Editor's Note: This transcript was automatically transcribed, so mistakes are inevitable. You can contribute by proofreading the transcript or highlighting the mistakes. Sign up to be amongst the first contributors.
For many, the idea of walking into a store and finding new clothes that actually fit is nonexistent. The designs are either too bulky or the length of the sleeves and pants are too long frustrated. Many leave and do their shopping online instead. Except that experience isn't much better. Servitude co-founders Nick Clayton and Camilla Olsen are on a mission to help turn that situation around. The idea behind the company is a radical one to recreate a designer's brain by using code.
Nick and Camilla joined EITE visionaries to discuss what that means and how they are using AI and machine learning to find clothes for everybody.
Enjoy this episode. It visionaries is created by the team at Mission Dog and brought to you by the Salesforce Customer 360 platform, the number one cloud platform for digital transformation of every experience, build connected experience, empower every employee, and deliver continuous innovation with the customer at the center of everything you do. Learn more at Salesforce.com platform.
This podcast is created by the team at Mission Dog.
Welcome to another episode of it, visionaries, I mean, Faizan, host of IT visionaries, and today we have two special guests. First, Kamilla, how are you? I'm good.
It's a beautiful day here in Palo Alto.
It is a beautiful day here in sunny Oakland, California, as well. So I am excited to chat with you and Nick.
How are you doing? Well, it's beautiful here in Ann Arbor as well.
Oh, there you go. We're spread all over and our producers are in New York.
We got we got the, you know, half of North America covered here. But we're excited to get into everything that's going on, its attitude and your backgrounds.
So first, for our listeners who don't know, can you share a little bit about the company of Servitude is a retail technology company. And what we do is we once we take what was once very doable only by highly skilled practitioners in the fashion industry, and we allow anybody to perform these tasks within an executable framework. That target audience is the the fashion designer.
So we do design and recommendations based on body shape. So our goal is so that anybody can get close to help them look their best.
That's the consumer goal. But for the retailer and the fashion designer, we want them to build the best assortment that they possibly can in an easiest way they can do it.
Nick, as CTO, what is the scope of your responsibilities? So it's the CTO and obviously primarily responsible for the product. So getting all of the data and the technology built, that lets us. Really facilitate these designers and let them. Design, but bring in that analytical side of looking at things like trend analysis and things like partnership.
And so we're taking a step back. How did you get started in technology?
Well, I started out in high school studying.
Always wanted to do something in math, but I took a started out an electron microscope and spent most of my time in the biomedical space.
But my last two companies that I did, there were predictive modeling companies in the pharmaceutical space. I took a break and went to the fashion design school. I had my own fashion label for five years when I learned about this problem of returns. Then one in six items goes to a store, ends up in a landfill. And that's astonishing in and of itself.
But the part that I learned as a fashion designer is that the reason that that happens is that the industry is focused on the hourglass body shape, not actual bodies. And there are only 20 percent of women have an hourglass body shapes. So I realized that I could solve that problem by bringing in predictive modeling techniques that I used back in the pharmaceutical industry. And so I didn't realize I was doing going to be doing I. But I wanted to solve that problem.
And Nick and I pulled together on this and that's what we ended up doing. So he tricked me into it.
OK. Give us a what's the what's the founding story here, Nick? How'd you how'd you, uh, how did you trick her into it?
So I've known her for a very long time. I actually grew up in Palo Alto and did musical theater with her son Mark growing up. But also growing up, I did a lot of coding and robotics and then I quit school and developed some computer vision and machine learning. Comilla reached out to me with sort of the basic idea of, hey, there's this problem in fashion. People are really only designing for the hourglass body shape. And I think there's ways that we can address that.
And as we started approaching the problem and looking at solutions, we drove further and further towards this sort of full, fully automated enabled solution, because that's just really what made the most sense for solving the problem.
So what we so we did here is we focus on the the real problem that women were having. So in my design studio, I had a lot of customers in Palo Alto who would come and say, what do you got coming straight on? And I see these sad faces and me feeling terrible, like I'd done a bad job, that my clothes didn't fit real people. And I knew I graduated with honors.
I knew I was a good student, but how could I be such a bad designer?
And I just felt horrible about it. But I realized that design school don't they don't teach you how to design for real bodies because it's designed for manufacturing to be efficient and we're supposed to figure it out or I'm not sure what they expect women to do, but we turn things and put it in the landfill.
But I spent actually years studying the problem, trying to to figure it out. But Nick and I spent a lot of time sitting down together, and we have another designer in the on our team to juggle her. And the three of us really literally sat down side by side for five years solving the problem together from a real customer perspective. And and that's how we ended up solving this big problem that other people have been trying to solve algorithmically. But we ended up putting the problem into the A.I., which is, I think, the magic that we have done that hasn't been done before.
And I think it Nick, if you can talk to that, I think that would be pretty interesting.
A lot of solutions out there sort of take a cookie cutter, approach their algorithms out there that if you throw enough data at them and the data is relevant, you'll end up with results that look OK. What we did is from the start when we actually designed the structure of the algorithms, we were using, not just the structure of the data and not just the data that was going into it. We brought design philosophy in to that stuff as well.
And so what that gave us is it gave us a system that actually is able to make inferences a lot quicker and with a lot less data because it is structured in such a way that. The structure flows naturally into the types of inferences that it needs to make. So that sort of domain specificity from the very beginning, rather than on boarding domain expertise, once you already have the system built, that process of engineers and designers sitting down side by side really, I think, made our selection much more powerful than it would have been otherwise.
Pardon me, but the kind of thing that we've created from the get go, we created a fashion designers brain. We've really put in encode how she thinks. So when we make recommendations, it goes with the same process that I would in my studio trying to fit somebody. We actually thought of all of those situations. And so what we're finding years later is that that same knowledge base and everything around it, and it is such an efficient way to address this problem that it can be used to solve many, many problems in the fashion retail space, because that thinking has been abbreviated in so many spaces in the industry.
And we're able to add in the necessary complexity in A.I. to solve where there's so much waste. And so we're now have in now just about to introduce new products on the design side using A.I. design and also to on the supply chain and then also in the on demand, manufacturing and redesign. And to me, the I'm not saying that to. To sell products, that's not your audience, but to to say how important it is to bring the domain expertise in the solving and to creating a solution because you're creating a real efficient approach that can be used in many more places than you initially thought.
Well, you know, like you said, the the problem is enormous in terms of, you know, five billion pounds of landfill waste and carbon, 15 million tons of carbon emissions annually. Like, clearly there's a massive kind of issue, but all of the different players that are involved don't necessarily know the totality of the issue or maybe they know it, but they don't really have a solution. I mean, you know, we've actually talked back to probably one hundred and fifty episodes ago about this idea of like how clothing and sizing and things like that haven't really changed too much over the years with with the stitch fix and and just talking about how, like it really needs to change, like how you from the elemental level of of how people are designing needs to change.
Otherwise it's going to continue to to have a ton of waste. How does your solution kind of help the folks who are actually designing to make the process easier to to limit the input for what could be a ton of waste?
So on the on the design side, so we have just we're just going into beta today with our design product. And let's let's describe that in the first case. So there are other approaches to doing fashion design. But and let's clarify the landscape, because people are calling the 3D products design.
But that's where we're talking about ideation in terms of fashion design and 3D comes after ideation. And that's to Perfecta, your design.
There's a Gane process and Nick can address that.
But we have an approach to design, which is very much how a fashion design and what approach it. So we take inspiration images. It's in our system is entirely image based. So we take images of inspiration. We take a designers archive, archival history of everything. She is design that's relevant to what she's about to do, and that gives us her history of design, details that she can draw from and it gives us her drawing style. And then we take a trend, which is what's coming up in the new season.
And it's a combination of those three things that we use to create new designs and iterate and we can actually iterate trillions of possibilities. What makes our system sustainable is that we fashion designer can only design for one or two, maybe three body shapes at a given time. And our taxonomy, we have seven hundred and twenty nine different body shapes that we consider.
So in our system we can test for all seven hundred twenty nine to make sure that the collection you're producing matches your customer population. So when we design or performs or ideation and puts together her collection, she can do an analytical step to test to make sure, well, does it match these all nine body shapes to match the different heights doesn't match the different proportions of torso and shoulders, what have you. And so by doing that, by making sure she has coverage for the population of people that exist in her customer base, that's how you keep clothes from going in the landfill, because she knows that what she's going to what you're going to ship will serve your customer base.
What ends up in the landfill are clothes for all of our last people that don't serve the whole population.
And so, you know, when you when you talk about those clothes, I'd imagine that, like, this is the first part of the problem is like, you know, I guess to take a step back, like, why would there just be like size zero, two, three, four, five, six, you know, and upward? It just seems like it's very like the entire kind of process is pretty like old fashioned, like it hasn't changed in a long time.
So I'm curious, like, obviously one piece you just described is the design piece. But then also in manufacturing, they still then they would have to create tons of other versions of these clothes. Is that also a sticking point?
Let me pause here for a second and stop you halfway through your sentence as I didn't address this specifically size. Part of our goal is to is for designers and retailers to have an optimal assortment. That's what they are, the collection that they're setting out to be manufactured.
There's a geometry, if you can just imagine that someone who is a you know what, a shift dress, a sheath dresses. It's something that's pretty form fitting to your shape and that would look great on an hourglass shape. But if you have a woman who has a large she's top hourglass. So she has larger bust or bottom hourglass. So she has larger hips, she would never fit on her. Right. What she fit it for one part of her body, but it wouldn't fit for the other part of her body.
So that's a bad dress for her. But for someone who's a bottom hourglass, something called an empire dress is ideal for her or a wrap dress because it's for giving on the bottom. So there's a concept of getting the right silhouette for matching silhouette for body shape and also matching design details for body shape. Like, for example, someone has a large chest. You don't give them a round color, you give her a V V neck is much more attractive.
So there is a choice in design details. So if you have a right in creating your assortment, getting the right mix, silhouette and design detail, you can serve your whole population.
Have you back a little bit of what I was talking about here? That serving that variety doesn't require that you increase the number of styles or use that you're creating, what it requires is that you're more intentional, intentional about the way that of those styles are distributed. So you can create an assortment that serves everyone with the same number of years as you were using before, just with a set of skills that are better distributed so that there is something for everyone.
And you're not overproducing dresses for one body shape when they're only 20 percent of the population.
So the problem is really an assortment problem, not a size problem. And so if you have the right assortment, you can still use your size system, but you just have to have the right silhouette and design details to match your population. And the thing is, it is a really complex problem for an end for humans to do without I. I tried it, it's really hard. And so one might think you can solve it once and just repeat that.
But the thing is, the fashion trends and styles change and it gets refreshed, refreshed every season. And so that needs to move around quite a bit. So it is a very dynamic system. So you have to keep solving it. And so our system takes all of the factors that are used each season to solve it.
So let's say you have someone who is from, you know, like retailer X and they are designing for like a new a new dress for the season or a new style. And they're leveraging your platform to design this. How does it go from that point on to the floors of the store? Would it be that instead of the rack that has one double zero one zero four twos, you know, three sixes like like how does it change the actual way that inventory works out?
Imagine that especially with, like, direct to consumer type companies, that you could have much better inventory, you know, not in stores, then you could have, you know, as an in-store experience and really like change the entire in-store experience potentially with having a better fit in general.
I'm not sure if I can answer your question with this, but I'm going to offer this previously explained to you how to make an assortment. I can say that we can do the same process to create the right assortment for a store. So a designer herself, she has her own customer base and so she needs to create a product for her customer base and then a store, a buyer there needs to buy from many different designers. And each designer has different parameters in their in their collections.
Right. And that all has to come together to match the store's customer base. And so our technology can be used to make all of that come together so that the store is serving their customer base as well.
Is that the problem that you were asking me about?
Yeah, no, that that's definitely part of it. I mean, it's a pretty it's a it's a complicated system. That's why it's so, so interesting that to leverage to solve the initial problem of creating thousands and thousands of variations. And then I was what I was getting to is. You know, at the end of the day, this affects the actual physical brick and mortar retailer in how they present the product to new customers. And I'm curious, like, how does how does technology play into that interchange?
Well, we have not gone there. We thought about it. And so in the individual mom and pop store, it's a harder of course, because they can't they don't hold as much inventory as anyone else would.
They would have more of a personality and help cultivate their specific personality of customer base.
Whereas you could see by using this kind of technology, it can be more differentiated at the store level. Right now, everyone's going after the hourglass body.
So that's why if you look at the average street, so many people look awful because they literally cannot find clothes that fit their body. In our research, fifty to fifty five percent of women literally have given up on shopping and hate it because their experience is so bad.
Yeah. So I think what we're doing really is open going to open up opportunities for the fashion industry to serve many more women well and allow fashion to be more differentiated and offer more and to have people be women than men to be more satisfied than sorry.
Speaking the feminine, if it's frequently that we're inclusive, we also serve men and children and our technology as someone with, you know, with a mother and a sister and a and a fiancee, I can definitely attest to the fact that, you know, until you go shopping with your significant other, you couldn't possibly imagine how different it is. I mean, it's crazy that we got to this point in general that there's like that sizing is the way that it is, because there's so many aspects of the human body that are different that it's like, how could you possibly say that?
Like, something is like a small, medium or large like feels there's that doesn't account for all of the different parameters.
It's the same thing every time we tell the story when we're often talking to men.
But there are often share very similar stories with us too. So it's it's women don't own the real estate on this.
Also, it's a pervasive story, but this application of where the opportunity is in solving the the waste in fashion is is where we can apply this technology to to facilitate.
So 3-D, for example, 3-D, I think, has a big opportunity to help fashion designers refine fashion design and is instead, to some extent, it's been accepted. But what the problems with 3-D is that it in fashion design development, it it takes a lot of time. And so there's been some resistance on that, whereas we give designers time because we actually save a month in product development, because we do a lot of things really fast so that we like to say we take all the we do all the left brain stuff so the designers can stay in the in the in the right brain and stay in the designer space, in the creative space, but on the retail side.
So here's an opportunity in the brick and mortar.
People are talking about using ARMM, VR instead of especially at a time recoded, that that could be a real game to try it on virtually, except that, you know, it takes time to for it to load up and for you to visualize it and it takes time to just visualize it and see and more often not. It's disappointing when you try things on even virtually.
And that's a whole story you got to go through. And so one of the things that we can do is that we can preselect what's in the store for the X experience so that when the customer wants to try on X or she'll have the optimal pieces that match with her body shape or his body shape so that the experience with x ray will be. More satisfying and likely to be successful. I want to get under the hood a little bit on on the A.I. specifically, Nick, can you share how the A.I. is is performing these these the massive amount of compute you're looking at?
Yeah, so what we do is we look at primarily two sources of input and everything is built on top of that. The first is body shape and proportion of people, and the second is silhouette and shape and design details of clothes. So we have a visual recognition system that allows us to look at photos of clothes and we can pull out from pretty much any photo photos, either from a retailer's storefront or from Instagram or social media. We can look at them all and we can say, OK, what's what are the clothes here?
What are the design details that are showing up in those clothes? And then we've created a taxonomy for that and the understanding of how those design details interact with each other and how those design details interact with the body shape and proportion. And so when we're looking at making a recommendation, for example, it's a simple calculation or someone's calculation. Once you have the underlying eye of what's the the score of that garment relative to the shopper's body shape and proportion, when you're looking at creating an assortment and creating a design, you're flipping that.
And so you're saying, OK, what's rather than looking at all of the garments that I have and making a recommendation, I'm looking at all of the people that I have and trying to figure out what kind of garments I should be making. And so we look at when we're creating those garments, pushing them towards the real body shape, distribution of your customers, trends that you're interested in. And so we can look at what's coming down the runway or what people are sharing on Instagram, and we can push the design generating system towards those types of designs.
So obviously, it's still going to stay within your style and your brand's DNA. But depending on how it's configured, you can push it a little bit or push it a lot towards your inspiration or your trends that you're trying to fit to.
As you're talking with designers who I would imagine are not a geeks in most cases. How do you kind of like talk through what what what the software can do for them?
Yes, we have always had a philosophy to make it really easy for designers and retailers alike to use our system. And that's part of why we've built this sophisticated visual recognition system that we have. As most people understand, images and designers particularly are fantastic in understanding images. That's kind of their job. So our system relies most heavily on images as an input. And so that's a very natural language for designers to talk in. So when we allow the designers to create their inspiration within our system, that's the same way that a designer would pull together a collection of images as an inspiration for their designs.
Normally they can take those same images, upload them into our system, and our system will pass out the fashion details from them, whether they're images of clothing or whether they're sketches or faces with particularly interesting design details. We can look at all of that type of input, and we're really trying to keep it in that creative space for designers since they aren't, as you say, necessarily. There are some with some background in statistics, but the vast majority don't.
Very few people really get what we're doing until they see the example of what we're doing.
Yeah, I mean, as with most projects, I think it's like once you see what's going on, then like, wow, I can't believe this.
Yeah, it's definitely a bit of a see it to believe it. Phenomenal.
When I think as you mentioned, you lived it. Camela, you you know exactly how it feels on the on the user side of trying to do this and and not having the ability to to scale what you were doing further than just, you know, kind of hammering away.
Well, that aspect of being able to design like we were and we did a text. Program in partnership with Target a couple of years ago and CEO of Target asked us, well, to look at their website and tell them what we thought. And then we looked at theirs and we've looked at a few others. And and our conclusion from that, and that's really what threw us down this road is, is that we realize that there are pockets of women who are walking out of stores because they literally cannot find any clothes that fit their body.
Nothing nothing can fit at all. It's not a size problem. It's because it just proportionally will not. They can't close the button. They can zip it. And it's not because they're fat. It's just because it's not made for their shoulders or their their their torso or their long legs or something.
And it and it's really it's really sad because there is a lot of people who've gone through life this way where they just can't find clothes that fit their body. And I'm really empathetic to that. And then when we looked at the numbers, it's it's a much bigger number than you'd think it was. And, you know, that's the problem, just needed to be solved. And so how much are you talking to?
Like larger retailers, like the targets of the world to try to facilitate, you know, designers or suppliers that they work with to leverage the platform?
So we're just getting so this is again, this is one of those things that when you talk to people about it. Yeah, it's a nice idea kamela, but you're going to have to show it to me when you have it done.
And so we just finished, we're just finished our we're going into beta now.
So I have an approach that targets the world. Yet with this product we've talked in theory with with many people about the idea, but I haven't shown them tried to sell this yet.
So we are going into beta with someone who's the local in this area.
But what I think what we're going to do first is to open this up on our website for designers at large to use so that we can add to our database and build start building a good database of information for the design, details and trends and many other aspects of what we're building.
And so with after that, at the end of the summer, I think we will then start talking to some of the larger companies.
Awesome. Well, let's get into our lightning round questions. These questions are fast and easy, just like the sales customer three sixty platform. You can go to Salesforce.com slash platform to learn more about the number one cloud platform for digital transformation of every experience.
Salesforce.com slash platform to learn more lighting. Real questions. Are you both ready? I'm scared. Yes, you should be just kidding.
OK, what hobby or habit have you picked up during shelter in place? I'm making Fred.
I've been playing a fair bit more magic the gathering lately. What is a. Book or podcast or something that you've really enjoyed.
Recently I have started listening to Michelle Obama's belongings becoming I'm a big fan of everything by Brandon Sanderson.
How about a hidden talent or passion?
And I always I'm really regret not accepting my adviser's request that I apply to medical school.
I don't precisely keep it hidden, but it's not necessarily obvious. I've been a singer throughout my life and greatly enjoyed it. Last question. This is a little different for both of you, so. What is your best advice for first time CEO, Camilla? Listen, more than you talk, and then what is your best advice for a first time CTO?
Nick, I don't think it's really important as the CTO that you be involved in the discussions that you have with customers so that you build a product that's really for your customers. Awesome.
Well, that's it. That's all we have for today. Any any final thoughts?
This is really fun. Thank you so much. Yeah, this was wonderful. Awesome.
Well, thanks for stopping by and we'll will be following along.
So hopefully we we get some good, good results after the beedis launched. We'll let you know. Thanks so much.
Take care. It visionaries is created by the team at Mission Dog and brought to you by the Salesforce Customer 360 platform, the number one cloud platform for digital transformation of every experience, build connected experience, empower every employee, and deliver continuous innovation with the customer at the center of everything you do. Learn more at Salesforce.com platform.