> ## Documentation Index > Fetch the complete documentation index at: https://agenticadvertisingorg-feature-feedback.mintlify.site/llms.txt > Use this file to discover all available pages before exploring further. # AI advertising standards landscape > AI advertising standards landscape: how AdCP, OpenRTB, MCP, and A2A relate. Comparison of protocols, standards bodies, and their roles in agentic advertising. The key standards shaping AI-powered advertising are AdCP (agent workflow coordination), OpenRTB (impression auctions), MCP (AI tool calling), and A2A (agent-to-agent communication). Advertising is moving from programmatic (machine-executed auctions) to agentic (AI agents managing full campaign lifecycles), and these protocols define how the pieces connect. This page maps the landscape: what exists, who maintains it, and how they fit together. ## Active protocols ### Ad Context Protocol (AdCP) AdCP defines how AI agents interact with advertising platforms. It covers the full campaign lifecycle: product discovery, media buying, creative generation, audience activation, brand governance, and delivery reporting. | | | | ------------------- | ----------------------------------------------------------------------- | | **Scope** | Agent-level advertising workflows | | **Transport** | MCP (tool calling) or A2A (agent-to-agent) | | **Maintained by** | [AgenticAdvertising.org](https://agenticadvertising.org) | | **License** | Apache 2.0 (open source) | | **Current version** | 3.0 (GA, April 2026) | | **Key tasks** | `get_products`, `create_media_buy`, `build_creative`, `activate_signal` | AdCP is transport-agnostic: the same task definitions work over both MCP and A2A. A platform implements once; agents connect via either transport. ### OpenRTB OpenRTB handles real-time impression auctions — the bid request / bid response cycle that powers most programmatic display and video advertising. | | | | ------------------- | ----------------------------------------------------- | | **Scope** | Impression-level transactions | | **Transport** | HTTP POST | | **Maintained by** | [IAB Tech Lab](https://iabtechlab.com) | | **Current version** | 2.6 / 3.0 | | **Key objects** | Bid request, bid response, win notice, billing notice | OpenRTB and AdCP are complementary. OpenRTB handles "should I bid on this impression?"; AdCP handles "create a campaign with this budget and targeting." See [AdCP and OpenRTB](/docs/building/concepts/adcp-vs-openrtb) for details. ### Model Context Protocol (MCP) MCP defines how AI models call external tools. Developed by Anthropic, it's the standard for connecting AI assistants to APIs, databases, and services. | | | | ----------------- | ------------------------------------------------- | | **Scope** | AI tool calling | | **Transport** | JSON-RPC over stdio or SSE | | **Maintained by** | [Anthropic](https://modelcontextprotocol.io) | | **Used by** | Claude, Cursor, Windsurf, and other AI assistants | AdCP uses MCP as one of its transport layers. An AdCP MCP server exposes advertising tasks (like `get_products` or `create_media_buy`) as tools that any MCP-compatible AI assistant can call. ### Agent-to-Agent Protocol (A2A) A2A defines how autonomous agents communicate with each other. Developed by Google, it enables multi-agent workflows where specialized agents collaborate. | | | | ----------------- | --------------------------------------- | | **Scope** | Agent-to-agent collaboration | | **Transport** | HTTP + JSON-RPC with SSE streaming | | **Maintained by** | [Google](https://google.github.io/A2A/) | | **Used by** | Multi-agent orchestration frameworks | AdCP uses A2A as its other transport layer. In an A2A setup, a buyer agent sends AdCP tasks to a seller agent as structured messages, with support for long-running operations via streaming. ## Standards bodies and organizations | Organization | Focus | Key outputs | | ------------------------------------------------------------ | ------------------------------ | ------------------------------------------------------ | | **[AgenticAdvertising.org](https://agenticadvertising.org)** | AI agent advertising standards | AdCP specification, JSON schemas, client SDKs | | **[IAB Tech Lab](https://iabtechlab.com)** | Digital advertising standards | OpenRTB, VAST, ads.txt, sellers.json, content taxonomy | | **[Anthropic](https://anthropic.com)** | AI safety and research | MCP specification | | **[Google](https://google.github.io/A2A/)** | AI and cloud | A2A specification | | **[W3C](https://www.w3.org)** | Web standards | Privacy Sandbox APIs, Topics API | AgenticAdvertising.org is an independent member organization. It is not a subsidiary or working group of IAB Tech Lab, Anthropic, Google, or any other company. Its members include platform providers, advertisers, agencies, and developers. ## Other standards in the ecosystem | Standard | Purpose | Maintained by | | ------------------------------ | -------------------------------------- | ----------------------- | | **VAST** / **VPAID** | Video ad serving and interactive video | IAB Tech Lab | | **ads.txt** / **sellers.json** | Supply chain transparency | IAB Tech Lab | | **Open Measurement SDK** | Viewability and attention measurement | IAB Tech Lab | | **Unified ID 2.0** | Privacy-preserving identity | The Trade Desk / Prebid | | **Privacy Sandbox** | Cookie-less targeting APIs | Google / W3C | ## How the layers fit together These protocols operate at different layers of the stack: ``` ┌──────────────────────────────────────────────┐ │ Strategy layer (AdCP) │ │ Campaign planning, budget allocation, │ │ cross-platform coordination │ ├──────────────────────────────────────────────┤ │ Transport layer (MCP / A2A) │ │ How agents call tools and exchange data │ ├──────────────────────────────────────────────┤ │ Execution layer (OpenRTB / platform APIs) │ │ Impression-level auctions, ad serving, │ │ creative rendering │ ├──────────────────────────────────────────────┤ │ Measurement layer (OMSDK / UID2 / Privacy) │ │ Viewability, attribution, identity │ └──────────────────────────────────────────────┘ ``` A typical workflow: an AI agent uses **MCP** to call **AdCP** tasks on a publisher's platform, creating a campaign. The publisher's ad server uses **OpenRTB** to execute impression-level delivery. **OMSDK** measures viewability. **ads.txt** verifies the supply chain. ## What's changing Three trends are reshaping how these standards interact: **Agent-mediated buying.** Instead of humans navigating dashboards, AI agents will manage campaigns across platforms. This creates demand for standardized agent interfaces — which is what AdCP provides. **Protocol convergence.** MCP and A2A are establishing the transport layer for agent communication. Domain-specific protocols like AdCP build on top of them. This mirrors how HTTP became the transport layer and domain-specific APIs built on top. **Vertical specialization.** Generic agent protocols (MCP, A2A) handle communication. Vertical protocols handle domain logic. AdCP handles advertising. The same transport-plus-domain pattern may emerge in other verticals as agent adoption grows. ## Getting involved Understand the protocol architecture and core concepts. Participate in working groups that shape protocol direction. Detailed comparison of AdCP and OpenRTB — how they complement each other. Technical comparison of MCP and A2A as AdCP transport layers.