WPP has Open, Publicis has CoreAI, Omnicom has Omni and Monks has Flow. Mumbrella argues agencies selling their AI platforms as the differentiator are selling the wrong thing, because clients buy outcomes, not infrastructure.
When every agency's slide says AI-powered, the slide is telling you nothing. The case study next to it still has to do all the work.
The agency AI platform has become advertising's latest arms race. WPP has Open, Publicis has CoreAI, Omnicom keeps expanding Omni and Monks has Flow, with independents launching their own at pace. Mumbrella's argument is that agencies pitching these platforms as their differentiator are selling the wrong thing.
The logic is hard to fault. When every holding company has a proprietary AI stack, the stack is no longer a point of difference. It is the entry fee. A capability all the majors claim in identical language differentiates none of them, and clients cannot meaningfully evaluate competing black boxes anyway.
What clients can evaluate is outcomes. Did the work get sharper, faster or cheaper. Did the thinking improve. Did the numbers move. The platform is plumbing, and plumbing has never won a pitch on its own.
There is also a quiet economic tension underneath the pitch. Agency AI platforms compress the hours that agencies have historically billed. Selling the platform hard while defending the retainer model built on hours is a contradiction most agencies have not resolved, and clients are starting to notice.
Why it matters
Australian marketers are sitting through these pitches right now, and the AI platform section has become the least useful part of the meeting. The practical question is not whose AI is better. It is whether the agency's use of AI shows up in your outcomes and your invoice, or only in their margin.
For businesses choosing between agencies, in-house builds and hybrid models, the honest baseline is that the underlying models are broadly the same for everyone. Advantage comes from how the work is done and what proprietary data feeds it, not from the badge on the platform.
WPP Open, Publicis CoreAI, Omnicom Omni and Monks Flow. When every major network has one, the platform is the entry fee, not the difference.
What to do about it
Make agencies pitch outcomes, not infrastructure. Ask what changed in output quality, speed and cost for clients like you, with numbers. Skip the platform demo.
Ask where the AI saving lands. If tooling makes the work faster, some of that efficiency belongs in your fee or your output volume. Get the mechanism in the contract.
Interrogate data handling, not model brands. Whose data trains what, where your performance data goes and what walks out the door if you leave matter far more than which model sits underneath.
Judge the operators, not the tools. The differentiator has moved back to where it always was. The people, the judgement and the proof.
The AI arms race will keep producing platforms, and the agencies worth hiring will keep proving it with outcomes instead.