The advertising industry is building standards for a world of AI agents

For years, advertising technology has been built around systems designed for people to operate. Buyers search inventory, compare options, negotiate terms, submit orders and coordinate with publishers through a mixture of platforms, emails and manual workflows.

AI agents are beginning to change that model.

Instead of simply helping an advertiser research a campaign, an agent can potentially discover inventory, evaluate an offer and negotiate a transaction with another agent representing a publisher or media owner. But once software starts operating across company boundaries, another problem emerges: the systems need a common language.

That is the idea behind the Ad Context Protocol (AdCP), an open standard developed and governed by AgenticAdvertising.org. The organization describes AdCP as a protocol that allows AI agents to discover, plan, buy, sell and measure media across advertising channels. Its ecosystem already includes more than 140 companies.

The development comes as the advertising industry experiments with a more autonomous model of media buying. AgenticAdvertising.org says the first agent to agent media buy took place in October 2025, involving real money and real inventory, with human oversight still part of the process.

But interoperability is only one piece of the puzzle.

An AI agent may be able to communicate with another system, but that does not automatically mean it should be allowed to purchase inventory, negotiate pricing or commit a company to contractual terms.

That distinction is becoming increasingly important as advertising moves from AI assisted workflows toward agentic execution.

From connecting agents to controlling transactions

ADvendio recently joined AgenticAdvertising.org to participate in the development of AdCP standards, positioning its technology around what happens after agents establish a connection.

The company’s argument is that agentic advertising needs an execution layer capable of translating an AI driven interaction into a transaction that respects the commercial rules already governing a media business. Its platform implements AdCP through an MCP server that allows AI agents to discover inventory, create media buys, update campaigns and retrieve delivery and financial information.

That creates a different role for advertising infrastructure.

ADvendio’s system includes a Guardrail Engine designed to enforce rules around inventory, pricing, creditworthiness and contract terms. The company also maintains a human approval requirement for external bookings, meaning an agent can discover or negotiate an opportunity without becoming the final authority over the transaction.

“We are delighted to join AgenticAdvertising.org to advance open Ad Context Protocol (AdCP) standards. We are on the verge of a technology revolution in advertising, but progress hinges on our ability to move past isolated AI solutions that can only work reliably within internal ecosystems,” said Julian Ahrends, CTO at ADvendio.

The distinction matters because advertising transactions carry consequences that go beyond an AI system producing a recommendation. A booking can affect revenue, inventory availability, contractual commitments and relationships between advertisers, agencies and publishers.

Advertising’s next interoperability problem

The emergence of AdCP reflects a broader issue facing agentic AI: individual agents may become increasingly capable, but their usefulness across an industry depends on shared protocols and rules.

AgenticAdvertising.org was created specifically to provide independent governance for AdCP and related standards, with participation from brands, agencies, publishers and advertising technology companies. Its stated objective is to prevent every connection between AI agents from becoming a bespoke integration.

ADvendio’s membership therefore represents more than another technology company joining an industry association. It places a media operations platform inside an emerging debate over how autonomous advertising transactions should actually work.

The technology may eventually make parts of media buying dramatically faster. But before agents can operate at machine speed across the advertising ecosystem, the industry still has to decide what they are allowed to do, how systems communicate and where human judgment remains mandatory.

That infrastructure is now being built alongside the agents themselves.