Licenses, training, and internal AI education programs are now standard equipment at large enterprises. Nearly nine in ten organizations report regular AI use in at least one business function, and the share attributing any earnings impact to it has held at 37%, unchanged from a year earlier. Among the companies where AI moves earnings the most, 73% have also redesigned their workflows around it. Whether AI reaches the earnings line depends on how a company changes the work inside a single function, then repeats it across the rest.
Nicolas Comestaz is Vice President of Global Data & AI CoE at multinational beauty brand Coty. He currently leads the company's global Data and AI Center of Excellence with a mandate to enforce education, adoption, use-case prioritization, and governance across the organization. Comestaz was previously Coty's Vice President of Global Media, Communication, and Performance, a role that covered more than 60 markets. Before Coty he led digital marketing and data transformation across LVMH's Maisons. He's spent more than 20 years in beauty, luxury, and consumer businesses, and served on the World Federation of Advertisers executive committee.
"The key focus for me is helping the organization move from AI experimentation to industrialization and business impact. We're working on the workflows, function by function and market by market, in order to integrate AI deeper," says Comestaz. Coty reached that position through a rollout that put tools, education, and governance in place together. Most enterprise AI programs get stuck at adoption, but Coty's current work sits at the next stage.
Licenses, training, and governance
Coty deployed generative AI licenses across the organization, ran sessions on what people could do with AI, and built a corporate education program it calls "Supercharged with AI." Enterprise AI maturity depends on data, people, and process, and Comestaz traces Coty's position to the data side of it. "We started the journey far before the hype around AI," he explains. "By nature we are a very data-driven organization, and we had several data analytics initiatives that were live in the organization the last two years."
Comestaz emphasizes that governance is built alongside training. Each function and each market has its own people running business transformation day to day, with guidance from the center on how to use data and AI behind it. "We have deployed a hub-and-spoke model," notes Comestaz. "The central team is the hub, gathering the center of excellence, the digital AI factory, and the data governance pieces. The spokes are composed of the functions and the markets, where we have representatives in each one of these functions and markets."
Coty's open testing phase ran alongside the initiatives the company took straight to scale, and Comestaz treats what that phase revealed about his own organization as one of its returns. "A lot of initiatives we moved directly into scaling or industrialization," Comestaz adds. "But in parallel we empowered the team to test just basic capabilities, because it was a way for us to identify the builders and the champions within the organization."
Building out the roadmap
Comestaz's team begins with a discovery step, asking a function to describe how its people spend their time. Those answers shape the training itself, so a session works through the team's own responsibilities. Repeatable AI results depend on the context a tool is given. "Instead of having a generic training, we ask the team to give us a flavor of the tasks that they are doing in the day-to-day," says Comestaz. "That allows us to go through a training that integrates how they could reshape part of their tasks with solutions that are at their disposal."
Teams that want to go further move into a second step Comestaz calls practice, handled by a smaller group from the function. Coty's teams leave the session with a ranked list of the function's own problems, each carrying a candidate solution. That ranking is what the function takes into its roadmap. "We are working with them in order to create a workshop, a boot camp, or hackathon with a selected group of people within that team, in order to clearly identify and document the pain points that they are facing and the AI solutions that we could associate to those pain points," Comestaz notes. "That gives us a mapping of the art of the possibilities."
In the third step, the team carries the ranked list back to the function's leadership. Sponsorship at that level puts a named executive behind the work before any of it reaches the people who would build it. "We assemble all of this into a roadmap that is sponsored by the leader of the function, making sure that we have a structured approach to manage the demand," explains Comestaz. "The support from the AI factory might help them to build the advanced solution that they might need."
A content solution at scale
Coty's content solution is one of two examples Comestaz gives of work that moved straight from pilot to scale. The second is a partnership Coty signed to track how its brands surface in AI assistants and to act on what it finds there. "We had some initiatives that were able to industrialize very quickly. One of the examples is the content piece," Comestaz says. "We tested the solution, and it's now part of our tech ecosystem used by all the brands of the organization."
In Comestaz' example, marketing teams hold three kinds of roles: leaders, coordinators, and the producers who make the marketing material. What each group reaches for during the day is what decides which AI tool would help. Leaders reach for dashboards, and for the notes of brand review meetings. "Instead of having a dashboard that is most of the time not necessarily user friendly, having an agent allows them to chat to data inside of this dashboard," adds Comestaz.
Coordinators would draw on a different set of tools, and Comestaz raises marketing activation as his example. Someone in that role could have information from a media agency and an influencer agency pulled into one view of the marketing plan, then steer it as a whole. Both examples show what a single function would get out of the process. Today, Coty continues to deploy the approach across its functions and markets, and each function it reaches runs through the same three steps. "It's really a framework we're using in order to help frame AI demand in a way where we are able to make sure we understand the quick wins and solutions that will bring the most value for the organization," concludes Comestaz.
