AI Meets Adtech
Face it, everyone agreeing with everyone else is boring. Conference panel conversations often devolve into polite comments made not to offend other panelists. No matter how much moderators try, some fail to get a certain tension going during the conversation. During an “AI Meets Adtech” panel at Streaming Media 2025 in Santa Monica in October, I did my best to stir things up with a group of industry thought leaders that included Google’s Inderpreet Sandhu, Tavant’s Filiz Bahmanpour, FOX’s Amit Shetty, IAB Tech Lab’s Shailley Singh, and former Disney tech ops expert Sarge Sargent.
The topics we covered on this panel included the following:
“Every ad server is going to have some machine learning in it to build its algorithms and keep feeding and training its algorithms,” said Sarge Sargent, former Director, Ad Platforms Technical Operations at Disney. Ad servers have always managed transactions this way, so in this case, AI/ML is already in use.
“I think that AI and advertising are all about tools to help us do things better,” said Amit Shetty, Sr. Director, Video Ad Technology, FOX Corporation. “So you are probably talking about the same set of things that we always were doing, whether it is measurement, delivery, ad fraud detection, or brand safety. AI is able to help you to do those tasks much better, hopefully cheaper though that part of it is still being sorted out. Basically, [it’s] a tool that helps various use cases, not a single thing.”
Takeaway: AI is fixing existing use cases.
Changing Times
When the conversation turned to direct and programmatic advertising, Tavant Head of Media & Entertainment Filiz Bahmanpour commented, “I think the last stat was 58 percent of video and TV inventory is now programmatic, but there’s a big portion that’s still direct I/O, which is highly manual. Deals are set up in the CRM and then replicated in the online management system and then replicated in the ad server menu with ad ops people and a large group of ad ops people and whenever there’s an issue within the campaign, the error troubleshooting is highly manual today as well.”
She added that Tavant is beginning to see “interest from some of the CTV players in being able to talk to a chat GPT application and inquire about how the campaign is doing. They want to experiment with it and hook up to their knowledge management, whether it's Confluence or another sort of ad server or CRM system.”
Takeaway: AI is improving the direct-buy workflow.
“Basically, you’re going to put a lot of people out of the jobs because AI can do some of those ad operational jobs,” Sargent said. “Troubleshooting campaigns and campaign management are a very manual process, but with agentic workflows, if your models are properly trained and you take the data, you can train your agents to actually do the same job that a team of people can do.”
Takeaway: Job losses are coming.
Live
Inderpreet Sandhu, Global Head of CTV Partner Development and Growth, discussed how AI-driven changes in adtech are impacting live streaming workflows and monetization. “Publishers have given us feedback like ‘We want Google Ad Manager to develop tooling to help us be more efficient,’” he said. “A lot of ad ops teams are very lean to start. They’re not very well staffed. A lot of people ask us to help them do more with the same amount of resources that they currently have, especially as they’re distributing content all over the world. A key example I’ve been really leaning into is around live streaming, and live workflows,” Sandhu continued. |How do you monetize a live event that’s distributed potentially across hundreds of countries around the world on all these different devices with different internet capabilities?”
Google's Inderpreet Sandhu (left), Tavant's Filiz Bahmanpour (right)
He recounted a recent use case involving DAZN, with whom Google partnered on the club World Cup. DAZN streamed “across nearly every continent in the world,” he said, “400-odd countries and dozens of device configurations and they really leaned into us to try and utilize as much of the platform to help them find issues before they became issues.”
Looking ahead, he said, “If live becomes more part of the regular content diet across the world, you’re going to need to have tools that are invisible on an everyday basis but are [also available] when problems or potentially opportunities come about. I think that’s what a lot of us on the platform ad-serving service side are building towards.”
Takeaway: AI will impact live ad insertion monitoring at a worldwide scale.
Brand Safety
FOX’s Shetty said he had seen “a big shift in the kind of insights that we could learn in terms of what’s in a piece of content, whether it’s text content, image, or video, and being able to dig into and understand the nuances of what it means. You have your general categorization—here are the dangerous categories, but you also had sentiment analysis. Is it in the context of news versus propaganda, or what have you? There’s so much benefit that FOX has been getting on that front.”
Referencing Sandhu’s point about the leanness of current ad ops teams and the need to get more out of them without allocating more budget or other resources, he said, “The tools that we now have allow for these ad ops teams to set up more quickly,” and leverage Generative AI to identify audiences of interest, gauge their potential for reaching enough inventory, and assuring brand safety. “Using Gen AI,” he said, “this is easier now to set that up than it used to be in the past.”
Takeaway: AI helps provide better brand security.
IAB Tech Lab Weighs In
Shailley Singh, EVP, Product & COO, IAB Tech Lab, said that he has begun to see the application “agent techbots or agent tech workers” in real-world ad insertion workflows. LLM-based generative AI, he said, “can understand a lot of things about content and apply it to converting new taxonomies and a brand suitability. It can also understand a brand’s objective by reading a brand’s webpage and their brand policies. It can then try to match the two. There are a couple of companies that are doing that.”
FOX's Amit Shetty (left), IAB Tech Lab's Shailley Singh (right)
Beginning in September 2025, he said, IAB Tech Lab began to see “that leading companies who provide data like LiveRamp, Optimal, and others are starting to talk about agents that are autonomous and can put your data together and put your audiences together by having access to all of your partner’s data sources”
Takeaway: AI creates the potential for the North Star goal, data interoperability.
Singh continued, “I have to imagine that when advertising is purchased, someone said, ‘We need to reach a certain group, and the data about this mythical target market in many locations and various data formats.’ Before AI got involved, some human had to do the work to make sense of this.”
With the advent of these new autonomous agents, he continued, “As a brand manager, you might say, ‘I’ve just learned that people who are Taylor Swift fans also buy a lot of cruise tickets. Can you go and find where I can buy these audiences or lookalikes?’ [And the agent could] go and access Experian, LiveRamp, all of these other data sources and it might take a day to process it and find this audience. Now, this agent may be talking to agents at the other end. Those data providers may have their own agents. That's the kind of world that we’ve started to hear about now."
Takeaway: AI agents can process disparate data at scale.
What Should You Know?
So, what do adtech pros need to know about the AI-driven changes that are coming, and in some cases already here?
The first, said Sargent, is “just understanding what AI is like, and how the different parts of AI apply to your specific problem. The second thing is to identify which tools to use, or even if AI makes sense to resolve that specific issue you have because it may not. You want to verify or validate what the AI is producing for you. Using ML to kind of do numbers crunching, you still have to validate that that is the actual right data set and that the model is correct.”
Related Articles
With an eye to providing more insight into real-world deployments and the operational side, online adtech firm Magnite volunteered to give me insight into how AI fits into their current operational processes.
05 Sep 2025
FAST and AVOD are struggling to fill inventory and CPMs are sagging in much of the CTV world. Is programmatic the problem or the solution? Is brand or direct advertising better equipped to turn the tide? Would a hybrid strategy work better? Fremantle's Laura Florence, Roku's Charlie Goodman, Philo's Aulden Kaye Yi, and Vevo's Melissa Sofo debate the best solutions to free streaming's fill and pricing problems in this clip from Streaming Media Connect 2025.
14 Mar 2025
Brand safety in the digital ad space has always been a fluid issue, and everyone in the industry is coming at it from a different perspective. In order to mitigate issues and prevent future failures, the advertising and digital media ecosystems need total transparency, and must create a set of stricter regulation guidelines.
29 Jul 2020