How This Gen Z Founder Raised $11 Million To Build The AI-Powered Closet From 'Clueless'

Forbes

Nothing to wear? Check out Alta, Jenny Wang’s AI-powered stylist and shopping app. Upload photos of items you own (or recent shopping receipts) to create a digital closet. The New York–based startup will then list outfit suggestions or—and this is where Wang hopes to make money—recommend new stuff to buy based on the weather, season and occasion. “Fashion is so interesting from a data perspective,” says Wang, a Harvard-trained engineer. “Alta is both utilitarian in that it helps you decide what to wear and buy, but it’s also just aspirational.” Amid the sea of AI-powered closet startups, Alta is backed by both venture and fashion heavyweights, including Menlo Ventures, which led the company’s $11 million seed round in June, LVMH-backed Algaé Ventures and Michelle Obama’s stylist, Meredith Koop.  Read the full story on Forbes:   Subscribe to FORBES: https://www.youtube.com/user/Forbes?sub_confirmation=1  Fuel your success with Forbes. Gain unlimited access to premium journalism, including breaking news, groundbreaking in-depth reported stories, daily digests and more. Plus, members get a front-row seat at members-only events with leading thinkers and doers, access to premium video that can help you get ahead, an ad-light experience, early access to select products including NFT drops and more:  https://account.forbes.com/membership/?utm_source=youtube&utm_medium=display&utm_campaign=growth_non-sub_paid_subscribe_ytdescript  Stay Connected Forbes newsletters: https://newsletters.editorial.forbes.com Forbes on Facebook: http://fb.com/forbes Forbes Video on Twitter: http://www.twitter.com/forbes Forbes Video on Instagram: http://instagram.com/forbes More From Forbes:  http://forbes.com  Forbes covers the intersection of entrepreneurship, wealth, technology, business and lifestyle with a focus on people and success.

Transcript

00:00Someone has to build something in the space, leveraging these transformer models and diffusion

00:05models. We didn't see anyone else do it, so we started the company.

00:17Jenni, thank you so much for joining me today. You are building the modern day Clueless Closet,

00:23so I'm so excited to talk to you about this kind of convergence of fashion and tech and how you are

00:28finding yourself in this industry right now, so thank you so much for joining me. Thank you so much

00:33for having me. I'm so excited to be here. So for those who have not seen the Clueless movie or who

00:38have not heard of Ulta, can you tell me a little bit about the company that you're building today?

00:41Yes, so Ulta is an AI stylist and what that means is every morning I look at my Ulta app and Ulta tells

00:48me what to wear based on my closet, my style, the calendar and the weather. And you can also ask Ulta

00:54to sell you for any occasion, such as I'm going on a trip to Columbus, Ohio, which should I pack or

00:59wear? And Ulta will also generate the outfits on an avatar that looks like you. How did you come up

01:05with this idea? Because I feel like this is something that like so many of us can use in our lives, but

01:10what kind of took this from something that you just thought one day like, oh, I wish I had something

01:14to just tell me what to wear to then building a company around it? As you said, it really came from

01:20a personal problem. I used to struggle every morning trying to decide what to wear. I think

01:24we all have that like pile of clothes on a chair trying to figure out what exactly the outfit for

01:29the day is. And at one point in time, I even had a spreadsheet. I was trying to track what I was

01:34going to wear. And this is actually a very complicated permutation. When you think about it,

01:38you're trying not to repeat outfits. You want to think about, okay, well, what did I wear the last

01:42time that I met someone? What is the weather today? Am I commuting on the subway? Am I walking,

01:47et cetera, et cetera. And so I just felt like, wow, like there should just be an AI to help me

01:53decide what to wear. And I think Clueless was such a seminal moment. I think this year is actually the

01:5930th anniversary of the movie. So it's like crazy that it's been so long and the technology is finally

02:06good enough to build a product like this where Ulta is a AI style companion. So it is both a utilitarian

02:14AI product, but it's also just fun. Like I love getting my outfits every day and seeing it on my

02:19avatar. Yeah. I love that. And it is so crazy to see how much has changed over the last couple of

02:24years, of course, from 1995, but even within the last like five years, you officially launched Ulta in

02:292025. But when did you first start toying around with this idea? Yeah. So we launched on the app store

02:35earlier this year, but I actually tried building a version of this a few years ago using the state of the

02:41art and AI then, which is neural networks. And it was just too difficult to do like that type of model

02:47can often be a black box. And you have so many, so many parameters when it think when you think

02:53about fashion, whenever I talk to my peers in AI, and some of them are working on AI for legal or AI

02:58for biology, etc. They're always amazed at how difficult fashion is as a problem. And so to break

03:06that down a little bit, like I love your pants. And they have all these little rhinestone studs on

03:11them, they have a certain wash, the denim, there's a there's a specific straight leg cut, like all these

03:17different nuances are three dimensional data points that you need to feed into a computer vision model,

03:25or just any sort of AI model that understands fashion. And I think in the prior era of AI models,

03:31like even when you think about what was AI 10 years ago, you were having models that were like,

03:35is this a hot dog? Or is this not a hot dog? You fast forward to what you can do today and what

03:40we're seeing the industry evolve to. And so when I first saw like the transformer models get used

03:46more broadly, and I think, of course, like chat GPT was when it first came out was one of the first

03:51use cases of this, I thought someone has to build something in the space leveraging these transformer

03:57models and diffusion models. And we didn't see any else do it. So we started the company.

04:02Yeah, I want to get into that transformer side of things in a second. First, I want to ask for

04:06someone who's coming to Alta and maybe like does not know anything about AI or how you guys are, you

04:11know, teaching these models how to think and how to work. Can you tell me one, like what a customer

04:16will experience when they click on to the Alta app, and then how that's actually made possible with the

04:22technology that you're creating? Yes, absolutely. So from a consumer perspective, how you use Alta is,

04:29you will add in your closet. And this is pretty easy, you can add in photos from your camera roll,

04:34they don't have to be necessarily photos of clothes alone, it can be photos from your Instagram,

04:38for example, or you can or you can go and take photos of your clothes. And then the AI will recognize

04:44the clothing and swap it out with the original product image if it can find it. So your catalog in

04:49your closet looks really clean. You can also forward your email receipts to Alta. And the AI will take around

04:54five minutes to parse the receipt and put into your closet. So that's on the closet upload part. And

04:59then we do everything else for you background removal, all of the tagging fabrics, dress code,

05:06colors, etc, on every piece of closet item. And then after that, the AI will start making outfits

05:12for you. And you can train your stylist by giving it feedback, you can say, Oh, you know, I really

05:17prefer a monochrome look on Mondays. Or I don't like wearing red furry sweaters with blue jeans. And

05:24these are these are real ways that users are training their stylists with natural language

05:29feedback. And then as the stylist gets to know you, it creates more outfits for you. You can also

05:33upload a photo of your face and your body to create an avatar. And then the Alta will generate

05:38the outfit on your avatar. And it's really cute. It looks like a little mini you and I love playing with

05:44my avatar. And it's something I think that has really unlocked for people, which is like,

05:49not only can you see a recommend an outfit recommendation, but you can see it visualize

05:54on yourself. I think that's a huge unlock. And then you can start asking Alta for prompts.

05:58So if you look at our social media, you will see our community has asked everything from Alta style me

06:04for a Costco run. Alta style me for meeting up with my ex or going on a first date where I want to look

06:13cool, but not too cool. Taking my three-year-old toddler out on a play date at the water park.

06:18It's a huge range of all these sorts of events that people need outfits for and therefore they're

06:24asking Alta. And then alluding to what we mentioned earlier, Alta can generate packing lists,

06:30Alta can make look books. So it's everything that you could possibly want in a fashion app.

06:34It's both utilitarian in the sense that it's helping you decide what to wear and also what to buy,

06:39but it's also aspirational and it's just fun. I think shopping and fashion is fun and AI is a new

06:45way for people to create content and to express themselves. And then how are you from the back end

06:50of things? Of course, the consumer can try and train the models or teach the models, but how are you from

06:55the back end really effectively creating what then the consumers can use? Yeah, absolutely. So under the

07:03hood, Alta is close to a dozen models. And the reason is because fashion is such a rich taxonomy,

07:11like everything from how the cool rhinestones of your pants are laid out to the fabric of the jacket

07:19that I'm wearing. Like these are all very visually dense attributes of clothing that you need to somehow

07:25encode into what a computer can understand. And so we spend a lot of time on even things like color. So

07:32there are around a million colors that the human eye can recognize and red and orange, you might not

07:40always say look good together, but a certain shade of red and a certain shade of orange on a certain

07:44person can look amazing. And so for us, even on the color itself is an in-house model where we think

07:50about what is the probability that this specific shade of red goes with this specific shade of orange

07:57on this specific person. And those are some of the underlying permutations and parameters that go

08:03into our model. And so we think about fashion both on the item level. So trying to teach computers to

08:10understand a piece of clothing, where this clothing can be worn, how to pair the clothing with other items

08:16in their outfit, but most importantly, try to understand someone's personal style. And I think that is to me,

08:22what's most exciting about consumer AI is that we can finally create products that to an end consumer

08:28feel like they're learning. Like when you use Altimore, you feel like, wow, my AI stylist is learning

08:32my style and I'm getting better and better recommendations. And I do think one important

08:38thing to note is from a consumer perspective, it doesn't really matter what models you're using

08:44under the hood. What matters is that you are providing a useful product that is also fun,

08:51entertaining, and makes them happy while they're using it. And so I also think that's really

08:56important for us to always keep in mind as builders of AI companies. Totally. When did you get into

09:02fashion in general? Because you said you have a coding background, but how did you get this other

09:07kind of side of your interests? I have always loved fashion. I grew up as an early user of

09:14platforms like Polyvore and Pinterest, and I would subscribe to Vogue and cut out my favorite

09:20outfits and put them on my dorm room wall in high school. The paper version of what you're doing

09:25right now. Exactly. And with glue sticks and scrapbooking paper. But I never thought I could do

09:33anything in fashion because I didn't know anyone in the industry. I studied computer science and math in

09:38college. And I was very interested in hackathons. So I think it was like my sophomore year, I won the

09:46Gucci hackathon. And so the CEO had flown in from Milan to give us our big check on stage. And that

09:52was the first time I'd ever met anyone in the industry. It was mind blowing to me. It's someone

09:56I still keep in touch with and is a mentor to this day. And from there on, it just started snowballing,

10:02like meeting more people in the industry, getting to know more brands, models, editors, stylists,

10:08which I honestly think all comes full circle because many of those individuals are now on our cap

10:13table. But I think the intersection of fashion technology is so interesting because it is two

10:19worlds that you feel like could be very different, but actually have a lot of overlap. And more so than

10:24ever, I am really excited to be in this intersection. Yeah, I want to talk about that cap table a little bit

10:29and just like your journey raising money and bringing people into this company with you.

10:34You guys have raised $11 million, correct? And that ranges from everything like Menlo Ventures to

10:40the Arnold family and like the LVMH family investing in you guys as well. What is the pitch that allowed

10:46you guys to attract both like the tech giants and the fashion giants of the world? That's a great

10:51question. And I'm so grateful for our investors. They're incredible. I would say the main pitch is that

10:59obviously fashion, e-commerce, retail, and you know, fashion is just where we're starting. We have

11:04very big goals to go into beauty and other spaces, but fashion is a massive industry. The way that

11:13people shop has changed dramatically in the last few decades from like the golden age of the department

11:20store to e-commerce, to mobile and to now with like AI applications. And you need a team that is

11:29not only like obsessed with fashion, but is also has a technical background to pull it off. And I,

11:34that was like the core of our pitch, which is that we wake up every day thinking to ourselves,

11:39how do we build the best possible AI stylist and personal shopping companion and unlock this experience

11:47for consumers? Many people think, oh, only someone very wealthy or a celebrity can have a stylist. And

11:53our opinion is that AI is such a democratizing technology. Now, any parent of any income demographic

11:58can provide their child with a tutor. And similar to like having a personal shopper or styling, like

12:03can we use AI to provide this concierge experience to any shopper? And in order to pull that off,

12:10you need a team that has fashion DNA, but is also obsessed with solving this very difficult technical

12:16problem. And then going back to the tech side of things, we mentioned before that the T in GPT is

12:21like really what's allowed you guys to thrive. Can you give me a little bit of an explanation of why

12:26that is and what the transition has looked like over the last couple of years? Yeah, absolutely.

12:31The models have changed dramatically in what they're able to do. And even now, well, the T in GPT has

12:40unlocked a lot of new models such as LLMs, but LLMs are not enough for fashion applications. And the

12:46reason is that like, we all use, you know, the chatbots and chat to be et cetera. That's for the

12:51most part, it's like text in text out and fashion is so multimodal. Like when you think about imagery

12:59and also video, like these are all ways that you can capture like how a garment might fit on somebody,

13:06how it might flow, how it might look when you're walking in it or posing in it. And so each model

13:12advancement has allowed us to unlock more and more features. The avatar feature that we have in Ulta,

13:17which is now one of our most popular features, like at our peak times, our community might be

13:23generating one avatar like every one, one and a half seconds, which is pretty crazy, is, was not

13:30possible a year ago. Like when we started the company, I was like, I really wish I could find a

13:34way to like create a mini me of myself like in Clueless, where she had her little like avatar in the

13:39computer, like generate that yellow plaid suit on her. And I was like, I wish we could do that.

13:44And in the early days of Ulta, when it was like stable diffusion V1, I would always try to generate

13:50models wearing Gucci t-shirts going back to like my gratitude towards Gucci. And no matter what,

13:58they would always come out with three C's like G-U-C-C-C-I. And I was like, this is terrible.

14:03And it's like now the models have improved so much where we can render even like the

14:09Lububu on your Gucci bag. And if that's the type of outfit that you want. And so each model

14:16advancement unlocks a new feature or a new way for consumers to interact with Ulta and an AI in general.

14:22Yeah. What is the revenue model today? How are you guys making money?

14:25Great question. The revenue model is affiliate. So whenever, for example, I asked Ulta for an outfit

14:32for Beyonce's Cowboy Carter concert and it pulled these Chanel Cowboy boots from my closet and a

14:39sparkly dress from my closet, but then it recommended a sparkly fedora from a brand that I hadn't heard

14:46of before. And so if you purchase via the app, then Ulta will take a commission. But I honestly think

14:52like this is just one business model. I think there are many ways to expand on this in the future,

14:58whether it's like product placement or more personalized branded content,

15:04more ways for people to potentially use premium features in order to access more expensive models.

15:12So just the beginning.

15:14The options are endless.

15:14Yeah.

15:15You mentioned earlier too that the way that different generations are interacting with the fashion

15:19industry has obviously impacted you and what you're building. How has Gen Z specifically and

15:23our spending habits and shopping habits and ways that we go about getting dressed every day impacted

15:28what you guys are building?

15:30I love that question because it's fascinating. When we were in our beta phase, most of our early users

15:38were professionals. So think like lawyers, consultants, finance, because they have to dress nicely for

15:45work every day. And so Ulta was part of their morning routines and it was helping them solve a problem

15:50in their lives. And I think once we release the app store, what's been really fascinating is how many

15:57Gen Z and younger users that have joined in this middle school, high school, college. And these are

16:04an AI native demographic, which I think retailers and brands need to be very aware of because they're

16:10growing up on AI tools. And so they expect to shop and be fed content and ideas in different ways.

16:19And so this new demographic interacts with Ulta very differently than the older demographic. And you

16:25can see it from the questions they asked. They will ask us to create features such as like if they're

16:31going to their first tailgates or their first game days, like they want to be able to like dress their

16:36avatars and like their school logo. They want to be able to have music at the same time as generating

16:42outfits as they're like shopping for new looks. So there's like so many different use cases and

16:48features from the younger audience that we're now actively building. But it reflects like how

16:53differently they shop and how differently they want to shop. Yeah, totally. And I read from one of your

16:59other interviews that around like 20% of our closet is like all we really use, which of course leads

17:05to a lot of waste of if 80% of our clothes are sitting in our closet and not making it out onto the streets at

17:11all. And I'm sure that especially with Gen Z caring more and more about sustainability and not

17:15contributing to this waste, that also plays a part in why they'd want to kind of reinvigorate what they

17:20already have, correct? Absolutely. Yeah. And that is one of the reasons why I think our community loves

17:28to use the product is they're like, wow, I have Ulta's put together outfits that I would never have

17:32thought to put together myself and it's helping me make the most of my closet. One of our more popular

17:37features is cost per wear, where if like you've told Ulta the price of an item or Ulta has pulled

17:42it in from online, like our agent has pulled it in from your receipt, then we can tell you, okay,

17:47you may have bought these shoes for 400, but now your cost per wear is only 30. And it's like,

17:52I always feel very, I always feel very satisfied seeing that number. But also I think probably by the

17:59time this interview comes out, we will have launched our partnership with Poshmark, which is really

18:04exciting because we'll be able to serve our community secondhand shopping suggestions within

18:10the app if they want to buy something vintage or used instead of something new. Totally. And speaking

18:16of like what you guys are working on now, where do you see this intersection of AI tech and fashion

18:20going in the future? I believe that the intersection of AI tech and fashion will only result in more and

18:29more personalized shopping experiences and to give everyone that concierge experience of having a

18:35stylist and a personal shopper. And I think in additional to being utilitarian where Ulta can help

18:42you pick what to wear, it can generate outfit ideas for you. Ulta is also just fun. And I think I call

18:49this category shoppertainment. I'm sure many people have used this term before me, but to me that means

18:55that like fashion is such an expression of yourself and you should have fun with it. And the best part

19:02of, you know, my job and my team's job, honestly, is when we get to see comments from our community

19:08saying that like Ulta has helped me feel confident in myself again. Like it is the reason that I get

19:13out every morning, like being excited to put on outfit and go to work. And I have even gotten emails and

19:19also like I've seen Reddit posts from our community saying I had long COVID and I really struggled with

19:25my sense of self and Ulta, like the outfits Ulta puts together and made me confident and made me feel

19:32like I was getting my own personal style back. And to me, like, I think that's why we do this. It's

19:37we're using AI to empower people to be their best selves, to have a positive impact on their lives.

19:44Because I think like what you wear, it may seem to, what you wear really affects how you feel

19:50about yourself and also affects how you show up for the people around you. And that just has

19:54so many impacts for your own career and your own life. Totally. And as a young female founder

20:01in this very, very male dominated tech world, what are some bits of advice that you have for other

20:06young female founders today? How have you gone about whether that's like the fundraising journey,

20:11just building the products and putting a team around you that can help you build it? What are

20:14some bits of advice that you feel like have really helped you succeed? I would say that's a,

20:18that's a great question. And the one piece of advice I would have, honestly, to anyone is just to,

20:24to keep on learning. And for me, that's could range for anything from asking our technical advisors

20:32for advice on new ways that I'm seeing other people train models or scale infrastructure

20:40on their consumer apps, or asking other investors or our community, like what sorts of features that

20:49they want to see, how are ways that they're learning about new apps, I feel like there's always so much

20:54opportunity to learn. And I always feel like the beginner in the room. And in terms of team building,

21:01I feel so grateful every day to be a part of a team that is so intelligent that I get to learn from

21:07every day. It's like, our team is our family. And it's truly the best part is getting to getting to

21:13see them. Yeah, I love that. As you move into the future, what are some things that you're excited

21:17about with Alta specifically knowing how much is changing at like such a rapid pace right now within AI?

21:24Yeah, AI is truly a firehose. I feel like I wake up every day and there's some new update or some new

21:31state of the art foundation model or way that people are doing things. I say for Alta specifically,

21:40we're very excited about a lot of the social features that we have coming up. I don't think

21:45anybody has really cracked multiplayer AI, like most consumer AI products right now are very much you and

21:50AI or you and the chatbot. So I'm excited to see what happens with our new social features and be a

21:56pioneer in that space. And as I alluded to earlier, I think fashion is the beginning of self-expression.

22:03There's so many other ways that you express yourself in addition to what you wear. And beauty

22:09is one such example. Like our community often asks us, what hair should I wear with this outfit? What

22:14makeup should I wear with this outfit? What perfume or cologne should I wear with this outfit? And

22:19those all represent, I think, new lines of business for us. And the retail landscape is changing and so will

22:25consumers purchasing behaviors of all of those sorts of products. My last question for you,

22:30a lot of the under 30 community is obviously focused on youth, people doing really big things at a

22:35really young age. How do you feel like youth plays into your career today? Do you think about it? Do you

22:40not think about it? What's the impact that it's had on how you've been able to succeed and grow?

22:45My honest answer to that question is, at least for me, the benefit of being under 30 is that I do not have

22:53kids or a pet or even a plant. Totally. You're free to focus on you and what you are creating.

23:00And so I think so much of Silicon Valley culture right now is about the whole 996, 997

23:06working hours. And I can do that because I have, I don't have any of those current obligations. And

23:13that might be one of the benefits of being under 30. Do you have like star goals with burnout ever

23:18because of that? I think I love building this product Ulta because when I see every day the

23:27emails and the DMs that we get from our users around the world saying how much they love the

23:33product, that makes me so excited to make Ulta better for them every single day. Totally. Well,

23:38thank you so much for joining me today. This was so fun to talk about all your building and I cannot

23:42wait to see what's next. Thank you so much, Alex.

Category

Transcript

Be the first to comment

Recommended