‘Waste in the system’: Why Georgia-Pacific is holding out on buying agents

“An agent will only automate a repetitive task,” he continued. “And if you have waste in the system, you’re automating a repetitive task where waste continues to exist. So before we even go to a buying agent we have to clean it up or else the buying agent is going to essentially deliver subpar results.”

The agent, in other words, is a means to an end, not the end in itself. Which makes the past two years of work — cutting supply-side platforms, scoring inventory before the bid and tightening control over what gets bought and from whom — less a side project and more the groundwork for it. 

“Would we consider buying agents in the future? Yes we would but it’s more important that we try to get the best quality impression at the lowest possible cost, and then at that point we try to automate that repetitive task,” said Shah.

It’s a thought not lost on other marketers. In fact, there were plenty of them saying something similar at the Digiday Programmatic Marketing Summit last December. Since then, the industry has split into two camps: One is testing agent-to-agent buying and trying to push it toward scale. Usually, it’s from the holdcos like WPP and Omnicom and ad tech vendors. The other, Shah’s camp, is more cautious. They’re the brand-side marketers wary of handing agents the keys while programmatic itself is still this dysfunctional.

“I think there’s a tendency to confuse automation with intelligence. Agentic buying won’t magically improve media performance if it’s simply automating inefficient workflows, poor data or bad commercial decisions,” said Ian Maxwell, CEO, Converge. “As ever in ad tech, the real issue isn’t the workflow, it’s the incentives. Agentic buying won’t fix a market where participants are rewarded for different outcomes. As things stand today, automating those decisions simply risks making the same mistakes faster.”

None of this means buying agents won’t become widely adopted. There’s too much momentum behind them for that. But adoption will come staggered, in pockets, rather than all at once. The risks are too big and the upside too uncertain for it to go any other way. There’s a reason so few agentic advertising tests have amounted to much yet. There aren’t even any fully-fledged standards for how one agent buys from another.

“The truth is the industry is thinking about integrating agents wrong,” said Mateusz Jędrocha, chief product officer at Adlook. “Yes, automating a task that already creates waste will continue to generate poor results. But the true value of AI agents isn’t in replicating what humans already do.”