Agentic Media Advertising Trading
Session details:
This paper introduces a modern communication architecture for agentic media advertising trading, in which AI agents participate directly in the buying and selling of advertising inventory alongside traditional B2B and email channels. The proposed framework enables seamless, structured, and authenticated exchanges between advertiser, agency, and media property agents through emerging protocols such as Agent-to-Agent (A2A) and the Model Context Protocol (MCP). Advertiser brands register with media organizations to obtain secure authentication tokens, allowing agency-side AI agents to issue briefs and RFPs that combine structured campaign parameters with descriptive objectives and performance standards. Media property agents autonomously review these briefs, identify missing or ambiguous information, request clarification, and return optimized product offerings and pricing. Agency agents evaluate the aggregated options, assemble candidate campaigns, and—subject to configurable managerial approval thresholds—transmit purchase instructions back to the media properties.
By enabling AI agents to interpret hybrid data inputs, generate targeted proposals, engage in iterative clarification, and autonomously execute low-risk transactions, the workflow significantly reduces cycle time and enhances decision quality. Dynamic approval controls ensure that campaign size, complexity, and strategic significance determine when human oversight is required, preserving managerial governance while maximizing automation efficiency. The adoption of discoverable communication interfaces via A2A and MCP ensures reliable, interoperable messaging across heterogeneous systems. This paper details the communication flows, decision models, and operational mechanisms that enable agentic media trading, demonstrating how expanded connectivity between buyers and sellers increases market liquidity, broadens demand access for media owners, and gives advertisers more flexible and optimized campaign choices.