AdCP the “interface” that traditional systems would execute on. June Cheung, head of JAPAC at Scope3, has said that AdCP and ARTF “solve for different problems.”

Everyone seems to have their talking points. But if all of that is true — if AdCP is an allocation layer above the auction — then why do we need AdCP? Traditional programmatic already distinguishes between strategy and execution. Agents can simply compress the gap between the two. They don’t need to replace deterministic auctions to do it.

That may explain why IAB Tech Lab CEO Anthony Katsur waves off AdCP as just another acronym, labeling it “deeply flawed” and “resource consuming.”

Katsur maintains that the Tech Lab’s existing work already addresses many of the problems AdCP aims to solve. And Tech Lab recently announced a road map for scaling agentic buying and selling across the digital advertising ecosystem, built on the foundation of ARTF and existing protocols.

“Agentic execution is already part of how digital advertising operates today,” Katsur said in a statement. Reading between the lines, it’s pretty clear: AdCP isn’t invited to the party.

He isn’t alone. Adam Heimlich, CEO of Chalice, which came on board ARTF early, previously said that ARTF is building on a “proven” foundation in the programmatic ecosystem, while AdCP “went far in trying to standardize something that’s unproven.”

What does all of this mean for media buyers?

At any rate, despite the current hype and flurry of hastily released case studies, widespread adoption of agentic media buying might be a long way off.

Vlad Chubakov, associate director of programmatic at Delve Deeper, said that he’s seen “no interest on the client side” because it’s too early.

“It will have a future in this industry, but the current state of agentic is that it’s nice to present at conferences,” he said.

In the meantime, “companies need to connect their agents to their data, which is not yet happening at scale,” LiveRamp’s Hogg said, in order “to make informed decisions that maximize performance and value.”

Further, the DSP exec said that buyers should “pay close attention to who is shaping the agentic interface layer itself to understand the incentives and governance behind them.”

“Much of the current narrative is being driven by individuals and organizations closely associated with specific protocols,” added the DSP exec. “That doesn’t make those efforts invalid, but it does raise important questions. Why adopt a new protocol rather than build on an industry-governed standard?”

Of course, none of it matters if agentic can’t compete on incremental outcomes, Joe Root, co-founder of Permutive, previously argued.

“AdCP runs the risk of publishers jumping to create agents when they haven’t fixed the foundational problem,” he said. “It could distract the industry from getting performance in line with the walled gardens.”