How a Brand Should Think About AI in Production.

The advertising industry spent roughly thirty years taking itself apart. Media split from creative. Production split from both. Specialized agencies multiplied. Holding companies grew into loose federations of businesses that technically shared a parent but often competed for the same clients. It made sense at the time. Specialization had real value. The model made sense when the disciplines were complex enough to warrant it and the volume demands were manageable enough to absorb the gaps between them.

AI is making that logic harder to defend.

In the last twelve months, every major holding company has announced some version of the same move. And notably, it is the production entities leading the charge. WPP consolidated its entire production operation into WPP Production, built on its AI platform WPP Open. Publicis Production launched Marcel Make, connecting audience data, creative performance, and content production into one agentic workflow. Dentsu launched Content Engine, pulling planning and production into a shared pipeline. Omnicom completed its merger with IPG and is consolidating two massive data and technology stacks into one. There are other agency networks creating solutions to sell, but while the details differ, the thesis is the same: the future is one system, not several talking to each other. AI is what finally makes that possible.

This is not just org chart shuffling. Production has always been the discipline closest to the actual content; the place where a brand idea gets translated across every tier, from hero campaigns down to always-on performance assets. But those tiers were often as siloed as the disciplines around them; different teams, different vendors, limited feedback between what performed and what got made next. AI is what connects them into one continuous system, where the making informs the thinking as much as the other way around. 

Everyone wants to know what AI means for live action, photography, and post-production. That is where the work gets made and those questions are worth asking. But they are harder to answer without understanding what is shifting at a higher level first; in how production is now connecting disciplines that were never connected before.

‍

The Failure of the Linear Workflow

The volume demands on marketing teams have changed fundamentally. Paid social, online video, retail media, connected TV, all of it requires more asset variations, distributed more frequently, optimized against more signals than the old production model was ever designed to handle. The traditional workflow, where strategy happens, then creative happens, then production happens, then distribution happens, was already straining before AI. Now marketing teams find themselves in an entirely different era.

What AI changes is not just speed. It changes what production can do when it is no longer the last stop in the chain. When content performance signals feed directly back into what assets get created next, and when production can scale asset variations across all three tiers without starting over each time, the system gets smarter over time, compounding what it learns across every campaign. The holdcos are betting that owning that end-to-end system is the next source of competitive advantage, but that shift brings a new set of questions that no one has fully answered yet.

‍

AI as a Commodity: The True Value of Data

The tech platforms offered across holdco reorgs differ in their specifics but share a common architecture. At the center is a proprietary AI layer built on years of accumulated campaign intelligence, consumer data, and market signals. That layer connects what were previously separate disciplines: strategy and insights, creative development, production, distribution, and measurement. The pitch to brands is that every data input informs every creative output, and the intelligence compounds over time.

Publicis Production's Marcel Make is built on CoreAI, Publicis's intelligence backbone, drawing on Epsilon's 2.3 billion consumer profiles and layering in creative performance data to inform what content gets made and for whom. WPP Open is designed to unify creative, media, production, and enterprise capabilities under one agentic platform. Omnicom's Omni platform is absorbing IPG's data infrastructure as part of the merger integration. The real bet across all of them is on the data itself, and their control over it.

AI itself is increasingly a commodity. The generative tools, the optimization layers, and the agentic workflows are accessible to any brand with the appetite to build. What is harder to replicate is the data underneath. Years of audience signals, campaign performance, and market intelligence is what makes these platforms actually intelligent. That is the asset everyone is competing to control, which is worth understanding before you decide how much of your creative performance history, audience data, and production assets to store inside someone else's system.

‍

What Consolidated Systems Offer & What to Watch For

For brands, a connected system offers things that were hard to get before. Faster time to market happens when brief, production, and distribution share the same pipeline. Creative informed by real performance signals instead of post-campaign analysis. Scale that no longer requires proportional cost growth. And clearer accountability, because when one partner owns the end-to-end, there is less room for work to get lost between outside suppliers and contributors.

For brands running high-volume, always-on programs across multiple channels, the efficiency gains when tapping into integrated AI solutions are real. The holdcos are not wrong that this is where client demand is heading. The platforms are easy to get excited about. The implications take longer to see.

To understand the impact of full integration, think about the Apple versus Android decision. You can switch. Nobody is stopping you. But after a few years inside one ecosystem, your photos are in iCloud, your apps are tuned to one OS, your muscle memory is built around one interface, and the people you message are on one platform. The switching cost is not a wall. It is a thousand small frictions that add up to the same thing.

Holdco AI platforms are starting to work the same way. When your campaign history, your creative performance data, your audience signals, and your production assets all live inside a proprietary system, you can leave. But you lose the institutional memory. The model that has been learning from your campaigns does not come with you. Neither do the custom workflows, briefing processes, and approval structures your team has built around the platform. A few years in, leaving is a lot more complicated than it looked at the onset.

Connected systems can also obscure as much as they reveal. When the same platform is running the work and providing reporting, you rely on the platform to tell you how well the platform is doing. That built-in bias is compounded by the fact that these platforms are constantly evolving; new capabilities, new models, new workflows. The evolution is part of the value proposition. It is also what makes them harder to evaluate and govern over time. That is not a reason to walk away, but it is a reason to make sure someone outside that system is monitoring the work.

Creative homogenization is worth watching for too. When the same AI stack is optimizing content for dozens of clients across the same media environments, differentiation becomes harder to sustain. Efficiency and distinctiveness pull in opposite directions, and right now the platforms are much better at promising the former.

‍

The Right Questions Every Brand Should Ask

More brands should first ask the build versus buy question explicitly before they default into a holdco platform. Building a platform takes time, energy, and budget, but may ultimately deliver the best strategy. And for those who go with a holdco model, it is worth understanding which parts of the platform are genuinely proprietary versus built on licensed or third-party technology.  

There is real risk to be considered here; the long-term viability, stability, and ownership of those underlying platforms can directly impact what a holdco can actually deliver, particularly in a fast-moving and still-maturing ecosystem. That distinction also matters when you start asking how the platform connects with tools you already have in house, because not everything is designed to talk to each other, and the gap is usually your problem to solve, not the platform's.

Then there is the question of visibility across the pipeline. When decisions are being influenced by the platform at every stage, from briefing and creative development through distribution, optimization, and reporting, you need an audit trail across the full workflow. If you cannot see what the system is doing and why at each stage, you cannot govern it — as these platforms become more agentic, that gets harder. When automated systems are making decisions across the pipeline, tracing back to where something went wrong and why is not a problem the industry has fully solved yet. It is worth asking how any platform you work with handles that before you commit.

The third question is about what happens to your data and the intelligence built on top of it if you want to move on or bring other partners in. Raw data export is one thing. But if your media performance, audience insights, and campaign learnings have been synthesized inside a proprietary platform, that synthesis does not automatically transition with you. Understanding data portability — what leaves with you, what stays with the platform, and how your reporting and insights translate into your own systems — should be part of the conversation before a brand signs, not after.

Start your next platform conversation with the right questions: 

• What is actually proprietary versus built on licensed technology? 
• What does the audit trail look like across the full pipeline? 
• How does this platform connect with what we already have?
• What does data portability look like at contract end? 
• Who is responsible for AI governance and rights management? 
• And who is measuring outcomes; the same party running the work or someone independent?

‍

Be wary. The networked agencies are no longer agents of the brand. They are making a land grab on data and bringing as much production in house as you will allow in order to increase their revenue numbers, which is not always in the best interest of the brands.”- Jillian Gibbs

‍

The Content Supply Chain Tug of War

There is real positive momentum here. When production becomes the connective tissue between data, creative, and media, operating as one connected system, brands can finally keep pace with the volume and complexity of content modern marketing requires across every tier. That part isn’t spin.

But most large marketers are not starting from zero. They have a CDP they bought three years ago, a DAM their teams actually use, measurement tools they trust, and in some cases their own AI workflows they have spent real money building. The holdco platform wants to sit at the center of all of it. So does Salesforce. So does Adobe. Amazon and Walmart Connect are building the same thing from the retail side —  closed loops that run from purchase data to creative to screen. There is a multibillion-dollar business opportunity in being your operating system, and everyone knows it.

The brands that fare best will be the ones who understand their own content supply chain before they plug into someone else's.

‍

Steve Intrabartola, a Senior Advisor at APR, specializes in integrated production focused on AI-driven creative and production workflows.

‍