From Programmatic to Agentic: Benjamin Masse on Building a Standard Without Bias

AAO President Benjamin Masse sits down with Justin Pearse of New Digital Age at Cannes Lions 2026 to talk governance, trust, and why the C-suite, not just the CTO, needs to own the agentic transformation.

If Cannes Lions 2025 was the year the industry talked about AI, Cannes 2026 was the year it started talking about agentic AI and, more importantly, started deploying it. That was the throughline of AAO President Benjamin Masse's conversation with Justin Pearse, Editor-in-Chief of New Digital Age, recorded on the Croisette for the NDA Podcast. Masse, Chief Product Officer at Triton Digital, founding member of the Ad Context Protocol (AdCP), and President of AgenticAdvertising.org, used the conversation to lay out where agentic advertising actually stands today, and what still needs to happen before it earns the industry's trust at scale.

A standard built to have no side

AAO exists, Masse explained, to give the industry something programmatic never quite had: a standard built by everyone, for everyone. The board is split evenly with ten advertisers, ten publishers, ten ad tech companies, and ten agencies, specifically so the resulting protocol isn't tilted toward buy side or sell side. It's also designed to work across all transaction types, not just the neatly ID-based, digitally native inventory that programmatic was built for. Radio, linear TV, out-of-home, and other traditional media have always had to bend themselves into a programmatic shape that doesn't quite fit. AAO's ambition is to start from a clean slate instead.

Clearing up what "AI" even means

Before getting to agentic specifically, Masse flagged a more basic problem: most conversations about "AI policy" collapse three very different things into one word: server-side pacing AI, generative AI that creates content, and agentic AI that executes tasks. Getting the terminology straight, he argued, is a precondition for getting the governance right. And governance, done properly, doesn't have to be as intimidating as people assume. The goal of agentic AI isn't to remove humans from advertising. The goal is to remove humans from the parts of the job that never needed a human in the first place.

The real revolution is in the back office

Asked where agentic is having its biggest impact right now, Masse didn't point to buying or serving. He pointed to the operational grind behind them. His framing of the ad tech industry as a "form-as-a-service" business landed as one of the conversation's sharpest lines: companies have spent years building applications full of complex forms and asking humans to wrestle with them, often without fully understanding what a given field or setting actually does downstream. That's not expertise. That is humans compensating for poorly designed systems. Fix the system, he argued, and people get their judgment and creativity back.

At Triton Digital/iHeartMedia, that's already showing up concretely: agents are helping with strategy, discovery, and planning, turning radio plans that used to take weeks into a matter of hours or days. Major cloud providers have deployed AdCP's seller agent more than 20 times each through their professional services teams. This is evidence, Masse said, that this isn't a future-state conversation anymore.

"Embedded human judgment," not "human-in-the-loop"

Masse pushed back on the industry's default language for AI oversight. Human-in-the-loop, he said, implies humans finding out about a decision after it's been made — like a child telling a parent about a broken lamp. AAO's principle is different: embedded human judgment, where approval is built into the process before an agent acts, not bolted on after.

The mechanics are concrete. If a buyer's weekly radio spend is set to move from $10,000 to $100,000, both the human buyer and the human seller have to approve the change, and that approval is logged in an auditable trace. On top of that, any deployment of buyer or seller agents requires a separate governance agent running in an isolated environment. This specifically prevents collusion between agents and confirms everything is operating within plan.

It's a business transformation, not an IT project

One of Masse's more pointed arguments: treating agentic as a CIO or CTO initiative is a category error. He compared its scale of impact to the arrival of the commercial internet. This is not an incremental step up from programmatic, but a shift that touches every function in a business. Citing a conversation with Prebid at the Opening Night with AAO mixer at Cannes Lions, he floated the idea of "CPMO" (product and marketing operating as one function) as a better model than treating product as downstream of IT. If the CEO and CFO don't understand agentic, he said, the whole company risks being steered in the wrong direction.

That's also why AAO's mission is explicitly twofold: the AdCP protocol itself, and education, including an academy with certification, aimed at both individual and corporate members. The risk isn't that agentic makes people obsolete, Masse argued; it's that people who don't understand the shift won't be able to demonstrate their value inside it, echoing what happened to some roles when the internet arrived.

Where agencies and sales houses still win

Does agentic threaten agencies? Masse's answer was a qualified no. Because embedded human judgment requires a human to set the inputs and rules an agent operates within, the knowledge, experience, and critical thinking agencies and sales houses bring across multiple clients remains valuable — arguably more valuable, since that cross-account pattern recognition is hard for any single advertiser or publisher to replicate alone. He drew a parallel to programmatic's own back-and-forth: the industry swung from "we don't need agencies" to "we need agencies," and he expects a similar recalibration here.

Trust, logged by design

On trust and transparency — the issues that have dogged digital advertising for years — Masse pointed to AdCP's design choices rather than promises. Every agent operating under the protocol needs a declared identity (brands publish a "brand JSON," agents carry their own declaration), transactions are logged, and agents must pass compliance checks before they're allowed to operate at all. Logs aren't public, but they're available for audit and if a party can't produce logs when asked, Masse noted dryly, that itself is a signal something was being hidden. AAO isn't a policing body, but the protocol is built to give the industry teeth when it needs them, and agentic's multi-agent structure also reduces the risk of over-concentration in any single vendor.

The real hurdle: getting the C-suite fluent

Asked what's most likely to slow adoption, Masse didn't point to technology. He pointed to education — specifically, getting agentic understood by business leaders, not just technologists. Everyone at Cannes is already fluent in agentic on the tech side, he noted; the harder conversation is with the CEOs and CMOs who need to see it as a solution to a business problem, not a technical upgrade. Since October, AdCP has shipped three full versions and an incremental release, reaching 3.1 by June. Masse credits this pace to member companies dedicating real, full-time resources because they believe it's solving something real.

His closing point tied it back to the health of the wider media ecosystem: making it as easy to invest in journalism, radio, and TV as it currently is to spend on a platform isn't just a nice-to-have, it's core to keeping quality content and journalism funded. Standards, done well, aren't just infrastructure. They're an outcome AAO thinks is worth building deliberately, with everyone in the room from the start.

Listen to the full conversation between Benjamin Masse and Justin Pearse on the NDA Podcast.

Learn more about the Ad Context Protocol (AdCP) and AAO membership at AgenticAdvertising.org.