Today’s conversation: Index Exchange CEO Andrew Casale and Dentsu chief trading officer Ant McDonagh break down the hope and reality of agentic advertising.
Some quick CES/Las Vegas observations:
The Sphere is the best ad canvas in the world. Kudos to James Dolan for building this.
What happened to Las Vegas is an ongoing debate. I fall on the side of it getting too greedy – they are casinos, after all – and catering too much to the high net worth. The prices here are insane.
CES’s staying power comes down to its place on the calendar. Most people I speak to find it invaluable as a whirlwind kickoff to the year to pack in 40-50 client meetings. Cannes plays a similar midyear role.
Media People stay in media. I’m always reminded at these big industry events that for all its challenges, people in media tend to stick with the industry over the years. There’s a reunion vibe at the Cosmopolitan’s Chandelier Bar.
Las Vegas hotels, which are the last redoubts of indoors smoking, have odor pockets that can surprise and alarm. I hit one that reminded me of living in Krakow in 1997.
The Rebooting partnered with Index Exchange to co-host the Innovators Unscripted gathering at Index’s sprawling 75th-floor suite that came with a massage room sensibly turned into a green room. We discussed why agentic advertising is a new tech that follows an old story in ad tech: great promise, unclear implementation, risk of domination by giants, and need for broader cooperation from others.
Agentic advertising’s promise and peril
The programmatic advertising market is having its AI moment in the form of agentic advertising. Its boosters, and even more skeptical industry veterans, follow the familiar pattern in artificial intelligence discussions by focusing on its potential. Agentic advertising has the potential to be “otherworldly” and represent a revolution in how the janky digital ad system operates, perhaps even leveling the playing field among those in the ecosystem that are not Giant Tech.
And yet, the reality is far more murky. For one, like any heralded ad tech advancement, agentic advertising must pass through a definitional gauntlet. Just what is real agentic advertising?
In a discussion The Rebooting held in partnership with Index Exchange at the Consumer Electronics Show this week, Index Exchange CEO Andrew Casale said many are conflating workflow optimization – despite programmatic’s billing as a replacement for manual, it still involves a lot of people pushing buttons – and decisioning, which is where the biggest benefits lie.
“You can get to a place where the advertising is like what you might see in something like Minority Report: the ads know who you are,” he said.
The definition matters because outside of X threads, use of AI agents is fairly limited. Eric Schmidt believes AI agents will soon be deployed by the millions and ultimately develop their own language. In the here and now, my A/R billing software has rolled out agents to… collect W-9s. I’m not holding my breath for these agents to discover the Rosetta Stone.
Yet AI is bound to have a major impact on the digital ad system. The current debate is whether it will be incorporated into the ecosystem like other advancements or mark a burn-the-boats moment.
Google, Meta and the other giant tech platforms that get most of the ad spending and buy compute by the billions will set the pace. They will lean on black-box systems, like Google’s PMax and Meta’s Advantage+ are radically simplified ad systems that use AI to find customers for advertisers across their properties. This is part of advertising’s shift to an outcomes-driven approach that has advertisers set their budgets and desired results and lets the robots take it from there. The tradeoff: Not much in the way of transparency or control, which is anathema to advertisers and their agencies.
“There is always reticence about handing control off to a black box,” said Dentsu global chief trading officer Ant McDonagh. We do see what the sales look like, but what we don’t really understand is how the AI drove the outcome. And for us, that’s a real challenge.”
Be that as it may, platforms have shifted the default expectations to be selling outcomes. That means the open web – another amorphous term that now includes streaming TV – must adapt. The dream: moving AI closer to the impression as part of the decision of what ad to serve will enable open web participants to offer a credible alternative to the powerful outcomes-based ad systems of the giants.
Doing that runs headlong into the definition feature of the open web: fragmentation. Every participant – publishers, agencies, ad tech vendors – has historically gone its own way, solving the same problems independently, guarding its own IP, and duplicating infrastructure in the name of differentiation. That model worked tolerably well in a world optimized around volume and reach. It is far less viable in a world optimized around outcomes.
“Collectively the open internet is enormous,” Andrew added. “Collectively, everyone’s data combined is also enormous. The challenge is that everyone has historically had to solve the problem independently, which means taking on the same redundant costs over and over again.”
In the near term, Andrew sees more pain as the market restructures. Some weaker publishers will be consolidated. Agentic advertising will not arrive to stop that. Over time, he sees the industry shifting from workflow optimization approaches to embracing so-called containerized agentic advertising, which is a more efficient way to move agentic tech to the complicated auction process. It’s a common-sense approach that recognizes the reality of the expense of GPUs, the energy demands and inherent redundancy.
“What if instead of everyone continuing to try to solve all these problems in a suboptimal way, we collectively come together—with standards—and use the same compute to solve everyone’s problem simultaneously?”