BLOG - What Is AdCP? The Ad Context Protocol, Explained
On 16 October 2025, the first AI agent-to-agent media buy executed – real money, real inventory, real campaign – with AI agents handling creative, targeting, and approval from both sides of the transaction. The protocol that made it possible is AdCP: the Ad Context Protocol.
AdCP is an open standard that gives AI agents a shared language for discovering inventory, negotiating deals, and executing campaigns across every advertising platform and channel. It is to agentic advertising what OpenRTB was to programmatic – the infrastructure layer that makes the whole system interoperable.
This is what you need to know about it, and why it matters for email publishers in particular.
What Does AdCP Stand For in Advertising?
In advertising, AdCP stands for Ad Context Protocol – an open technical standard for AI-powered advertising, governed by AgenticAdvertising.org.
The acronym carries some baggage. Search “what is AdCP” and you’ll surface results about the Acoustic Doppler Current Profiler, an oceanographic instrument that measures water currents. Same letters, entirely different field. In the advertising context, AdCP refers exclusively to the protocol for agentic media buying and selling.
AdCP was launched in October 2025 by a founding coalition of ad tech companies including Yahoo, PubMatic, Scope3, Optable, Swivel, and Triton Digital. Since launch, the governing body – AgenticAdvertising.org – has grown to over 104 member companies spanning publishers, agencies, data providers, and technology platforms.
The protocol is open source, maintained under Apache 2.0 licensing, and governed through four equally-weighted voting classes: brands, agencies, publishers, and technology providers. No single company controls its direction.
What Is AdCP in Advertising?
AdCP is the common language AI agents use to buy and sell media. It standardizes how a buyer-side agent describes a campaign – audiences, budget, objectives, creative requirements – and how a seller-side agent responds with inventory, pricing, and deal terms. Both sides of the transaction speak the same schema, so the negotiation can happen directly, without human intermediaries translating between systems.
The problem AdCP solves is structural. Today’s advertising ecosystem runs on hundreds of fragmented platforms, each with its own API, its own workflow, its own reporting format. An agency buying across display, audio, CTV, and email touches different toolsets at every step. AI agents can automate that work – but only if the platforms they’re talking to share a common interface. Without a standard, every new channel requires a bespoke integration. AdCP removes that constraint.
In practical terms, AdCP enables a buyer agent to start with a brief – “reach first-time car buyers in Germany this month” – and have that request resolved directly against publisher inventory, with deal parameters negotiated and confirmed between agents, in a fraction of the time a manual workflow would take.
The protocol covers the full campaign lifecycle: inventory discovery, media buying, creative distribution, audience activation, and performance measurement. It works across programmatic and direct transaction models. Most global ad spend still flows through direct and IO-based deals – AdCP is built for all of it, not just the bidstream.
AdCP vs MCP: What's the Difference?
MCP – the Model Context Protocol – is an open standard built by Anthropic that defines how AI models connect to external systems. It handles the transport layer: how agents discover tools, exchange context, and interact with APIs in a structured, predictable way. MCP is the foundation.
AdCP is the advertising layer built on top of it. Where MCP is general-purpose, AdCP is specific – it extends MCP with the data structures, tools, and governance frameworks that advertising transactions require. Inventory descriptions, campaign briefs, deal terms, creative specifications, audience signals, measurement outputs – all defined in AdCP’s schema, all transportable via MCP’s underlying infrastructure.
The relationship is additive. MCP lets an AI agent connect to external systems. AdCP tells it exactly how to behave inside the advertising ecosystem: what to ask for, what to offer, what a valid deal looks like, and what governance rules apply before a campaign can launch.
For publishers and ad tech platforms, the practical implication is this: if your systems are MCP-compatible, AdCP implementation is a shorter path than starting from scratch. The transport is already there. What AdCP adds is the advertising-specific vocabulary and the governance layer that keeps humans in the loop on decisions that matter.
That governance principle – Embedded Human Judgment, in AgenticAdvertising.org’s terminology – means agents handle the operational work, while approvals, guardrails, and strategic decisions stay with people.
AdCP and Email Advertising
Email is one of the most valuable advertising channels in existence – and one of the least accessible to automated buyers. A newsletter publisher with 500,000 opted-in subscribers, strong open rates, and verified first-party audience data has inventory that premium advertisers want. But without a standard interface, that inventory is invisible to the agentic buying systems now being built across the industry.
AdCP changes that. An email publisher running a dedicated email ad server can expose their inventory through AdCP-compatible tooling, making it discoverable to buyer agents operating across every channel. The email ad serving infrastructure – the system that manages campaign delivery, targeting, frequency, and reporting inside the email – becomes the interface point through which agentic buyers find, evaluate, and transact on email inventory.
For email ad serving specifically, this matters because email has always operated outside the programmatic bidstream. Email advertising runs on direct deals, sponsorships, and publisher-controlled placements – exactly the transaction models AdCP is built to support, alongside programmatic. An email ad server that implements AdCP can field campaign briefs from buyer agents, respond with available inventory and audience data, and execute deals without the manual back-and-forth that currently makes email advertising operationally intensive.
The audience quality is already there. The first-party data is already there. The email ad server is already the operational hub. AdCP is the layer that makes email inventory legible to the buyers who are about to start looking for it through AI agents.
AdCP Founding Members: Who's Building the Standard
AdCP is governed by AgenticAdvertising.org – an independent, non-profit industry body with over 104 member companies across every part of the advertising ecosystem. Founding members committed time, resources, and governance participation to establish the protocol from the ground up. That founding cohort includes Yahoo, PubMatic, Scope3, Optable, Swivel, and Triton Digital, alongside a growing coalition of publishers, agencies, data providers, and technology platforms who have joined since launch.
Passendo is a founding member of AgenticAdvertising.org – and the founding email-native ad server in the coalition.
That distinction matters. The AdCP founding member community has been built primarily around display, CTV, audio, and programmatic infrastructure. Email advertising – with its direct-sold deals, first-party audience data, and publisher-controlled inventory – represents a structurally different surface. Having an email ad server at the founding table means email is built into the standard from the start, not retrofitted as an afterthought.
For email publishers, that means the infrastructure being designed to connect AI buyer agents to inventory is being shaped by someone who understands how email ad serving actually works. Passendo has MCP compatibility and is positioned to be the bridge between the agentic buying ecosystem and the email advertising channel as adoption accelerates.
The protocol is open. The standard is being written now. The founding members are the ones writing it.
Make Your Email Inventory Visible to Agentic Buyers
The shift to agentic media buying is already underway. AI agents are being built to discover, evaluate, and transact on publisher inventory – and the publishers who are discoverable when those agents come looking will have a structural advantage over those who aren’t.
Passendo is an AdCP founding member and the leading email ad server for publishers. If you’re running email advertising and want to understand what AdCP-readiness means for your inventory, your deals, and your revenue – talk to us.
FAQ
What is AdCP activity?
In the advertising context, AdCP activity refers to the actions AI agents perform using the Ad Context Protocol – discovering inventory, sending campaign briefs, negotiating deal terms, distributing creative, activating audiences, and measuring outcomes. Each of these steps is defined by specific AdCP tools and data structures that buyer and seller agents use to communicate. Outside of advertising, “AdCP activity” is a term from oceanography referring to measurements made by an Acoustic Doppler Current Profiler – a different instrument entirely.
Is AdCP replacing OpenRTB?
No. AdCP and OpenRTB operate at different layers. OpenRTB standardises real-time bidding on individual impressions across programmatic exchanges. AdCP standardises agent-to-agent communication at the campaign level – the negotiation, planning, and deal-making that happens before and alongside programmatic activation. AdCP is designed to work with existing programmatic infrastructure, not replace it.
Who are AdCP's founding members?
The founding members of AdCP are Yahoo, PubMatic, Scope3, Optable, Swivel, and Triton Digital. Passendo joined as a founding member of AgenticAdvertising.org, the governing body, representing email-native ad serving. The coalition has since grown to over 104 member companies.
Is AdCP open source?
Yes. AdCP is open source, published under Apache 2.0 licensing with an express patent grant from contributors. The specification is governed by AgenticAdvertising.org through four equally-weighted voting classes – brands, agencies, publishers, and technology providers. Anyone can implement the protocol; membership in AgenticAdvertising.org is about participating in its governance and development.