Las Vegas hosted AdExchanger’s Programmatic AI event, and after two days of sessions and conversations, we left with a clearer picture of where the industry’s head is at and where the real work is happening.
The conversation has shifted
The most noticeable change from years past: nobody needed to be convinced that AI matters. The interesting debates were happening further upstream, around execution, differentiation, and what “useful AI” actually looks like in practice.
One phrase that kept surfacing was the shift from “human in the loop” to “human in the lead.” It reflects something we think about a lot at Dstillery: automation that accelerates doesn’t mean automation that replaces judgment. The organizations moving fastest aren’t chasing autonomy for its own sake. They’re embedding AI into the places where manual work creates the most drag.
There was also a lot of honest conversation about “pilot purgatory,” the gap between experimenting with AI tools and actually integrating them into scalable workflows. That gap is where many companies are still stuck, and it came up repeatedly as a more urgent challenge than any technological limitation.
Mark Jung on agents actually used in the wild
A highlight for us was Dstillery’s VP, Head of Product & AI, Mark Jung’s session, “Agents Built For Audience Buying Workflows… Actually Used In The Wild.” The framing was intentional: not a roadmap, not a demo of what’s coming, but a look at how agentic AI and Dstillery’s DS-1 platform are already running inside real audience-buying workflows today.
Mark walked through how Dstillery’s agentic AI advertising platform, DS-1, helps teams move faster on audience discovery, reduce activation timelines, and cut down on the manual operational work that tends to slow campaign execution. From audience creation to activation, the session showed how agentic AI can streamline workflows that traditionally take days into processes completed in minutes. The core argument: you don’t need fully autonomous systems or perfect industry standards to start generating real operational value. The work is happening now.
That landed well in the room, and it matched the tone of the broader conference. Less interest in flashy demos, more interest in what’s actually solving problems.
What we’re thinking about next
Consumer intent signals are getting more fragmented as AI assistants become a bigger part of how people discover and decide. That makes downstream behavioral modeling more important, specifically the kind that can work across fragmented signals. It’s a dynamic we’re watching closely.
There was also consistent emphasis on data quality and governance as AI takes on a larger operational role. The reality is that AI is only as strong as the data and systems behind it, and as automation scales, those foundations become even more important. For us at Dstillery, that’s where the real opportunity lies: building AI that’s not just fast, but reliable, scalable, and grounded in high-quality data.