AI Agents Need Human Conductors to Navigate Connected TV and Commerce Media
Autonomous software programs are moving from theoretical concepts to active participants in the programmatic advertising supply chain – and that’s forcing the media marketplace to rethink things.
While the backend infrastructure to support these sophisticated tasks is accelerating, the transition requires a delicate balance between computational efficiency and traditional human oversight.
The macro trends of agentic artificial intelligence and the underlying infrastructure are pushing the industry faster and further than it knows how to run, said Mike Dupree, svp, demand, Magnite, in this video interview with Beet.TV.
Orchestrating the new automated workflow
Market forecasts indicate a massive capital influx to support this shift across the broader economy. Worldwide spending on artificial intelligence is projected to reach $2.59 trillion in 2026, marking a 47% annual increase largely driven by infrastructure investments, according to a Gartner forecast.
Dupree said the advertising sector has seen the concept of agentic software evolve rapidly, noting that in six short months, the market has gone from conception to actual tangible activation.
“We see a broad spectrum of readiness from both our supply and sell side partners and our buy side partners in terms of how people are approaching agentic,” Dupree said. “But we do know it is actively reframing and reshaping how we think, plan, buy, optimize, measure, report.”
“We want flexibility to bring whatever technology or choice you have to the table, which is why Magnite launched our orchestration layer right around Cannes,” Dupree said.
Maintaining human oversight in an automated ecosystem
Despite this automation, Dupree said maintaining a human element in the loop remains an essential part of the equation. “You still need somebody with context,” Dupree said.
“You still need somebody with taste, and you still need somebody with risk management potentially to look after and oversee this agentic activation layer.”
“AI may know how to play all the instruments, but you still need a conductor and that conductor needs to know when to start, when to stop, when to pause and let things breathe, and then when to compose,” Dupree said. “That is certainly the job of all of us as we continue to embrace this technology and apply it to our day-to-day.”
Merging high-intent commerce with big screens
Beyond artificial intelligence, though, Dupree said the convergence of retail data and connected television is creating a new dynamic for advertisers. He said marrying the storytelling canvas of television with high-intent commerce signals allows marketers to cut through legacy targeting methods and drive measurable outcomes.
“If you’re an apparel company, instead of just spraying and praying, you can now target and measure your ability to reach high-intent shoppers and see what that did in terms of offline or online sales afterwards,” Dupree said. “It is incredibly powerful and something that’s always been reserved for more native digital that is now being applied to the most important canvas of all, which is the big screen.”
To prove the efficacy of these campaigns, Dupree said the sell-side platform is pushing toward outcome-based measurement focused on identity resolution. “You can’t measure anything that you don’t have the signal on,” Dupree said. “We’ve built a conversion API which can be bespoke to pretty much any conversion event offline or online depending on what the customer’s needs are.”
Autonomous software programs are moving from theoretical concepts to active participants in the programmatic advertising supply chain – and that’s forcing the media marketplace to rethink things.
While the backend infrastructure to support these sophisticated tasks is accelerating, the transition requires a delicate balance between computational efficiency and traditional human oversight.
The macro trends of agentic artificial intelligence and the underlying infrastructure are pushing the industry faster and further than it knows how to run, said Mike Dupree, svp, demand, Magnite, in this video interview with Beet.TV.
Orchestrating the new automated workflow
Market forecasts indicate a massive capital influx to support this shift across the broader economy. Worldwide spending on artificial intelligence is projected to reach $2.59 trillion in 2026, marking a 47% annual increase largely driven by infrastructure investments, according to a Gartner forecast.
Dupree said the advertising sector has seen the concept of agentic software evolve rapidly, noting that in six short months, the market has gone from conception to actual tangible activation.
“We see a broad spectrum of readiness from both our supply and sell side partners and our buy side partners in terms of how people are approaching agentic,” Dupree said. “But we do know it is actively reframing and reshaping how we think, plan, buy, optimize, measure, report.”
“We want flexibility to bring whatever technology or choice you have to the table, which is why Magnite launched our orchestration layer right around Cannes,” Dupree said.
Maintaining human oversight in an automated ecosystem
Despite this automation, Dupree said maintaining a human element in the loop remains an essential part of the equation. “You still need somebody with context,” Dupree said.
“You still need somebody with taste, and you still need somebody with risk management potentially to look after and oversee this agentic activation layer.”
“AI may know how to play all the instruments, but you still need a conductor and that conductor needs to know when to start, when to stop, when to pause and let things breathe, and then when to compose,” Dupree said. “That is certainly the job of all of us as we continue to embrace this technology and apply it to our day-to-day.”
Merging high-intent commerce with big screens
Beyond artificial intelligence, though, Dupree said the convergence of retail data and connected television is creating a new dynamic for advertisers. He said marrying the storytelling canvas of television with high-intent commerce signals allows marketers to cut through legacy targeting methods and drive measurable outcomes.
“If you’re an apparel company, instead of just spraying and praying, you can now target and measure your ability to reach high-intent shoppers and see what that did in terms of offline or online sales afterwards,” Dupree said. “It is incredibly powerful and something that’s always been reserved for more native digital that is now being applied to the most important canvas of all, which is the big screen.”
To prove the efficacy of these campaigns, Dupree said the sell-side platform is pushing toward outcome-based measurement focused on identity resolution. “You can’t measure anything that you don’t have the signal on,” Dupree said. “We’ve built a conversion API which can be bespoke to pretty much any conversion event offline or online depending on what the customer’s needs are.”
Topics Videos
