The first agentic buys
Over the past year the advertising industry has started testing agents in real transactions. In November 2025 IAB Tech Lab published a framework for agents to make real-time decisions (ARTF), and in February 2026 folded it into a broader programme, AAMP, with toolkits for buying and selling agents and a registry to identify them. The organisation has said the programme will also cover linear TV, radio and out-of-home.3
Out-of-home already has cases. In June, Broadsign and Draft Digital completed what they describe as the first end-to-end out-of-home buy run by agents, on Global's inventory in the Netherlands. The agency's agent prepared the brief and the plan, the media owner's agent queried and negotiated the available inventory, and the booking was activated programmatically. Brief to booked plan took under fifteen minutes, with human review.4 The same week at Cannes, Displayce opened its planning data to agencies' AI assistants through MCP5, and VIOOH reported that its selling agent had put together more than a hundred deal packages in the first half of the year.6
- ARTFIAB Tech Lab publishes its real-time decision framework for agents.
- AAMPAgentic standards programme with buying and selling agents and an agent registry. Will include out-of-home.
- Broadsign, Draft Digital and GlobalFirst end-to-end out-of-home buy between agents. Under 15 minutes from brief to booked plan.
- DisplayceThree digital out-of-home agents available from AI assistants through MCP.
- VIOOHIts selling agent builds more than 100 deal packages in the first half.
All these transactions ran on digital screens already connected to programmatic platforms. None involved paper billboards, static urban panels or building wraps.
What an agent needs to buy
An AI agent is software that takes an objective, decides the steps and carries them out using other systems, within limits set by a person. Programmatic buying automates a bid with fixed rules. An agent can also read a brief, query several suppliers, compare, propose and, if authorised, book.
For that, the seller has to offer its information in a format a machine can read. McKinsey's report on agentic commerce sets out three conditions for sellers: catalogues with structured data an agent can read, interfaces that verify the agent's identity and purchase intent, and payments with delegated authorisation and programmable spending rules.1 Out-of-home adds a fourth condition, proof that the Campaign has run, because payment depends on it.
| Condition | Programmatic digital out-of-home | Static out-of-home |
|---|---|---|
| Readable inventoryEach unit described with structured, comparable data. | AvailableScreens are listed on sell-side platforms with location, format and availability. | ManualSpreadsheets in each media owner's own format, sent by email. |
| Identity and permissionsKnowing which agent is acting, for whom, and what it can commit to. | In progressBuying agreements on the platform; IAB Tech Lab's agent registry is under way. | ManualPrices agreed with each agency at the start of the year and confirmed by email or phone. |
| Rule-based paymentDelegated authorisation, spending limits and automatic settlement. | PartialThe programmatic platform settles what was delivered. Each buyer sets its own spending limits. | ManualInvoice after the Campaign, with payment terms negotiated case by case. |
| Proof of executionVerifiable evidence that the Campaign has run. | AvailablePlay logs from each screen. | ManualProof-of-posting photos sent by email, with no link to what was booked. |
What static out-of-home lacks
In static out-of-home, availability usually sits in the media owner's spreadsheet, the price depends on what was agreed with each agency, the booking is confirmed by email and proof of posting arrives as photos. Each step depends on a person who knows the process and who to ask.
An agent cannot work like that. Without inventory described by shared criteria, it cannot compare units from two media owners. Nor can it negotiate without written rules on who can commit to a price, or close the purchase if payment depends on an invoice and on terms argued over in every Campaign. And if proof of posting is not linked to what was booked, nobody can check what it did.
An agent only proposes units it can buy. When agencies start planning with agents, static out-of-home is likely to appear less in their plans.
What Gartner expects
Gartner expects a third of enterprise software applications to include agentic AI by 2028, up from less than 1% in 2024, and at least 15% of day-to-day work decisions to be made autonomously. In the same analysis it predicts that more than 40% of agentic AI projects will be cancelled by the end of 2027 because of rising costs and unclear business value. It also estimates that of the thousands of vendors marketing themselves as agentic, only about 130 really are.2
Enterprise software applications that include agentic AI
Day-to-day work decisions made autonomously
Agentic AI projects that will be cancelled by the end of 2027
40%+
Vendors that genuinely offer agentic AI, out of thousands that claim to
~130
The adoption and cancellation figures describe the same pattern. Agents deliver in orderly processes and fail in those that depend on case-by-case conversations. For an agent to operate, the process has to be in a system first.
What to get ready
There is no need to wait for the standards to be finished. What an agent will need is the same thing that saves teams work today.
Media owners
- Describe every face with a shared taxonomy: type, dimensions, coordinates, orientation and lighting.
- Keep availability in a system that updates on every sale.
- Write down the pricing rules: open rate, prices agreed with each agency and who can change them.
- Hold the Campaign's funds in escrow from confirmation and release them when the agreed terms are met.
Agencies
- Decide what an agent can approve and what needs human review. Draft Digital kept the agency's four-eyes check in the first agentic buy.
- Set spending limits per Campaign and per client.
- Require every proposal, change and proof of posting to be recorded and linked to the Campaign.
- Work with suppliers that expose their data through an API or MCP.
Where OS.ooh fits
OS.ooh is the operating layer that covers these conditions in out-of-home. Each company works with its own roles and permissions, and every Campaign is recorded from proposal to payment. On that foundation an agent, the agency's or the media owner's, will be able to operate with rules both sides know.
Sources
- McKinsey & Company (QuantumBlack), "The agentic commerce opportunity: How AI agents are ushering in a new era for consumers and merchants", 17 October 2025. mckinsey.com
- Gartner, "Gartner Predicts Over 40% of Agentic AI Projects Will Be Canceled by End of 2027", 25 June 2025. gartner.com
- PPC Land, "IAB Tech Lab names its agentic ad initiative AAMP to end market confusion", February 2026. ppc.land
- Broadsign, "From brief to activation: inside the first fully agentic AI-powered OOH campaign", 22 June 2026. broadsign.com
- PPC Land, "Displayce launches agentic DOOH agents via MCP at Cannes Lions 2026", 25 June 2026. ppc.land
- VIOOH, "VIOOH scales agentic AI across H1 2026", 26 June 2026. blog.viooh.com



