The Beauty of AI: Estée Lauder's Sowmya Gottipati
23:31 Share

Today we're airing an episode produced by our friends at the Modern CTO Podcast, who were kind enough to have me on recently as a guest. We talked about the rise of generative AI, what it means to be successful with technology, and some considerations for leaders to think about as they shepherd technology implementation efforts. Find the Modern CTO Podcast on Apple Podcast, Spotify, or wherever you get your podcast. Skincare products are inherently physical, not virtual.

How can companies use AI to help make choosing skincare products possible online? Find out on today's episode. I'm Soumya Goodparty from Estee Lauder, and you're listening to Me, Myself, and AI. Welcome to Me, Myself, and AI, a podcast on artificial intelligence and business. Each episode, we introduce you to someone innovating with AI. I'm Sam Ransbotham, professor of analytics at Boston College.

I'm also the AI and Business Strategy Guest Editor at MIT Sloan Management Review.

And I'm Sherwin Kodobande, senior partner with BCG, and I co-lead BCG's AI practice in North America. Together, MIT SMR and BCG have been researching and publishing on AI for six years, interviewing hundreds of practitioners and surveying thousands of companies on what it takes to build and to deploy and scale AI capabilities and really transform the way organizations operate.

Sherva and I are excited today to be talking with Somya Kotapati, head of global supply chain technology for Estee Lauder. Somya, thanks for taking the time to talk with us. Welcome. Glad to be here. I'm really excited to be here talking about AI, one of my favorite topics. Yeah, great to have you. Let's start with your current role at Estee Lauder. What do you do now? I've been with Estee for about two and a half years. I've been with Estee for about two and a half years.

We are a prestige and luxury beauty brand company, and we have around 30 brands under our umbrella. Estee Lauder being the flagship brand, but we also have Clinique, MAC, La Mer, several others. We are a global company with our products across all various regions from Asia to Northeast and Latin America, Europe, and everywhere.

I'm responsible for global supply chain technology. So I am responsible for the technology that powers the entire supply chain globally. That includes inventory, supply, demand planning, manufacturing, distribution centers, fulfillment, transportation, end-to-end supply chain, all the technology capabilities that support that.

All right. I didn't hear anything about artificial intelligence in that, though. So how does artificial intelligence connect with that supply chain? I took the supply chain role just about four months ago. So prior to that, I was responsible for brand technology for the Estee Lauder brand. That was really where I was directly involved in a lot of AI applications.

How we use AI at Estee Lauder, starting with the consumer experience, how we're enhancing our consumer experience with the AI technology and providing real-life applications, which I can talk to a little bit with the virtual try-on and so on and so forth. And then personalization is a very important area and the application of AI is very significant.

And then AI is also being used to create new products such as skincare and fragrance. This is where you can use data to inform ourselves on what kind of ingredients and what type of products that people like. And so we can inform our product planning using AI.

and what I call agile enterprise. So there are a number of areas where AI is applied to run an efficient organization such as supply chain and R&D. These are all various areas that we're using AI in.

Sounds it's quite prevalent across the whole value chain. I did read up on something, I think a talk you had given on how AI is being used to help personalize the fragrance. Is that right? That's right. Yep. Can you comment on that a bit? Absolutely. That's definitely one of my most exciting projects that I worked on. It's an industry breakthrough, so I feel very proud about it.

It's a fragrance recommendation engine, but it takes advantage of neuroscience and AI and olfactory science. We are bringing all three sciences together to make that happen. You know, human brain has approximately 400 olfactory receptors.

And we are working with a company that can actually replicate those receptors in a lab environment. So if you take a particular fragrance, we can actually tell which olfactory receptors in your brain are activated by that fragrance. So are these like neuromorphic chips or are these silicon-based software?

It's not a software. These are actually biosensor testing. That's pretty cool. Yeah, it's really cool. So we would be able to tell receptor 67 and 92 and 86 are triggered by this particular fragrance.

And let's say that fragrance is predominantly lavender based. By the way, your brain can't really tell the difference between lavender and woody. So I might be able to bring woody fragrance to you and same receptors might get triggered because they are evoking the same emotion in your brain. So because same receptors are being triggered, we can tell, oh, by the way, just because you love lavender, you might like this other fragrance that may be woody, which smell totally different.

But they have the same effect on your brain or they trigger same emotional reaction in your brain. That's very cool. And that's live?

That's live. Yeah, we are piloting that in China right now. And we're trying to expand it to other areas as well. The way we implemented it, it's interesting because we started off with online because selling fragrance online is very difficult because you can't hurry a smell, right? At least we haven't, we don't have that technology yet, maybe 10 years from now. But this is why we came up with this technology to see maybe can we use facial recognition and

And the patient recognition can identify the emotion that you're feeling based on very subtle changes in your face when you smell it. And based on that, we can recognize how you're reacting to each of those fragrances. You get a score. And based on that, we could tell whether you liked or didn't like it or moderately on a scale of 1 to 10, how you're liking it. And we use that data.

So as a customer, you're looking at my facial recognition and then deciding what's the right fragrance for me. Correct. One of the things that Sam and I have been probing into over the last several years is the collaboration between human and AI and how...

It is so much more accretive, that collaboration to the pure tech or the pure human and how they complement each other. And it seems to me that fragrance and makeup and these things are so personal. And I have to imagine that in the AI solutions you talked about,

There must be or should be a fair amount of human intervention or collaboration. Can you comment on any of that? I get the recommender systems and how it works and the receptors and all that. But is there a human side to this as well that maybe the experts and maybe the customers are interacting with the recommendations of AI and adapting it?

Absolutely. Historically, when you try lipstick, how many can you try? Maybe three, four, five. You can't do more than that because after a while, the skin starts drying up and it's uncomfortable. But now with virtual try-on capability, you could try 30 shades of lipstick in 30 seconds. Same thing with foundation. We have...

56 shades of foundation, which are like so slightly different. We take pride in providing high touch service and in each of our stores, we have beauty advisors. Their job is to work with the customer and recommend products.

various foundation or lipstick, etc. How do you try so many different foundations? You can't, whereas AI can narrow it down for you. This virtual trial applications can narrow it down to two or three. And from there, beauty advisor can actually work with the customer. So beauty advisor is there to help them choose what actually looks better and have that conversation and also explain

Why recommend this for your skin based on the results that you want to achieve, whether it be acne or dryness, etc. So we don't see that going away, that human-machine interaction. It will always be there. Is there a feedback loop whereby the machine gets smarter? For example, the beauty advisor says this or the customer, now you've narrowed down from 60 shades to 3 shades.

But based on the final choice they make, I assume the algorithms are getting smarter from that interaction as well. Yes, absolutely. There are two ways it happens. One is...

We have a consumer data platform that has information about what you previously bought, what you liked, what your situation is, et cetera. So it feeds into that so that next time when you come into the store or when you interact with us, we can say, hey, by the way, last time you bought this, so I can reserve that for you or I would recommend something else. And the second thing is when we rolled out virtual trial applications, we started off with

million faces for the data modeling. Now it has 100 million faces. So that algorithm and the engine is constantly improved over a period of time. And these are faces of actual customers, right? Actual customers, that's right. How does that beauty advisor work with the platform to get that feedback back into the system? I guess you can see what they actually ordered or what they chose or what they preferred.

How do they get that input back in? I could tell, Sam, you're intrigued by the beauty advisor concept. Yeah.

We have very strict privacy laws. So in the store, when people are buying in the store, a lot of times we actually do not gather their personal information. But whereas when they're buying stuff from us online or social platforms where there is a login and that kind of mechanism, then you have that information. So we know exactly what they bought and that information gets passed on.

One of the things we're seeing, maybe it's a teaser of our new work to come out, but one of the things we're seeing is that the ability to understand and explain why an algorithm or an AI solution makes a recommendation or pulls out a particular insight or an action, just the ability to sort of understand it rather than it's a black box, helps organizations get a lot more adoption

I can speak in our supply chain world. In the last year, we rolled out an application to do our supply planning and demand planning. Before, it was spreadsheets and those kinds of things. The moment we started using the AI application, we saw accuracy of 30% increase in our forecasting accuracy.

Exactly. Yeah. Some of my clients deliberately will settle for a less precise or less accurate recommendation so that they get the adoption going. Maybe they go for less precision to trade in a little bit of explainability or ability to override. And so that way, at least people will begin to trust it more. I don't know whether you do something like that. Yeah. I have not come across that, but that's a very interesting point.

There's an angle there, too, that trades short-term and long-term, Shervin. Like, let's say short-term, they take a compromise solution that isn't quite as good. And then they can come back in three months and say, hey, you overrode this and it didn't turn out as good as you thought, sunshine. It's a longer game. It's not just each one-off decision that short-term optimal.

I did just that with my son, actually. He was going to a school dance and he was outside and all he had was a t-shirt on. And I said, wear a jacket, wear something. He says, no, I'll be fine. And I'm like, okay, you're going to get sick. And he got sick. But hopefully he learned. And he's like, dad, you were right. And I'm going to make him listen to this podcast so that he knows now I've said to the whole world.

Well, Shervin, if you can drag your kids into this, I have to tell the anecdote that I, this will shock you, Shervin, but I track the time that my kid's bus arrives every day. So I've got seven years worth of data now of what time that bus arrives. And so my next step is I'm predicting that. I'm trying to say, okay, what do we think today? Is today going to be an early day? Is today going to be a late day? And then we can leave the house at the right time, but maybe we miss it someday. So I'm not sure. How's that working? Yeah.

How's that working out for you? I'll have to come back on a later episode and see how that plays out. But at least it's real time. You know, I'm trying to use the dog food of the things that we talk about on the show. Speaking of things we talk about on the show, how's that for a segue? So we have a segment where we ask our guests a series of rapid fire questions. And so the idea is you just hear this question and you give the first response that comes to your mind.

And Shervin, are you doing these today or am I doing them today? No, I'm not because I don't have it in front of me. Okay. All right. So, Sumya, what's been your proudest AI moment?

I think I already spoke about this, the fragrance application we built last year. In my current role, that is really my proudest moment. But before that, when I was in my previous job, when we cracked the code on the computer vision, combining computer vision and with natural language processing to break down video processing, because that's really the beginning days of AI, but we were able to build something like that. So that was really cool.

This is one thing I think it's the coolest thing about technology. Technology transcends industries. It almost doesn't matter what industry it is. Technology is so pervasive. So I feel so happy that we are able to apply the same technology for totally different applications. And that's the beauty of it. If you'd come up with something that topped the fragrance example, I was going to be super impressed. That was already a pretty proud one. So what worries you the most about AI?

What worries me the most? I think it's the data privacy and the bias, those definitely, and the tracking. One hand, when I use Google Maps, I like the functionality. I like what it does. But at the same time, I know Google knows exactly where I am at every second of the day. I don't like that, right? So the privacy and the data tracking piece, absolutely a problem.

What's your favorite activity that involves new technology? Reading a book. Do you have any recommendations for us? You can extend your answer with a book recommendation. Recently, it's a very short book that I read, a diary. It's about a Pakistani girl during partition between India and Pakistan. And this little girl who

Didn't talk, but wrote a diary every day. It was a really moving and interesting book. I really liked it. So what was the first career that you wanted when you were a child?

Oh, I wanted to be a pilot, actually. You are one, right? Yes, I am. Okay, so that would get a checkmark next to that one. You got that one. Well, I'm more of a recreational pilot, but I actually wanted to be a professional pilot, but that's okay. I'll settle for recreational. I don't know. You've gone from AT&T to NBC to Estee Lauder. There's a next step.

So what's your greatest wish for AI in the future? What are you hoping for?

This is more of an answer from my personal side of things, which is I just hope we use AI for environmental causes more, you know, better crops with better yields and water conservation. And I hope there is a lot more advances on that side of things as opposed to shopping or, you know, personalized experiences.

That's particularly interesting coming from someone who's so interested in both of those aspects that you think that these other aspects might be even more promising. Some of you, great talking with you. And these are fascinating applications. I think...

I think that most people who listen to this are going to remember the smell example. I mean, I think that's just so there's something very visceral about that. I think we'll connect with lots of people and spur some thinking. So thanks for taking the time to talk with us. We really enjoyed it. Thank you. Oh, thank you. This is so much fun. Thank you so much.

And we can have some beauty advisors to work on Sam while we're talking. This is a podcast and nobody knows that I've got a face for radio. Tell them to bring all the shades and foundations and we'll see what they could do. You know, you should go to Estee Lauder.com and try the 30 shades of lipstick in 30 seconds. I'll go with you, Sam. And the foundation. See how it looks on you. I'll go with you. I'll just Google something shades of gray. No, don't go there. Not that. Don't go there.

We've come to the end of season four of Me, Myself, and AI. We'll be back on August 2nd with new episodes. In the meantime, we hope you'll listen to our back episodes and join our LinkedIn community, AI for Leaders, to keep the discussion going. Thanks for listening.

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 listeners 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 your insights,

and gain access to valuable resources about AI implementation from MIT SMR and BCG, you can access it by visiting mitsmr.com forward slash AI for Leaders. We'll put that link in the show notes, and we hope to see you there.