L’Oréal expands CreAItech with OpenAI models and ChatGPT try-on plans
L’Oréal adds OpenAI models to CreAItech and plans Maybelline try-on in ChatGPT, signaling a shift toward AI-led product discovery.
L’Oréal is expanding its generative AI marketing production system, CreAItech, through a new partnership with OpenAI that it says will support product discovery and creative production at scale.
The company outlined the partnership alongside plans for Maybelline to integrate its virtual try-on experience directly into ChatGPT, and it also intends to provide up-to-date product information to OpenAI so ChatGPT can reference L’Oréal-provided notes when users ask about its products.
Table of contents
Jump to each section:
- How the OpenAI partnership fits into CreAItech
- Why product discovery and AI-led search are central to the deal
- What the numbers suggest about L’Oréal’s AI operating model
- What marketers should know about brand content in generative AI platforms
How the OpenAI partnership fits into CreAItech
L’Oréal described the OpenAI agreement as a “foundational partnership” that gives its teams access to OpenAI’s latest models inside CreAItech, the company’s in-house system used to produce still images and video for earned, owned, and paid channels, including social and e-commerce placements.
CreAItech is designed to help internal marketing teams move faster while selecting different models for different creative needs. L’Oréal says the system already incorporates tools and models from Seedance, Google (Gemini, Veo3, and Imagen), Adobe, and Stable Diffusion, and it signed a deal with Nvidia in 2025 to improve GPU efficiency as usage scaled.
L’Oréal’s chief digital and marketing officer, Asmita Dubey, framed the model mix as a practical decision, pointing to differences in strengths across models and noting an interest in OpenAI’s image model for text-to-image rendering quality.
The partnership also extends beyond marketing creation. L’Oréal said the deal includes access to GPT-Rosalind, a reasoning model built for the life sciences sector, indicating that the relationship touches R&D tooling alongside marketing production.
ContentGrip
Why product discovery and AI-led search are central to the deal
A key element of the partnership is how L’Oréal products may appear in AI-generated answers. L’Oréal plans to provide current product information directly to OpenAI so the models behind ChatGPT can draw on the company’s own notes when users ask about L’Oréal products, rather than relying only on general web training signals like product reviews, forum discussions, and wiki-style pages.
This matters because L’Oréal is explicitly positioning the deal around product discovery, not just faster creative output. Gartner’s Greg Carlucci connected the move to changing discovery behavior in beauty: generative AI is reshaping how consumers find and engage with brands, and personal care brands are seeing declines in website traffic as consumers shift toward AI-led search experiences. Gartner also cited that nearly one in five consumers now use generative AI tools to find information.
L’Oréal is also already testing paid advertising in ChatGPT. Dubey said the company has run ads in the app since April and is testing the product with CeraVe, SkinCeuticals, and Garnier brands in the U.S., while emphasizing it is still early and without sharing performance results.
What the numbers suggest about L’Oréal’s AI operating model
L’Oréal’s investment and usage data signals a deliberate operating model change, not an isolated pilot. The company’s 2025 tech investment reached around €1.5 billion ($1.98 billion), and it aims to equip its roughly 10,000 marketing staff worldwide to build assets quickly and switch between tools based on the brief.
In its full-year earnings call, CEO Nicolas Hieronimus said CreAItech reduced production costs by 40% and supported the creation of 50,000 marketing assets. If those figures hold over time, they point to two outcomes that matter for marketing leadership: a lower unit cost per asset and a higher feasible volume of creative iteration across channels where creative fatigue is common.
L’Oréal also said it has trained 70,000 staff in AI use, suggesting that enablement is being treated as a company-wide capability build rather than confined to a specialist team.
On the media side, the company increased worldwide advertising and promotions spending by 10% versus 2024 to $15.48 billion, and noted that advertising and promotions represent 32% of sales. Dubey said roughly 60% of spending goes into paid social, and the company works with 500,000 influencers and creators a year. L’Oréal also uses AI systems to help craft briefs and manage creator relationships at that scale.
Measurement remains a work in progress. L’Oréal uses a proprietary tool called BETiQ to track and optimize marketing ROI and wants to connect creative production and media decisioning through automated effectiveness feedback, but Dubey characterized that “flywheel” as more ambition than current reality, in part because AI success can be measured in many ways, from hours saved and volume produced to ROAS and click-through rate.
What marketers should know about brand content in generative AI platforms
This deal highlights a shift from using generative AI purely as a production accelerator to treating it as a distribution and discovery surface that brands need to manage deliberately.
- Model choice is becoming a creative strategy, not a technical detail
L’Oréal’s CreAItech approach assumes different models will be used for different tasks. For marketing teams, that pushes creative ops toward a “routing” mindset: choose the right model for the job, validate outputs, and keep governance consistent across tools. - Product data quality is now part of the marketing stack
Feeding current product information directly into systems powering conversational answers changes the burden on marketing, product, and e-commerce teams. The value is not only better answers, but fewer mismatches between what the brand intends to communicate and what an AI assistant synthesizes. - Testing paid placements in AI apps will need new baselines
L’Oréal is already testing ads on ChatGPT, but called it early days. Marketers should expect that benchmarks from search and social may not transfer cleanly, especially if the user journey starts with a conversational prompt rather than a keyword or feed scroll. - Content volume gains are only useful if measurement can keep up
A 40% production cost reduction and tens of thousands of assets only translate into stronger performance if teams can learn which variants work and why. L’Oréal’s ambition to connect creative generation to effectiveness data reflects the bigger challenge: aligning creative speed with reliable, decision-grade measurement.
The larger signal is that brand leaders are planning for a world where discovery happens inside AI interfaces, not only on brand sites or traditional search. That raises the stakes on structured product knowledge, rapid creative iteration, and consistent governance across a growing mix of models.
It also suggests that “marketing AI” is moving from a set of tools to a sustained operating model: training, internal platforms, partner ecosystems, and new feedback loops between content and media.
Finally, this approach implies closer coordination between marketing, R&D, and IT. When the same AI partnerships support both consumer-facing discovery and internal capability building, the practical marketing advantage often comes from execution discipline rather than any single model.
