Agentic AI is arriving inside media buying desks, but the architectures being built around it suggest the industry is more concerned with governance than speed.
WPP this week launched its Buyer Agent for Video, an AI-driven planning tool that evaluates inventory, applies audience and performance intelligence, and surfaces recommendations across premium media owners including Disney, Netflix, NBCUniversal, Paramount, and Fox. The agent recommends; humans approve. Financial commitments and campaign activation require explicit sign-off. WPP co-built the governance framework with IAB Tech Lab and Prebid.org, establishing shared protocols for how agents communicate with media owners, with audit trails and approval thresholds integrated throughout.
“Automation for us isn’t the goal,” said Lauren Wetzel, global president of data and technology solutions at WPP. “Better decision making is the goal.”
On the same day, Yahoo DSP announced its Agent Network: a 23-partner ecosystem connecting advertisers to AI-powered tools from DoubleVerify, Integral Ad Science, Kochava, MiQ, Publicis Groupe’s LiveRamp, Snowflake, and others across audience targeting, activation, creative, and measurement workflows. Yahoo frames the approach as open and interoperable, allowing advertisers to bring their own agents alongside Yahoo’s native capabilities.
The shared emphasis on governance over automation is the signal. When the first wave of agentic tools arrived in sales and customer support, the story was speed and scale. In media buying, where spend commitments run into millions and publisher relationships are contractual, the story is control. Both WPP and Yahoo building governance frameworks in the same week suggests the industry has learned from unchecked automation playbooks and is building oversight in before it becomes a compliance incident.
Both products also signal that the agentic stack is maturing beyond the data layer into the activation layer where budgets actually move, and where brand safety and spend accountability create real institutional risk if something goes wrong.
Source: Digiday
