Article image
Cover generated with Midjourney, edited in Photoshop.

Fox's Agent Wants to Pick Your Substitute

Software can book linear TV now. Somebody still has to say no.

Markus Brinsa 21 Oct 22, 2026 6 6 min read Download Web Insights Edgefiles™ seikou.AI™

Verified Sources

The moment in the new Fox deals that I keep coming back to is a small one. An advertiser asks for a spot. The spot is sold out. Instead of dropping it from the order, Fox's agent proposes a replacement that reaches a similar audience. The counteroffer travels back to the buyer's system, both sides confirm, and the schedule stays full.

Simulmedia, which has been running these buys with Fox since June, describes the payoff in exactly those terms. The substitute keeps the buy anchored to the audience "so the schedule doesn't end up with a gap."

A gap is bad for the seller and usually bad for the advertiser. But a full schedule and the right schedule are two different things. The distance between them is where agencies will earn their fees in agentic TV, or quietly stop earning them.

What Fox actually announced

On October 15, Fox said WPP, Omnicom, Horizon Media, Stagwell and Simulmedia are executing live linear-TV transactions through AdStudio.

A Fox agent routes each partner into its own permissioned account, and a governance engine monitors agent activity against established protocols. Fox says it and the agency teams keep control of key decisions.

The groundwork was laid in June, when Simulmedia reported an approved, activated agent-to-agent buy with Fox. Its agent sent a buysheet to Fox's agent, which checked inventory and validated pricing against quarterly rate cards. The order passed through Fox's Pricing and Planning approval before it aired. Simulmedia now says its system, VAMOS (Verified Agentic Marketplace Order System), talks to Fox over the Agent2Agent protocol Google introduced.

None of this is magic, and Simulmedia doesn't pretend otherwise. It says the agents run on order automation that has handled linear buying for years. What's new is that software on both sides can carry the exchange without a person retyping every handoff. Stagwell's trading chief noted in Fox's release that speed has been scarce in linear. In that market, this matters.

Quotes from the agencies, numbers from Simulmedia

Read the partner material closely and one thing stands out. The only operating figures in the entire package come from Simulmedia, a TV-buying technology company rather than an agency. In a post published October 16 and updated October 19, it reported 10 campaigns on air across five Fox properties and more than 300 spot-level orders through VAMOS during the current quarter.

The four agencies sent statements, quoted in Fox's release. Omnicom says its Omni platform plans audiences against linear and connected-TV exposure data and transacts at the spot level. Stagwell says The Media Machine turns a client brief into a media plan and connects directly to AdStudio. Horizon says it is building direct agent-to-agent communication with Fox through Databricks, on top of the orchestration layer it launched in June. WPP Media's US chief media officer called the live transactions a technical milestone on the way to industry standards.

Four agencies. Zero numbers.

I don't hold that against them yet. Different routes to the same seller is what an early market looks like. Horizon routing its connection through Databricks is a reminder that Simulmedia's A2A link describes one company's setup. But a planning capability, an integration layer and a completed order prove different things. So far the agencies have mostly shown the first two.

Who represents the advertiser?

Back to the sold-out spot. Fox's agent recommends alternatives based on the advertiser's target audience. That sounds sensible until you remember who is doing the recommending. A seller has every reason to propose inventory it can supply. Audience matching tells you the replacement reaches similar people. It tells you nothing about whether the program environment is acceptable, whether the timing still fits the campaign or whether another network would have sold the same audience for less.

Simulmedia says counteroffers are weighed on the buy side against the audience the campaign is meant to reach. Good. Audience is one condition, and advertisers have others. They need to write them down before the agents start trading: what a substitute must preserve, which changes require a human decision, and what an agent may never accept on its own.

No agent should infer permission to pay more because a replacement looks attractive. Any change that touches a placement restriction belongs in front of a named person with the authority to approve it.

Simulmedia also lists two-way negotiation as work still ahead. To me, that reads as if the counteroffers so far mostly start on the seller's side, which puts even more weight on the buy-side check.

This is the agency's job now, and WPP already said so in June, writing that advertisers "need confidence that their agents operate in their interests." I agree, and confidence comes from evidence. An agency that wants credit for representing the client should be able to show how its recommendation weighed the seller's offer against real alternatives. It should also disclose any commercial arrangement that could tilt the choice. Where principal media deals are in play, that disclosure is the whole point.

An approval click proves almost nothing

A permissioned account should mean more than access. It should come with spending limits and clear escalation rules. The person approving a material change should see exactly how it differs from the original instruction.

Simulmedia says every agent exchange in VAMOS is recorded for audit and that commercial decisions stay with people on both sides. That's the right foundation.

A log only helps if a buyer can reconstruct the original request and follow each proposed change through to the final order. An approval entry tells you someone clicked. It doesn't tell you they understood the tradeoff.

There's a bigger gap underneath. Among the work Simulmedia says is still ahead is connecting what was planned with what was bought and what was delivered. Hold that next to the claim from Simulmedia's own founder, Dave Morgan, in Fox's release. He says advertisers can now optimize linear campaigns "like digital," renegotiating mid-campaign on closed-loop outcome data. Outcome-based optimization depends on exactly the plan-to-delivery link his company hasn't finished building. I'd want to see it running before I believed the comparison.

Contracts need to catch up too. Someone has to own an order that breaches instructions, and there has to be a defined way to correct a disputed transaction. Faster exchanges don't settle any of that. They just bring the dispute sooner.

Show me the error rate

Ten campaigns and 300 orders tell me the plumbing works. They don't tell me it works better. None of the announcements offer error rates, labor savings, or campaign results attributable to agentic buying. The only published counts come with no start date and no transaction value.

WPP, which is building a Buyer Agent for Video inside WPP Open, gave a timeline in June that is more sober than October's press release. It said testing would move from alpha and beta to a model that can support large-scale TV and video investment over the following six to nine months. It said findings and proposed standards would follow in early 2027. By WPP's own estimate, real scale lands somewhere between December and March.

What I want in the next disclosure is simple. I want transaction value, the share of eligible orders that actually ran through agents and how often a human had to reject or rework what the agents produced. Efficiency should be measured from brief to confirmed order, exceptions included. Any claim of better results needs a comparison that accounts for the inventory and audiences that changed along the way. A substitute spot that performs well says something about the spot and nothing yet about the agent that booked it.

Fox and its partners have built something real, and I'd rather say that plainly than pretend to be unimpressed. The agents took the paperwork out of linear TV. What's left is judgment: whether the deal on the table deserves the client's money, and a record that shows why it got it. If the record can't show that, the agent was the only one doing the work.

About the Author

Markus Brinsa is the Founder & CEO of SEIKOURI Inc., an international strategy firm advising enterprises and investors on AI risk and governance — turning emerging failure patterns into control structures, deployment standards, and decisions that stay defensible after rollout and under scrutiny. He created Chatbots Behaving Badly, a publication and podcast investigating real incidents where AI systems gave bad advice, manipulated, or failed in ways that mattered. He writes across AI failure, enterprise risk, governance, and the structural shifts underneath them. Thirty years bridging technology, strategy, and cross-border growth across the U.S. and Europe.

©2026 Copyright by Markus Brinsa | SEIKOURI Inc.
``