Artificial Intelligence and Business Strategy
In collaboration withBCG
Katia Walsh began her career as a journalist in her native Bulgaria and is now the global chief strategy and AI officer at retailer Levi Strauss & Co. Over the course of her career, she has developed a passion for three things: the power of information, the power of technology, and the power of machine learning. Her enthusiasm for these subjects is evident as she describes how she is ensuring that a well-known legacy clothing brand remains relevant through technological transformation.
In this episode of the Me, Myself, and AI podcast, Katia explains how she has organized digital transformation and employee engagement at Levi Strauss around five C’s: connections with consumers, commerce, creation, careers, and culture. She also describes the machine learning boot camps the retailer has offered to nontech employees to boost innovation and outlines how the company thinks about responsible AI practices.
Read more about our show and follow along with the series at https://sloanreview.mit.edu/aipodcast.
Give your feedback in this two-question survey.
Sam Ransbotham: Many retailers are personalizing their product offerings, but few do it using AI products inspired by employees with limited technical background. Join us when we talk with Katia Walsh, chief global strategy and AI officer at Levi Strauss & Co., about how the company’s AI boot camps upskill its workforce and inspire innovation.
Welcome to Me, Myself, and AI, a podcast on artificial intelligence in business. Each episode, we introduce you to someone innovating with AI. I’m Sam Ransbotham, professor of information systems at Boston College. I’m also the guest editor for the AI and Business Strategy Big Idea program at MIT Sloan Management Review.
Shervin Khodabandeh: And I’m Shervin Khodabandeh, senior partner with BCG, and I colead BCG’s AI practice in North America. Together, MIT SMR and BCG have been researching AI for five years, interviewing hundreds of practitioners and surveying thousands of companies on what it takes to build and to deploy and scale AI capabilities across the organization and really transform the way organizations operate.
Sam Ransbotham: Today, Shervin and I are excited to be talking with Katia Walsh, chief global strategy and AI officer at Levi Strauss & Co. Katia, thanks for taking the time to talk with us. Welcome.
Shervin Khodabandeh: Thank you for joining us, Katia.
Katia Walsh: My pleasure. Thanks for having me.
Sam Ransbotham: Can you tell us about your current role? What are you doing for Levi Strauss now?
Katia Walsh: I’m responsible for a fusion of strategy and artificial intelligence, and to tell you a little bit more about that, it’s really building an integrated capability that connects emerging technologies, data, and AI in one holistic capability in service of our strategic goals.
Sam Ransbotham: How long have you been doing that?
Katia Walsh: I’ve been the chief strategy and AI officer for Levi Strauss & Co. for the last 2½ years. I joined very, very soon before the pandemic began, so it has been an induction by fire, and it has been also a great opportunity to show the power of technology in times like this.
Sam Ransbotham: Yes, it’s a great time and a challenging time, I’m sure. You didn’t start off, though, in this role. Tell us what was happening before 2½ years ago. I believe you actually started as a journalist in Bulgaria. Connect the dots between that and Levi Strauss.
Katia Walsh: I did grow up in communist Bulgaria at a time when Levi’s signified so much more than fashion or clothing; it was really the flag of freedom. It was about independence and democracy and the unattainable. I did start as a journalist for one of the very few independent publications in the country, and very early in my life, I learned the value and the power of information, and information is really data. Newspaper stories are data; there is a field now called data journalism. I didn’t know that at the time, but I learned about the power of information, and I developed a real passion for it.
And then I had the opportunity to come to the United States on a full scholarship and continue my education, and that was in the heyday of the internet, so that’s when I developed my second passion — for the power of technology to amplify the power of information or data. Then I continued my education. I got into academia and developed my third passion — for machine learning and statistics. I’m not an engineer, I’m a statistician, but through my education in statistics and machine learning, I developed this third passion about the power of machine learning to help us to drive desired outcomes.
Sam Ransbotham: What are some of those outcomes you’re trying to drive right now at Levi Strauss?
Katia Walsh: This is an industry, whether it’s apparel or fashion or retail in general, outside of the likes of Amazon. This industry has been quite analog, manual, imprecise, and traditionally not the best citizen of the planet. So my mission at Levi’s, together with the teams with which I partner and create and grow, is to help the company become … what used to be analog is now digital, what used to be manual is now automated, and what used to be intuitive is now precise. We are driving the full digital transformation of the enterprise, but also a disruption of the entire industry. This is the biggest thing to happen to this whole industry since the first industrial revolution.
Shervin Khodabandeh: That’s really quite fascinating, Katia. Share a little bit more about some examples of that revolution that’s happening in the industry, either at Levi’s — whatever you’re free to share — or anything outside so that our audience can have a better sense.
Katia Walsh: Yes, Shervin. I’m happy to share what we are doing at Levi Strauss & Co. There are three C’s that we center on when we deploy these great capabilities around digitization, data, and AI. And the first C is always about connections with our consumers. Levi’s as a company, in its 169 years, for most of that time had been a manufacturer. It had not had a direct connection with its consumers. But we recognize the importance of deepening the connections we have, not just with consumers but with our fans. What is great about Levi’s — an iconic brand like it — is that our consumers are not just consumers; they are ardent fans that literally tattoo the brand on themselves. And so we want to deepen that connection, and we want to use everything technology has to offer to us to do so. And of course you know what I’m talking about. It starts with some things like personalization, [which] everyone else is doing, but we would like to think we are doing even more.
One example of that is, we are completely personalizing the online experiences. So when you go — whether it’s on the app or the e-commerce site — what you see should be somewhat customized to your previous browsing behaviors, to your needs and desires, to everything we know, always shared with permission. So deepening the connections with our consumers is a big part of what we do.
Another C that we center on is, we want to make sure that we use technology, digitization, data, and AI to create smarter commerce. And this is where anything around internal efficiencies can be very helpful to the company. It may not immediately or directly touch the consumer, but it certainly has an impact on the consumer.
One example of what we’ve been able to do during and after the recent pandemic crisis was pricing optimization. When we were facing the depth of the lockdowns, as a global company, we certainly felt the impact of the pandemic all over the world. We had, at one point, two-thirds of our stores that had to be closed. On the one hand, being global gave us diversification, and where we had to be closed, in other parts of the world we were able to be open, and that gave us some learnings. On the other hand, we did feel the impact all over the world, and we used AI to determine the optimal price at which our products would sell anywhere in the world, through what channel, at what price, to which consumer. And that was very helpful, because we did not have to actually discount. A lot of our competitors did not have this incredibly powerful tool, and they had to discount as they were facing piles of inventory. But because of the strengths of the Levi’s brand, and because of the application of machine learning, we were able to predict that our products would sell at full price, so that helped the consumers to get what they needed, but it also helped the financial margins of the company.
And then the third C that we are also making smarter, where we apply artificial intelligence, is creation, the very nature of what this company does. And one example is that we have recently started to use AI in the design process. We now use convolutional neural networks, for example, to create new designs that process thousands and thousands of images — for example, van Gogh’s Starry Night, or David Hockney’s artwork or Jasper Johns’s artwork. And we can now create trucker jackets, which is a legendary product that Levi’s literally invented, but on that there is now van Gogh artwork, which we will be producing and selling to the world.
Sam Ransbotham: I think I need some Dalí pants.
Shervin Khodabandeh: [Laughs.] It’s quite fascinating, particularly the design example. It tees up the next question that I have in my mind: the role of the human here, particularly since you mentioned design, and I have to imagine that in the past this has always been a very human-centered process. So how is Levi’s bringing humans and AI together to achieve outcomes that neither one could do on its own?
Katia Walsh: I would venture to say that humans are actually the most important part of artificial intelligence, whether it’s human-centric design, which we of course aspire to, or it’s humans that are making machines smarter, and of course in turn, machines help us become even better. In the case of the AI-powered design that I mentioned, what is even more fascinating is that this work at Levi’s was pioneered by one of our young designers who had no formal training in machine learning or computer science. He’s one of the 101 graduates of our industry-first machine learning boot camp that we pioneered in 2021.
For that boot camp, we took a number of people across the entire company everywhere in the world, from 24 locations, from every single function, including retail stores and design. We fully democratized this process of teaching machine learning so that we could get the change agents we needed for digital transformation across the company and also help ourselves in this ongoing war for talent in AI. So the creator — going back to the AI design part — the creator of this was actually a graduate of this boot camp and is absolutely central to the design process.
Shervin Khodabandeh: Can you comment more about this university you were talking about to really educate and upscale and reskill?
Sam Ransbotham: I think you said 101 graduates so far; is that right?
Katia Walsh: That’s right. We’ve had two classes — we call them cohorts — from our 2021 application process. We had about 450 applications. This is not a program for everyone, just to be clear, because it takes people out of their day job for eight weeks. And I have my colleagues to thank for making their people available for eight weeks. So we’ll take people for eight weeks out of their day job. It was incredibly immersive and intensive and hands-on. We called it a boot camp for a reason. They literally had no time to do anything else, and they were exhausted when the time for graduation came.
They worked with real data to solve Levi’s problems, and we were actually able to deploy the models they created after the boot camp — models that were looking at prediction of demand, as I mentioned earlier; that’s incorporated into that work. The AI-powered design will also be taken further, but also other things like personalization of our marketing messages; that is something that the boot camp graduates worked on. So we created this for Levi’s people with Levi’s data to solve Levi’s problems, and we are now in the process of selecting the next graduates, who will start in April and graduate in May, and then we’ll have another cohort — another class — in the fall of 2022 as well. So this is an ongoing effort, and I’m very proud of that.
Shervin Khodabandeh: This is really great; it’s fascinating. So it seems like in a few years, you’ll have several hundred folks that I’m assuming are going to be embedded in different lines of business, right? These are not your technical folks that are maybe in the technology or AI or data science or engineering groups. What a great way to upscale and immerse folks in the business about the power of AI. Is the ambition to continue this until almost everyone’s gone through this, or what’s the ambition?
Katia Walsh: Well, Shervin, first of all, you are right that most of the people who graduate from the boot camp go back to their roles. They’re not necessarily changing their job. There are people who want to become very advanced data scientists and, of course, we don’t want to deprive them of that opportunity, and we do give them that opportunity when the time comes. But the vast majority stay in their existing roles and thus upgrade their own roles. And in that context, I want to mention two other C’s that we are also targeting through this combination of digitization, data, and AI capability we are building. I did mention earlier the connections with consumers, the commerce that we are making even smarter, and, of course, the creation process.
We have now two other smarter C’s. One is careers. People who go through this boot camp do change their outlook and the abilities to have a career, whether it’s at Levi’s or outside of Levi’s, although I’m proud to also say that in this time of the Great Resignation, the very vast majority of the graduates have stayed with Levi’s, and I do credit the boot camp, at least to a certain extent, for that. And then the other smart C is culture. These people are now helping us change the culture in the whole enterprise globally. They think differently; they know the language they speak; they connect with data scientists, engineers, and product managers. And so, collectively, through all of that, we are transforming the company for its next 169 years.
Sam Ransbotham: Katia, can you give us another example of someone who’s graduated from the boot camp that’s gone on to do a good project?
Katia Walsh: We have lots of examples, actually. We have 101 examples, at this point. But one other example that I think is also particularly compelling is about a graduate who is a retail store manager, someone who had never seen code in her life before. Her job had been as a stylist for 11 years, so she’s very close to the consumers. She talks to consumers all the time; she helps them make exciting decisions, she recommends what they should pair with what. But now, she has acquired skills that have enabled us to create a model that bundles items in our vast array of inventory that work very well with each other, that create outfits.
And so, through this automated process, this retail store manager in our Denver premium outlet store is able to proactively go ahead and suggest something that’s not only her own personal idea but is based on the recommendations of a machine learning model. She’s able to recommend bundles of items, entire outfits, to our consumers. And, of course, because it’s a model, it’s always learning, she’s always getting new data, and it’s always getting even better.
Shervin Khodabandeh: What a great story. Send that woman to Sam for some fashion.
Katia Walsh: I’ll give you one other example of a machine learning boot camp graduate and what she did, a woman who works in our Las Vegas distribution center. So she went back to her old job, where she had been facing an ongoing problem for years, and the problem was that every day, the distribution center experiences downtime — something goes wrong. Equipment breaks, a part wears out, and the distribution center would have to go anywhere from 15 minutes to two hours with no work — just complete downtime while we would continue to have to pay, of course, the cost of labor, and we would be missing out on the profits of shipping orders out.
So this young woman went back to her job with the skills she had learned and she said, “Well, I can now tackle this problem.” So she created a predictive maintenance model that now predicts with a great deal of accuracy what equipment is going to malfunction in the next 30 days. Moreover, she designed an app that shows those predictions in an easy-to-view way and automatically dispatches technicians to go ahead and preventatively check the equipment, so now there’s no downtime in this particular distribution center.
Sam Ransbotham: What are some other things that you think that are involved with responsible use of AI and ethical and trusted use of AI within Levi Strauss? I know this is something you’ve been thinking about and working on.
Katia Walsh: [I’ve been] very much thinking about and working on AI for good. It’s a very powerful tool, as we know, but like any tool, it can be used for good and for not-so-good. And one of the reasons I’m so happy to be at a company like Levi’s is that it really does bring a lot of values that are transcending an industry or geography or even an era. So within the context of data and AI, we have set a [code] of ethics for everyone who works with data in the company. Over time, that will be everyone in the entire enterprise. Everyone who works with data in the company at this time has to actually sign a code of ethics, not unlike the Hippocratic oath, to make sure that we always protect our consumers, our company, our shareholders, because how can you delight consumers if you’re not protecting them?
And then there are other things that we are doing to ensure that we use data and machine learning with utmost care. For example, as you well know, there is still a lot of opportunity, unfortunately, for bias in models and algorithms and the outcomes from that, so while we know that you cannot ever completely eliminate bias in life, we are doing our best to minimize it, and there are three ways in which we do so. One: through the people. The more diverse people we have who work with data and bring data and create algorithms, the more likely we’ll have these implicit checks and balances to ensure that we minimize bias.
The other reason we bring such diverse data is not only because it can enrich our models but because it can also help minimize bias. So that’s why the data sets are so different, and sometimes we bring together data sets that have never met in the past, and it’s amazing what you can find out when that happens, but that also helps minimize bias.
And the third way in which we work to minimize bias is the diversity of tools. We purposefully deploy a great deal of open-source tools. We make sure that, yes, while it’s beneficial to work with certain vendors, we also always want to stay on top of what’s next. And what’s great about open-source tools is that they’re worked on literally all over the world by anyone who has the skills. That is one of the reasons we also deploy open-source tools: to ensure diversity.
Shervin Khodabandeh: I want to go back to one of the comments you made a bit earlier about modernizing, digitizing, infusing with deeper, better data and analytics those functions in the company, particularly in a sector like fashion retail, that maybe generally have traditionally been quite analog. Like, you talked about merchandising and planning, and I’ll add to it maybe forecasting and pricing and things of that nature. Can you share some stories about how you’ve been able to bring folks who grew up sort of in a different era or are much more used to the old way of doing things, and how do you bring the new and old together in a collaborative way?
Katia Walsh: I think what you’re talking about is, in general, how one can shepherd a change in a company or in an organization in general, and this has been my entire career. I spent 20-plus years at the intersection of technology and data and analytics and machine learning, and most of that career has been actually helping companies transform themselves to meet their strategic goals. It’s particularly challenging with technology, and especially when you look at a particular technology like AI, because it can be seen as so intimidating.
And so one of my aspirations and missions, actually, has been to humanize it — to make it closer to people, to give it a face, to help people understand that not only is it not there to replace jobs, for example, but it is there to help them succeed even more, to make them even smarter. And that’s one of the reasons I introduced this machine learning boot camp, not only at Levi Strauss & Co., but at other companies as well. So humanizing this capability is very important, and how exactly we do that may depend on the specific context of a company or an organization, and certainly even within a time that can change. But I really believe in exciting and inspiring people about it and helping them get on board because they want to, not because they feel threatened.
Shervin Khodabandeh: Are there lessons there for others who are playing similar roles as an agent of change in terms of how they might go about implementing changes of this sort?
Katia Walsh: Yes, one thing that I have found very helpful — this is now the third company in which I am transforming the enterprise … not alone, and that’s actually one of the lessons: You cannot do it alone. You have to make sure that you garner that support, and that support has to be throughout the organization — certainly at the top leadership level, but also throughout, grassroots and sideways, as well. So getting that support is really critical. But in addition to that, I’ve had this motto of “think big, start small, and scale fast.” I suppose the “start small” can be seen as a compromise, but I don’t see it as a compromise. I see it as an opportunity to deliver immediate value. And that’s a big part of getting people on board.
When we are able to show value very quickly, even if it’s not the biggest value in the world, it has to be meaningful to excite people. But if people are able to see it very quickly and very concretely in their own business unit, function, geography, that certainly gets people on board because they see, “Wow, this is helping me solve a problem that I’ve been looking to tackle all this time,” or, “This is helping me meet my commercial goal that otherwise I might have struggled to achieve.” That helps a great deal. And what it also does is, if they’re people sitting on the sidelines, it makes them want to say, “Well, how come I don’t have this? I want to get this as well.” So it creates a little bit of healthy competition. So to summarize, some of what I found helpful has been certainly ensuring support throughout the company, and one way in which it has helped me to do that is to think big, start small, scale fast, show that immediate value, and get people on board.
Shervin Khodabandeh: And educate and humanize, as you talked about earlier.
Katia Walsh: That also.
Shervin Khodabandeh: It’s not a simple answer. It’s a real transformation.
Katia Walsh: Yes. The other thing is that when we build these capabilities … I want to actually address a potential fallacy. A lot of companies [embark] on a digital transformation, as we are calling it now. We’ve been digital for the past 25 years as a world, but I guess COVID has accelerated this need to modernize businesses in every industry. But a lot of companies embark on that without having a clear vision or view of why they’re doing that. And that’s why the fusion with strategy is so important: because it gives that “why.”
But the other thing to keep in mind, too, is, it’s not about building technology. Yes, technology matters a great deal. It’s a key enabler. But we have to be careful about “build it and they’ll come.” We have not built the perfect data ocean. We have it; we have gotten it started. It will never be perfect, because data is never perfect. It’s always coming at us like a tsunami. I wouldn’t say that we have built the perfect platforms yet, either — again, because they’re developing so fast in such a dynamic field. But what we are doing is to show that value very consistently and hopefully growing it over time. And that’s what the essence of transformation is.
Sam Ransbotham: Katia, it’s great talking with you. Thanks for taking the time to talk with us. We’ve really enjoyed it.
Shervin Khodabandeh: It’s really been great. Thank you, Katia.
Katia Walsh: My pleasure, Sam and Shervin. Nice to meet you.
Sam Ransbotham: Thanks for listening. Next time, we chat with Kobi Abayomi, senior vice president of data science at Warner Music Group. Don’t worry — neither Shervin nor I will sing.
Allison Ryder: Thanks for listening to Me, Myself, and AI. We believe, like you, that the conversation about AI implementation doesn’t start and stop with this podcast. That’s why we’ve created a group on LinkedIn, specifically for leaders like you. It’s called AI for Leaders, and if you join us, you can chat with show creators and hosts, ask your own questions, share insights, and gain access to valuable resources about AI implementation from MIT SMR and BCG. You can access it by visiting mitsmr.com/AIforLeaders. We’ll put that link in the show notes, and we hope to see you there.