R2B2 has completed what it says is the Czech advertising market's first purchase of digital advertising inventory carried out through autonomous AI agents. The pilot project involved Omnicom Media, publisher MAFRA and television provider Skylink, part of Canal+, and replaced traditional manual media buying with AI that identified inventory, negotiated terms and completed the transaction.
The test offers a concrete look at how agentic advertising-where AI agents act on a buyer's brief-could cut manual work from campaign execution. For marketers, the pilot signals a shift toward faster, more direct access to inventory without surrendering control over where ads appear.
How the agentic ad buy worked
Omnicom Media interacted with an AI buying agent through ChatGPT, providing campaign requirements in natural language. The agent then recommended inventory, negotiated with the publisher side and finalized the purchase. On the sell side, R2B2 acted as the agent connecting the buyer to MAFRA's ad infrastructure via its Model Context Protocol (MCP) gateway, using the Ad Context Protocol (AdCP) industry standard for agent-to-agent communication.
No spreadsheets or manual negotiations were involved. R2B2 said its cloud-based SaaS platform integrates server-to-server, avoiding client-side scripts and any impact on website performance. Publishers still approve campaigns and creative before activation, keeping existing brand safety checks intact.
What the participants are saying
Jindřich Jiráček, head of performance at Omnicom Media, described the approach as part of a natural evolution. "We view agentic buying as the natural evolution of digital advertising. This technology has the potential to dismantle long-standing barriers between campaign briefing, inventory selection, and execution," he said, adding that it opens new automation options while preserving human strategic oversight.
Lukáš Alexandr, CTO of R2B2, said the goal was to create a secure environment that doesn't disrupt established approval workflows. MAFRA's commercial director David Korn confirmed the pilot showed the publisher's infrastructure can support AI-driven transactions "posing no risk to web operations."
Skylink's acquisition marketing manager Tomáš Jelínek emphasized the importance of maintaining control: "What is crucial for us is that even when using advanced technology, we maintain control over where and how our brand is displayed."
Expanding the agentic buying model
R2B2 plans to add more publishers and develop the technology further. Alexandr predicted that "autonomous, agentic purchasing" will become the new market standard for trading ad space. That would mean tools familiar to many marketing teams-like ChatGPT interfaces for briefing-could directly feed into programmatic-style transactions without needing a demand-side platform the way buyers have traditionally used them.
The pilot's use of AI Agents & Automation in real ad inventory trading could shorten the path from brief to live campaign, reducing the administrative hours marketers now spend coordinating with publishers and platforms. As AI for Marketing matures, early tests like this one matter because they show where practical integration is happening, not just theoretical possibility. You can read more about the technology behind the pilot on R2B2's website.
Why this matters for marketing professionals
Agentic buying doesn't remove the need for marketer oversight-humans still set the brief, approve creative and decide strategy. But it could collapse the time between identifying an audience opportunity and activating a campaign. For performance marketers and media planners, that means less administrative friction and more time spent on message, offer and measurement. The pilot also signals that the stack for AI-driven transactions is being built now, and publishers are testing readiness. Teams that understand how to brief an agent, guard brand safety, and evaluate AI-recommended inventory will be better positioned when this method becomes more common.
Your membership also unlocks:
