IAB Tech Lab CEO Anthony Katsur called the current moment a “wild west” gold rush in his opening keynote.

My headlines? There’s a lot of energy, optimism, and product development, but we’re still early. Everyone is experimenting, very very few are transacting at any kind of scale (the energy is most focused on CTV), and the honest answer to when real demand flows through agentic pipes still sounds like it’s closer to 2027 than 2026. 

Katsur made the case that without shared rules you get financial hallucinations, “semantic drift” (a goal degrading like a game of telephone between agents), “token bloat” (wasted compute), and general uncertainty — and it very much feels like we’re still in the middle of the argument. Katsur frames 2026 as “crawl” and 2027 as “walk,” and I think that’s the right expectation.

I presented the publisher view coming out of INMA World Congress Berlin. My caution was that agentic trading is one challenge inside advertising — and advertising itself is just one challenge for publishers given the forces remaking the business, LLMs chief among them.

The most optimistic idea of the day (from the publisher perspective) came from Joe Root of Permutive. Joe made the case that publishers hold genuinely valuable first-party signals (job titles, family size, financial indicators). But as an impression travels through the programmatic pipes, those signals get stripped. By the time a DSP “decisions,” it often has little more than an IP address and time of day. 

That makes the open Web look “signal poor” when it isn’t. Agentic buying, he argued, finally lets the demand side connect to the operating system publishers have built and moves the point of decision to the signal-rich sell side. That only works if publisher first-party data is clean, structured, and ready to surface. 

Stephanie Layser of AWS made the point that accelerators like her team’s “Aid for Advertising,” built on a natural-language interface, show the barrier to standing up workflows is dropping fast. It’s that kind of potential speed-to-market that makes me hesitate when advising publishers to “wait and see.”

One of the surprises of the day was seeing the buy side now making many of the same arguments publishers are. But given the increasing aggregation of ad spend with a small set of giants (Google, Meta, Amazon), it makes sense that everyone’s under the same pressure.

OMD’s Ben Hovaness put hard numbers on programmatic waste (roughly 30 cents on the dollar in transaction costs, bid-request duplication in the hundreds per impression). His Programmatic Governance Council is a principals-led table of agencies, publishers, and advertisers, with ad tech facilitating rather than setting the agenda, aimed explicitly at rebalancing spend toward the open Web. 

WPP’s Lauren Wetzel leaned into the idea of publishers as premium partners and shared how WPP is rebuilding from a federation of agencies onto a single data-and-technology stack. Her POV is that programmatic widened the gap between brands and publishers, and the fix is to treat publishers as “intelligence partners, not inventory suppliers.”

Without shared standards, she warned, agentic just automates fragmentation.

My conclusions from the day: Despite the enthusiasm from the rest of the ecosystem, I still can’t see where meaningful agentic demand comes from in the near term. Most of the agentic trading being promoted right now is better classed as experiments, tests, PR opportunities, or all three. 

The exception is focused overwhelmingly on CTV. And agentic doesn’t innately fix the problems that actually tax the open Web, like transparency and take rates. An agent running across the same leaky supply chain can just as easily add another layer as remove one.

But that doesn’t mean do nothing. In this “wild west” era, the publishers best positioned are:

1. Legible to new buyers before the rules settle.

2. Helping write the rules.

3. Experimenting with seller agents while waiting for standards to settle. 

Get your first-party data into a state you can actually expose, get into the groups and councils shaping these protocols, and push for the transactions to take place as close to your own inventory as possible — because that, as Joe Root argued, is where both the signal and the better economics will live.

If you’d like to subscribe to my bi-weekly newsletter, INMA members can do so here.

Banner photo: Adobe stock image by Divyesh.

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .