Skip to content
AI Tools August 3, 2026

An Agent Reading a Campaign Brief Now Does What a Data Sales Team Used to Do

Publisher audience data was never unsellable. It was unsellable at the price of a human explaining it to every agency planner. That constraint just disappeared.

By The State of AI Marketing newsroom
Share
Editorial illustration for: An Agent Reading a Campaign Brief Now Does What a Data Sales Team Used to Do
Credit: JAC Growth Marketing

Charlie Morris runs partnerships and data strategy at Mediavine, an ad network sitting on top of more than 18,000 publishers. For most of the last decade, the audience data underneath those sites has been a thing his industry described as valuable and mostly failed to sell. AdExchanger reported on Monday that the buyer finally showed up, and it’s software.

Morris told the outlet the integration lets publishers curate new audiences built for one specific campaign. It unlocks “a data asset that’s been there but untapped for quite some time,” he said.

The setup is two agents talking. Optable runs a sell-side agent that reads a campaign brief and assembles matching audiences out of publisher first-party data. PubMatic runs a buyer agent that queries those audiences in real time, and the deal gets curated into a private marketplace without a person building it. The two companies announced the partnership on March 12 and wired Optable’s audience agent into PubMatic’s AgenticOS, a set of more than 20 agents the company launched at CES in January.

The data hasn’t changed in a decade. What’s new is a buyer willing to read it without being walked through it first.

Publisher first-party data has always had a sales problem rather than a supply problem. A mid-size publisher knows things about its audience that no third-party segment captures. Explaining that to 50 different agency planners, one meeting at a time, costs more than the resulting deals are worth. Morris named that cost directly:

“discoverability can happen in real time without us deploying a massive sales team to educate every agency on how we’re coming to market agentically.”

That’s the whole economic shift in one sentence. The data was never unsellable. It was unsellable at the price of a human explaining it.

Vlad Stesin, CEO of Optable, framed the same point from the product side when the partnership launched:

“This partnership ensures that publisher first-party data doesn’t just inform planning — it creates a layer of audience intelligence that directly powers activation and programmatic buying decisions.”

Now the number that keeps this honest. Optable’s integration has powered roughly half of the 30 or so end-to-end agentic campaigns run through PubMatic. PubMatic reported 30 live end-to-end agentic campaigns back in April, alongside more than 1,000 AI-enabled deals. Four months later the campaign count is in the same neighborhood. The deal volume is real and the end-to-end volume has been close to flat, which is what an infrastructure story looks like before it becomes an adoption story.

Indy Khabra, co-founder and co-CEO of Livewire, put the shift in emphasis this way:

“A lot of the AI narrative is very much demand-oriented. But I think you’re going to see more conversations balancing the demand and the activation side with the supply side going forward.”

For a founder-marketer this looks like ad-tech plumbing, and mostly it is. The part that isn’t: your own first-party data has the same sales problem Mediavine’s publishers had, and it’s about to get the same solution whether or not you’re ready for it. Every B2B company with a customer list, a webinar registrant file, or an event audience is sitting on segments it can’t activate. Describing them to a buyer took more effort than the outcome justified. When the buyer is an agent reading structured data, the description cost goes to roughly zero.

That cuts both ways, and the second edge is sharper. A machine that can discover your audience in real time can also compare it in real time against every other audience available at that moment, on price. The collapse of open-web ad supply already told publishers what happens when buyers get better comparison tools than sellers get differentiation. Agentic curation makes the comparison instant. Data that’s genuinely distinct gets found faster; data that overlaps with three cheaper sources gets priced accordingly, and nobody has to sit through the meeting where that becomes clear.

There’s also an auditability question this hands to buyers, and it’s the same one we raised about vector-based targeting. When an agent assembles an audience from a brief, the resulting segment has no name a human wrote and no definition a human approved. It has a rationale the agent can produce on request, which is not the same thing as a definition you can hold a vendor to. Publishers building this should expect the question. Advertisers buying it should ask.

The reason this matters more than the campaign count suggests is that the machine here is replacing a sales meeting nobody wanted to take rather than a person’s judgment. That’s the version of AI adoption that tends to stick, because the thing it removes is a cost everyone already resented paying. Watch the roughly-30 number through the fall. If it’s still roughly 30 in December, the plumbing works and the industry decided it wasn’t worth rewiring for.

Quoted in this story

  • Charlie Morris, VP, Partnerships and Data Strategy, Mediavine (source)
  • Indy Khabra, Co-founder and Co-CEO, Livewire (source)
  • Vlad Stesin, CEO, Optable (source)

Want your perspective in coverage like this? Get quoted.

Sources

This story is part of our running coverage: the full picture →

Get Net Effect.

The net effect of AI on your marketing: the stories that matter, twice a week, in five minutes.

More from AI Tools