Every retail media conversation about AI eventually turns to agents. Fewer conversations ask what good looks like once they're running, or what it takes to trust one enough to let it touch a media buy. This roundtable pulls that question out of the abstract. It digs into agentic media planning and buying, leaving the agentic shopping conversations elsewhere.

Xnurta builds proprietary agents that automate retail media decisions from the platform side. WPP Media is scaling agentic workflows across a 70-person team, most of whom didn't sign up to become prompt engineers. Together, co-leads Marcus Gverzdys and Jaime Martin will walk through what's working: how to evaluate an agent's performance against a human's, where the trust threshold sits before an agent moves from insight-gathering to buying, and how agencies are explaining this shift to CPG clients who are watching every dollar.

The conversation also turns the lens toward RMNs and CMNs. As agentic buying scales, these networks face a wave of new inbound traffic — the retail media equivalent of an early 2010s startup's AWS bill spiking overnight. What does it take to build an environment that's ready for that?

Expect a practical conversation, not a hype cycle: crawl-walk-run timelines and an honest look at whether the barriers slowing teams down are technological or cultural: including the quiet fear that training an agent means training your own replacement.