What does it take to promote “media high quality” in 2026 when AI is producing the content material and serving to place the advertisements?
That’s a query that Lidiane Jones, the newly appointed CEO of Integral Advert Science, should grapple with as she takes over from Lisa Utzschneider, who, it was introduced earlier this week, is stepping down after greater than seven years on the helm.
Jones is the previous CEO of courting app Bumble. She additionally beforehand ran Slack after its acquisition by Salesforce and spent greater than a decade in product roles at Microsoft.
Now, she’s bringing that have to an organization that desires to place itself not as an advert tech vendor however as a “deep tech firm.”
“You want always-on verification to drive high quality and belief for each advertisers and publishers,” Jones mentioned. “The extent of accuracy you might want to do that at scale, and the quantity of information IAS processes on daily basis, is mind-boggling.”
Non-public fairness agency Novacap acquired IAS final yr for $1.9 billion, taking the corporate non-public within the course of. There are plenty of advantages to getting off the general public markets, mentioned Utzschneider, who’s shifting right into a particular advisory position on the IAS board from now till the top of the yr.
“It provides us room to run sooner,” she mentioned. “Should you’re not tied to quarterly earnings, you could be a lot extra aggressive.”
As soon as the transition is full, Utzschneider may even advise Novacap on its portfolio corporations extra broadly. Along with IAS, Novacap owns TV promoting firm Cadent, in addition to varied know-how, monetary providers, digital infrastructure and industrial companies throughout North America.
Jones, three days within the job, and Utzschneider spoke with AdExchanger.
AdExchanger: The open net is in a bizarre and shifting place proper now. It’s contracting. How do you develop on this setting?
LISA UTZSCHNEIDER: Our purpose is to be wherever manufacturers are working their digital promoting, not simply on the open net. We’ve got options at each the pre-bid and post-bid stage, and we’ve constructed deep integrations throughout the open net with all the main DSPs, throughout the social platforms and into rising channels like audio, gaming and LLMs.
Wherever manufacturers wish to join with customers, we intend to be there with our verification and optimization.
Talking of huge language fashions, IAS has hinted at AI partnerships coming quickly, and OpenAI is the apparent assumption, though I do know you’re not naming names. Zooming out, what does a verification layer inside a chatbot really measure?
LIDIANE JONES: Shoppers are shifting quick, and so we’ve to maneuver sooner so we will be the place they’re. That’s our guidepost for product innovation and the options we provide manufacturers and publishers.
We’re already seeing brokers and AI techniques performing actions all through the stack, whether or not that’s pre-bid or post-bid, and in addition publishing throughout many channels. The variables are totally different, however we nonetheless have to ship belief. We’ve got an important bench by way of our know-how power and engineering expertise, so there’s extra to return.
However what I’ll say is that we wish to be a belief infrastructure for the AI period.
On AI‑generated content material extra broadly, some distributors are leaning towards blanket blocking – DoubleVerify has its “slop stopper,” for instance – whereas IAS takes a extra contextual strategy. Is DV unsuitable?
LU: After I converse with manufacturers about generative AI, they’re very clear on a number of issues. Firstly, they care about how their model reveals up. They need their model specs to be correct. The crimson of Coke needs to be the precise crimson of Coke, the font needs to be the actual Coke font.
After which additionally they care about ROI and outcomes.
Having a black-and-white definition of the standard of generative AI content material as we speak could be very troublesome. It’s extra vital to deal with the outcomes that the manufacturers themselves are targeted on.
LJ: Suitability is so nuanced by firm and by model. One of many issues IAS has already constructed into our options is giving manufacturers extra management to speak that nuance so we are able to characterize their intent as precisely as doable.
What’s the IAS imaginative and prescient for AI brokers?
LJ: There’s a broad spectrum of AI brokers, and we’re intently observing how customers are participating with them and once they’ll begin to depend on brokers extra to behave on their behalf.
We have already got a complicated platform that may establish whether or not one thing is an individual or not, and we’re how we are able to use that know-how to consider high quality, about intent and about who an agent is representing. We expect we’ve bought a fairly good benefit to assist manufacturers on that entrance utilizing the know-how we have already got.
I feel we are able to all agree that issues like viewability and fraud detection are – or must be – desk stakes by now. To not be facetious however, if that’s the case, what precisely is IAS promoting?
LJ: The house is so complicated. There’s the open net, social, walled gardens, CTV. Having an impartial verification supplier that permits manufacturers to have constant belief and high quality throughout all of that can proceed to be vital and, I’d argue, turns into much more vital as complexity will increase with AI.
You can’ve made that very same “that is all turning into commoditized” argument 10 years in the past, and simply take a look at how a lot this phase has grown over that point.
This interview has been flippantly edited and condensed.
For extra articles that includes Lisa Utzschneider, click on right here.
