This post was created in partnership with Taboola

Key takeaways

  • Agentic AI is shifting from automation to autonomy, but trust, transparency, and governance remain essential.
  • Organizations need clear workflows, quality data, and ecosystems of interoperable AI tools rather than a single platform.
  • As agentic AI pilots expand, marketers and investors will expect proof that AI improves performance and delivers ROI.

From autonomizing tasks and overhauling workflow to taking over decision-making and transactions, organizations are betting big on agentic AI-powered systems and tools to level up their performance marketing capabilities.

During an ADWEEK House Cannes Lions group chat co-hosted with Taboola, panelists discussed how agentic AI is flipping the script on performance marketing processes and where it could take the industry next.

(L-R) Omnicom Media's Jarrod Martin, Accenture's Ami Palan
(L-R) Omnicom Media’s Jarrod Martin, Accenture’s Ami Palan

The state of agentic AI

Krishan Bhatia, chief business officer at Taboola, began the conversation by exploring the key attributes of impactful agentic AI in action today.

“The autonomy is probably the most important aspect of delivering on the promise of AI,” Bhatia said. “The whole point is that AI routinizes, or basically autonomizes, routinized tasks of RFPs and plan building and optimization in flight, and it’s working.” However, he noted, when it comes to governance, guardrails, and security layers, it’s still a work in progress.

Matt Groshong, business development director of adtech and martech at NVIDIA, said the real value of agents will come when they have agency.

“Where we go with that depends on how advanced your data is that sits behind it. It depends on how comfortable you are with that transaction,” Groshong explained. “But more importantly, it has to be able to describe why it made a decision. Maybe you have the ability to undo that decision, and proper transparency, so that you can get to a point where you actually trust it.”

Taboola's Krishan Bhatia
Taboola’s Krishan Bhatia

A new perspective on process

Jarrod Martin, CEO of Acxiom and chief transformation officer of Omnicom Media, acknowledged that figuring out the best way to leverage agentic automation may require some trial and error.

“We think about traditional automation being very prescriptive about the steps we take in the hope that we get the outcome that we want,” Martin explained. “And an agent is where we’re being very prescriptive about the outcome we want, and hoping that they can go through the right steps to get to that outcome.”

However, he added, there can be wide variation in the means and methods the agent uses to reach that outcome. “Some of those may be very efficient, and some of those may be very inefficient,” Martin said.

Introducing agentic AI into your process with forethought and planning can make for a smoother transition, explained Steve Ellis, chief operating officer of Paramount Advertising.

“It is very helpful if you have your process documented as it exists today, and that your data is in great shape,” Ellis shared. “Those things make all of this go much, much faster, but also end up much, much better.”

(L-R) Paramount Advertising's Steve Ellis, Proem's Imran Khan
(L-R) Paramount Advertising’s Steve Ellis, Proem’s Imran Khan

Finding the right combination of AI tools is important as well. Because organizations have different structures and performance marketing goals, there is no one-size-fits-all single platform.

“There are different roles that so many of these agents are playing, and so you’re seeing naturally in the marketplace so many of these partners actually come together and launch something unique that hasn’t existed in the past,” said Ami Palan, CEO of Accenture Song Americas. “It’s very unique to the architecture of the particular client and what they’re trying to accomplish, so very outcome-focused, and that oftentimes means multiple ecosystems.”

Groshong pointed out that with agentic AI, everyone is learning at the same time. “This collaborative effort across an ecosystem is how we actually get to ground truth about what something is doing,” he said.

NVIDIA's Matt Groshong
NVIDIA’s Matt Groshong

Next-generation digital advertising predictions

As to what a post-agentic AI marketing world might look like a year from now, Palan said she envisions it being used less for discovery and evaluation and more for decisioning and transacting.

“The way that is going to be successful is when these agents can build trust with the customer, and that’s going to be the way the customer feels confident that what’s being done—not just recommended—but transacted on their behalf, is something they actually want,” she said.

From a business perspective, Imran Khan, founder and chief investment officer at Proem, predicted that organizations will begin looking for their ROI in the near future.

“All of the companies are investing pretty significantly in AI,” Khan shared. “We all see the promise of AI, and I think the next question is for that dollar that you are spending, are you getting the return that it promised?”

Only time will tell, but Bhatia is hopeful he’ll be reporting positive results for the customers currently in beta testing with Taboola right now. “I think next year I want to be sitting here and having proved it to all of those, plus some, and be operating at scale throughout our entire platform.”

Featured Conversation Leaders