iHeartMedia and independent agency Butler/Till completed what the two companies are calling the advertising industry’s first agentic streaming audio media campaign, buying premium podcast and streaming audio inventory through paired AI agents rather than a human trading desk. Each company deployed its own agent, connected through a shared MCP server, to plan and execute a real campaign for a U.S. agricultural solutions client running through July and August. Total spend was modest, just under $10,000, but the structure was the point: this was direct agent-to-publisher buying, not the SSP-mediated automation Butler/Till had tested previously with partners like PubMatic.
For marketing leaders, the result matters more than the scale of the test. The AI-driven buy delivered streaming audio at CPMs 42% lower than the client’s direct-buy benchmark, and 48% of the podcast impressions landed in premium non-skippable mid-roll placements, against an estimated 33% share under the client’s traditional plan. Butler/Till’s Kristie Murphy, associate director of programmatic, and Scott Ensign, chief strategy officer, worked the buy alongside iHeartMedia chief business officer Lisa Coffey. The companies now plan to extend the same agent-to-agent structure into broadcast radio.
The original insight here is less about the discount and more about the placement mix. A human trader optimizing for a CPM target will often default to the cheapest available inventory; the fact that the agentic buy skewed toward premium mid-roll rather than away from it suggests these systems are weighing placement quality and CPM together rather than trading one for the other. That is the harder problem enterprise buyers have been trying to solve as agentic ad buying scales, and it is different from the coordination question raised when an AI agent handled its first live sports ad buy earlier this year: this time, both sides of the transaction, buyer and publisher, were represented by an agent, not just one.
Source: iHeartMedia
