Sophie Blake
A coalition of major ad tech and AI firms has introduced the Ad Context Protocol (AdCP) — an open technical standard designed to enable AI agents to communicate directly across the digital advertising ecosystem. The initiative, unveiled on October 15, aims to modernize the infrastructure of media buying and selling as AI-driven systems increasingly take on planning and negotiation tasks once handled by humans.
The founding members — Yahoo, PubMatic, Optable, Scope3, Swivel, and Triton Digital — are joined by more than 20 additional companies, including The Weather Company, Magnite, LG Ad Solutions, Samba TV, MiQ, and Butler/Till. Each will contribute to the ongoing development and testing of the new framework, which is open-sourced and publicly available on GitHub.
AdCP’s design is based on Anthropic’s Model Context Protocol (MCP) and other emerging agent-to-agent (A2A) systems, creating a shared digital “language” for machine communication. The protocol’s first release includes standards for Audience Activation, Curation, and Media Buying, allowing AI systems to exchange structured information about audiences, inventory, and campaign objectives.
According to the consortium, AdCP is meant to serve as a bridge between today’s programmatic infrastructure and a future agentic ecosystem, much as OpenRTB standardized real-time bidding more than a decade ago. While OpenRTB facilitated ad exchanges and auctions, AdCP is designed to coordinate AI systems capable of negotiating and transacting across multiple platforms.
“Agents need to be able to talk to each other and interact,” said Anne Coghlan, co-founder and COO of Scope3. “AdCP provides the foundation for that communication — allowing publisher, advertiser, and platform agents to exchange information efficiently.”
In practice, AdCP could enable an advertiser to instruct its AI assistant with a command such as: “Find women interested in rock climbing in the UK.” The buyer’s agent would then communicate directly with publisher agents, which respond with inventory and audience options that fit the request. Humans would still approve major campaign elements such as creative and budget allocation, but the agentic systems would handle much of the negotiation and execution automatically.
The system reverses the traditional model of real-time auctions. Instead of publishers selling inventory impression by impression, advertiser agents can initiate direct conversations with publisher systems, negotiating based on audience segments, engagement metrics, or outcome-based goals. Proponents say the approach could reduce the number of intermediaries in the ad supply chain, simplify transactions, and improve transparency around how budgets are spent.
The group behind AdCP says its governance model will be nonprofit and open, allowing representation from publishers, advertisers, agencies, and ad tech vendors. The goal is to ensure neutrality and prevent any single company from controlling the protocol’s direction — a key challenge that has hindered past attempts at standardization.
Early demonstrations from Scope3 and Swivel have shown how agents could execute media buys through AdCP, including negotiating available inventory, adjusting spend levels, and logging every step for auditability. The protocol’s asynchronous structure also accommodates human oversight, allowing for manual approvals before transactions are finalized.
Advocates argue that AdCP could address one of programmatic advertising’s biggest criticisms: its opacity. By recording all agentic interactions, the system offers a verifiable trail of decisions and transactions, potentially reducing waste and improving trust between advertisers and publishers.
For now, adoption remains voluntary. The consortium expects to release expanded modules in 2026 covering creative generation and performance attribution, as AI systems take on a greater share of the advertising workflow.
If the industry embraces the framework, AdCP could become the connective layer of the next generation of digital advertising — one where AI agents, not humans, handle the bulk of media negotiation and optimization.
Comments
Loading…