Published February 26, 2024 · By CampaignsLive · Industry
In the first half of 2023, all three of the major advertising holding companies — WPP, Publicis, and Omnicom — announced significant AI initiatives. WPP signed a multi-year partnership with NVIDIA to build generative content production tools. Publicis announced Marcel-era enhancements and a partnership with Microsoft. Omnicom announced its own internal generative platform, eventually branded Omni Assist, alongside a series of partnerships with model providers.
Two years later, the announcements have been worked through into actual production capability with varying success. The pattern of what the holding companies have actually delivered against the announcements is worth examining, because the holding-company response has shaped how a meaningful share of the world’s brand creative work has been produced since.
The shape of the announcements
The three holding companies’ announcements differed in detail but shared a common shape. Each described a vertically-integrated AI content production stack, organized around the holding company’s internal data and brand-client relationships. The pitch to clients was, in each case, that the holding company could deliver creative work at higher volume, lower cost, and faster turnaround than traditional agency production, by virtue of the AI tooling sitting underneath.
The pitch had three components, each of which was plausible on its face and each of which has aged differently.
The first was that brand-client data would feed into the model. The holding company would use its visibility into the brand’s historical campaign material, audience data, and brand guidelines to produce more on-brand AI output than a general-purpose model could. This was the differentiation story relative to a brand using ChatGPT or Midjourney directly.
The second was that the holding company would centralize generative tooling for its internal agencies, giving every client of every brand owned by the group access to the same generative capability. This was the scale story relative to individual agencies building their own tooling.
The third was that the holding company would handle the rights, compliance, and disclosure complexity that individual brands and smaller agencies could not. This was the de-risking story.
What actually got delivered
In 2024 and 2025, the patterns of what each holding company actually produced became visible. The patterns are not identical, but they share a common structure.
The strongest deliverables, across all three groups, have been in pre-production and stakeholder-alignment tooling. The holding companies built credible tools for client-facing concept visualization — turning brief documents into mood and concept material that brand-side stakeholders could react to, weeks before traditional preproduction would have produced anything visible. The tools shortened the pre-shoot creative discovery phase from weeks to days. This was a real and verifiable productivity gain, and the holding companies were able to demonstrate it to clients.
The weakest deliverables have been in production-grade asset generation. The holding companies’ tools, by 2025, were producing concept-grade imagery — useful for internal review, client alignment, and pitch decks — but the production pipeline downstream of those tools was largely unchanged. The actual hero asset, the actual film, the actual print execution still went through traditional production. The AI tooling sat in the front half of the workflow.
The intermediate deliverable — variations, format adaptations, scaling a hero asset across placements — was inconsistent. Some clients of some holding companies got meaningful productivity gains here; others did not. The variation tended to track the specific agency and brand combination rather than the holding-company-wide tooling.
The rights and compliance story
The most consistent deliverable, across all three holding companies, was the rights and compliance story. By 2025, all three had built internal AI use logging, talent-scanning consent workflows, model-output provenance tracking, and disclosure-norm enforcement that approached the standards the SAG-AFTRA negotiations had set for the broader industry.
This was the most expensive piece of the work and the least glamorous. It was also the piece that most brand-side teams could not have produced themselves. The holding companies’ scale made the compliance infrastructure economical in a way that smaller agency operations could not match.
The compliance work has been, in retrospect, the most defensible argument for the holding-company stack. Smaller, more nimble agencies have arguably matched or exceeded the holding companies on the creative-tooling side. They have not been able to match the compliance side at scale.
What this means for brand-side teams
A brand evaluating its creative-production options at the end of 2025 sits in a different market than the one in 2023. Two years of holding-company investment have produced an infrastructure layer that is mostly invisible to the brand — compliance, rights, model provenance — but that is real and useful. They have also produced a creative-tooling layer that is, mostly, replicable by smaller operations using off-the-shelf tools.
The practical implication is that the choice between holding-company production and a smaller, more direct relationship with creative tooling specialists has shifted on its axis. The case for the holding company is no longer about superior creative tooling; it is about compliance and integration. The case for smaller, more direct relationships is no longer about cost; it is about creative judgment and tool fluency.
Brands that handle their own rights and compliance — or that work with specialist legal counsel — have less reason to default to the holding-company stack than they did in 2023. Brands that need the compliance layer outsourced have a stronger reason than they did in 2023.
The broader pattern
The holding-company AI play is one piece of a broader market pattern. The 2023 announcements were largely bets that AI would be a vertically integrable competitive advantage — that the holding company that built the deepest AI stack would dominate the next five years of brand creative production.
What has actually happened is more nuanced. The AI tooling itself has commoditized. The compliance and integration around it has not. The competitive advantage has moved from “having the tools” to “having the operational infrastructure to use the tools at scale, cleanly, in regulated industries, with talent management and rights compliance built in.” That is a different shape of advantage than the original announcements suggested. It is also less interesting to the trade press, which is part of why the original announcements have not aged as well as the actual deliverables have.
For the labor and rights infrastructure the SAG-AFTRA agreements set up around the same period, see The SAG-AFTRA AI Clauses, Translated for Brand Teams.