Puma’s AI head says the brand is still giving ‘the keys to the consumer’ as it invests in tech like a digital concierge

The man powering Puma’s AI initiatives — Ivan Dashkov — is excited to test out different ways to use AI. But he also knows that some consumers are more hesitant to adopt the technology. “I think we’re doing a better job recently, of being able to show people tools and get them excited about it,” he told Modern Retail. “I think that’s always a challenge: how to get everybody going in the same direction with AI and see it as a positive.”

Puma is in the midst of another turnaround effort, more broadly. The company reported a loss from continuing operations of €643.6 million for 2025, down from a profit of €280.7 million the year before. In a recent letter, CEO Arthur Hoeld acknowledged that the brand had “become too commercial over the years.” “This is reflected in low brand heat, poor sales quality and a product range that did not cut through,” he continued. In 2025, he said, “The aim is to establish Puma as a Top 3 sports brand worldwide, return to above industry growth and create healthy profits in the medium term.”

Dashkov, Puma’s head of emerging marketing tech, sat down with Modern Retail to unpack the brand’s latest AI initiatives and what fans can expect in 2026. This interview has been edited for length and clarity.

How does your new AI store concierge work?

“It’s a very large screen — I think it’s about seven feet tall — and there’s what appears to be a human inside. That human is powered by AI … [and] can talk about any product in the Puma store. He’s actually a specialist in our running product. … He can double click into our different types of running shoes and what’s best for you, depending on your style of running, where you’re running and your goals for running. But he knows every single other product in the Puma store, so he can point you to lifestyle products and basketball products and talk about those and the different technologies in those, as well.

A great thing about him, too, is that he makes the store experience a little more inclusive, as well, because he can speak over 100 languages. So, if somebody comes in from France and wants to have a conversation with a store associate, but one of our associates doesn’t speak French, they’ll still have that opportunity to have that interaction with the digital human. … [The concierge] also calls into our inventory API, so it knows what’s in the store and how many of each size is left. So, when a consumer is asking for a specific size, he can say, ‘We do have a size nine-and-a-half. I’ll bring that right out to you.’

This is a project we’ve been working on for probably about a year now. We’ve worked really closely with Nvidia and a company in their portfolio called LiveX, … and we tested this in our office for the last three months.”

What did the testing process look like?

“In our office, we have an employee store, so we set [the concierge] up next to that store. We got really great feedback. Some of the best feedback we got is that — and this is going to sound silly — it’s not creepy. A lot of these AI interactions that [people] have still feel a little off, but this one feels organic and natural. Another piece of feedback I got from one of my friends at work was, ‘Hey, I’m a little introverted, so I like having this other option of being able to talk to a digital person.’ That was an interesting piece of feedback that I wasn’t expecting. But it’s important to give consumers different ways to interact with our brand. And some people might prefer going this route, versus a different route.”

I’m curious: Do any of your store employees have any reservations about co-existing with the digital human?

“We did a couple of site visits, … and all the feedback from the store employees has been positive. I think they’re interested, as well. AI is just such a buzzword that we hear in different parts of our lives, over and over. I think they’re excited to have a new technology that they can also interact with and learn from.”

Do you plan to roll out this concierge to other stores?

“We’re still very early. …. We want to see how this first one goes, fine-tune it, continue training it, see what questions it’s getting, where the holes are and fill those holes. Once that experience feels really good and we’re seeing good metrics, then we can potentially talk about what other stores we want to roll out to. Right now, the focus is on Vegas and getting this to be fine-tuned. … There’s definitely going to be ways that consumers use [the concierge] that we’re not going to expect, and we’re going to have to rethink how we utilize this.”

I’d love to talk about other AI initiatives you’re proud of, from the last couple years. For instance, you started letting customers design products via AI. Can you tell me more about that?

“In 2024, we created a process for people to design their own Manchester City jersey, using a text-to-image AI tool that we produced called the Puma AI Creator. We wanted to empower consumers and let them go out and create their own jersey and make it accessible to anybody. … You could type in anything and have it appear in the jersey, but [we] also had guardrails to make sure that we were protecting our brand and Manchester City, as well.

Initially, our testing took two to five minutes [to generate the initial design], but we worked with Google and Nvidia to get that down to 30 seconds. We still wanted to fill that time, so we added a Tinder-style voting system, so you could see everybody else’s kits. … [After the design fully loaded] you could make the design bigger or smaller and change it from a V-neck to a crew-neck to a button-neck. You could change the trim color, the badge color and the sponsor color, and add your favorite player on the back.

So, we produced this, and the results were fantastic. In 10 days, we had 54,000 users, 180,000 kits, participants from 206 countries and territories, and 1.7 million kits rated in that voting system. If you do the numbers, the average person voted over 30 times. It was a contest, so there will be one winning kit, which will be announced here in a few months. That one winning kit will be [Manchester City’s] official third kit for the season. … People really gravitated toward classic and simple-looking kits, so we stayed true to that and made sure we respected the fans’ wishes.

We did this again with [professional football club] Olympique de Marseille, a couple of months ago. We improved the AI engine from the back end. … From the first [AI jersey initiative], we learned that prompting — ‘Type in anything you want in a jersey’ — is really hard. For this one, we made the prompting more of a chat feature, so you can type in your prompt, and the AI will respond to you and give you suggestions to try and make your prompt stronger.

That helped vary the designs. … We also added a gallery feature that has a remix feature. So, you can click on anybody else’s jersey and remix it. … We saw 173,000 kits created from nearly 33,000 users. The average session time was eight minutes per session. [We had participants from] 177 countries and territories, and 900,000 kits rated.

Digital Human and AI Creator are two of the bigger projects that we’ve worked on, but Puma’s done a lot in the AI space. There’s a shoe we did a couple of years ago with our designers, and we used AI to help design, as well. And then, from a creative-services standpoint, we use AI to produce some of the images on our product-display pages on our website, and also to do some of our lower-level creative, as well.”

How do you decide which AI projects to prioritize?

“AI is touching every single part and function of the business, so I don’t think you can really run away from it. Every single part of the organization really needs to think about how AI works [in order] to enhance their workflows and whatever they’re producing. … Historically, I’m somebody who thinks from a consumer lens. I used to work at the NBA, and I did social there, in the early social days. Our mantra back then was, ‘How do we bring the fan courtside, using our social media channels, and give them that sneak peek?’ … I took on that mantra when I started this group here at Puma. ‘How do we use new technologies to interact with our fans in meaningful and engaging ways?’

When AI first became a buzzword, we came up with this AI Creator idea, so that people can design jerseys. It was a fun moment, especially because, at that point — and this still happens today — people were scared of AI. They still push back on it, especially when you see it in creative. … When we were thinking about AI Creator, our idea was, ‘We don’t want to produce something using AI and just give it to fans. We actually want consumers to come in and use this tool to co-create with us and their favorite club and produce something unique.’ In the case of Manchester City, it’s going to be the first-ever AI-created jersey on pitch [the field] in the Premier League, which is a pretty big deal, in my opinion. So, we really gave the keys to the consumer to run with the AI experience.”

What other emerging-tech projects do you have in the works for this year? Are you still building games in Roblox?

“Yes. Talking about the Roblox side of it, we have a very robust digital-goods business. We’re selling Puma products across different video games. So, we’re working with games like ‘Fortnite,’ like ‘NBA 2K,’ but also games you’ve probably never heard of, like ‘Yeeps,’ which is a VR game that Gen Alpha loves. We just launched there, … so we’re in multiple VR games. That’s been a really interesting place for us to be. We’re providing other touchpoints for our consumer to interact with the Puma brand.”