AdCP is an open protocol, launched in October 2025 and built on Anthropic's Model Context Protocol, that defines how AI agents discover, compare, and activate advertising inventory — a shared language for agent-driven media buying.
Definition
Founded by six companies including Scope3, Yahoo, and PubMatic, AdCP specifies core agent tasks across discovery, comparison, and campaign activation. A deliberate design choice: its workflows are asynchronous, so a human can approve before an agent commits spend. It aims to let buyers use their chosen AI agents to connect directly to publishers and platforms instead of being locked into each vendor's interface.
Why it matters
Protocols decide what's possible before products do. If agent-to-agent buying standardizes on open protocols, the interface layer of ad tech commoditizes and value moves to whoever owns the decision intelligence. For buyers, the near-term takeaway is simpler: human-approval checkpoints are being built into the standards themselves — supervised execution is becoming the industry's default architecture, not a vendor quirk.
In agentic execution
Adlyse's model — agents execute, humans approve — is the same architecture AdCP encodes at the protocol level. The industry is standardizing on the pattern Adlyse runs in production.
Related terms
See these concepts running on your own accounts.
Start with a free audit and see what an agentic workforce would change first.
Start a free audit