L'Oréal's AI content engine — and dynamic DOOH
L'Oréal deepened its OpenAI partnership at VivaTech, powering its CreAItech engine to mass-produce campaign imagery and video. The under-discussed read for beauty DOOH: the real bottleneck on dynamic, dayparted screens has always been creative supply — not screens.
The conversation about dynamic and programmatic DOOH usually fixates on screens, SSPs and data. The quieter constraint is creative supply — you can target a salon screen by daypart, weather and locale, but only if someone produced enough on-brand, correctly-specced assets to fill all those moments. That’s why a beauty-industry AI story is also a DOOH story. At VivaTech, L’Oréal deepened its OpenAI partnership, putting the latest models behind CreAItech, its in-house engine for generating campaign imagery and video at scale. The biggest beauty advertiser on earth is industrialising exactly the input that dynamic beauty DOOH has been starved of.
What happened
L’Oréal used VivaTech to move its OpenAI relationship from experiment to infrastructure. The piece that matters for media is CreAItech — an in-house generative system, now running on OpenAI’s newest model, built to spin up marketing imagery and video at production scale. Chief digital and marketing officer Asmita Dubey framed the ambition as being “more demanding of AI to augment” the company’s consumers, métiers and employees. This is category color, not a DOOH announcement: L’Oréal didn’t say a word about out-of-home, and we assert no DOOH figure from it. But the capability it’s building is precisely the one place-based screens depend on.
Why it matters for beauty DOOH
Every argument this site makes for dynamic creative and moment-marketing runs into the same wall: dynamic only works if the creative exists. A weather-triggered SPF message, a morning-vs-evening skincare rotation, a store-locale variant — each is a separate asset, in each screen’s spec and aspect ratio, refreshed often enough to stay relevant. Historically that volume was too slow and too expensive to produce, so most beauty DOOH defaulted to one hero loop. An industrial GenAI content engine collapses that cost curve.
That intersects two things we’ve already covered. It compounds the 3D-and-motion creative finding — if motion and 3D drive the most attention but are the most expensive to make, AI production is what makes them affordable at scale. And it sharpens the AI-in-DOOH signal-vs-noise picture: the durable AI value in DOOH may be less about clever buying agents and more about content velocity — the unglamorous ability to generate, version and localise enough creative to actually feed programmatic’s targeting.
It also tells you who moves first. L’Oréal is the archetypal endemic beauty advertiser — the brand whose product belongs in the salon, the spa and the beauty aisle. When the category’s largest spender builds owned creative-production capacity, it lowers the marginal cost of putting tailored beauty creative on place-based screens, and the endemic advertiser is exactly the demand beauty DOOH inventory is built to serve.
The honest caveat
This is a bridge, not a benchmark. L’Oréal’s announcement is about marketing content broadly — social, commerce, CRM — with no stated out-of-home component, and the DOOH connection here is our extrapolation, tagged as such. AI-generated creative also raises real questions the medium hasn’t settled: brand-safety review at volume, rights and likeness, and whether “more creative” becomes “more sameness.” None of that changes the structural point. The binding constraint on dynamic beauty DOOH has been the supply and refresh of creative, not the supply of screens — and the category’s biggest player just invested heavily in fixing its half of that equation.
Related: Dynamic creative & moment marketing · FOOH vs. 3D DOOH for beauty · AI in DOOH: signal vs. noise · DOOH creative spec reference · The endemic advertiser map · Loop & creative-length study