Driving the Right Data Is the Way Forward for Supply Chain | Black Swan Textiles | PI Apparel

So right now, I am very honored to have met Keith Hoover this morning and introducing him for the keynote for the day to kick off day 2. So he is the president and said sometimes even janitor at Black Swan Textiles. His early career back in the '90s onward included working with amazing worldwide, globally renowned brands such as Ralph Lauren, Target, and then as the VP of Innovation at Under Armour. We all know those names quite well, I imagine. He started Black Swan Textiles in 2017. His experience included a lot of amazing on-site visibility into manufacturing, into being in the factories, and he took all of this, you know, the many facets of business that he was exposed to, and has been able to use that to imagine impact and create a business and have success with Black Swan. I personally can't wait to hear all about it and look forward to all of us meeting Keith Hoover.

Thanks. Good morning. It's a pleasure to be here in, um, at the Union League Club, um, especially with Pi Apparel. I've been participating with PyApparel for several years, watching the sessions, and there's definitely some themes that we've seen progress on and others that not so much. So we'll just keep banging away. But before we get started, we'll just quickly— here again, this is places I've worked in some form or another, basically in product development. Either in color or fabric or something to do with connecting with our supply chains, who we've seen grow all over the world. Today we'll talk about a few things. And as we get started, let's go— let's take a trip back in time, something that we still can kind of identify with today. [FOREIGN LANGUAGE] So this is before the advent of PLM and 3D, so it It's not really a commentary on them. But let's take— what's actually going on here? This is the user, the guy. And that's input, the keyboard. And the monitor is output. And all the technology has changed. It's still the same. So let's put that in context of our industry. That's the merchant at the brand. He's the user. Input is product development sourcing, maybe design. Output, the supply chain, the vendor. So what happens when there's a problem? Well, you smack the vendor. And then— You pound on product development and source. Then you get product development to pound on the vendor. Then you issue a chargeback. But during all of this, the one thing that doesn't get touched the core processing unit or the process. I can guarantee you there's nothing wrong with that keyboard. There's nothing wrong with the monitor. It's the process that's causing the problem. And that's true in IT. That's true in our industry. It's true in just about everything we encounter in life. So this also could be social media pounding on the brands, who pounds on the supply chain, but you get the picture. So as far as product development goes, there are models that literally, well, all the major brands, all the enterprise-level brands follow. And it starts with kind of the merchant identifying what their products should be in the line in the upcoming season. It's kind of a crystal ball thing. And then you've got product development sourcing trying to build skews to that line. And then when somewhere off in the distance there's manufacturing. But there's, there's no clear cutoffs between these phases. There's a lot of overlap which tends to cause problems because we have in this current model of fashion seasons, we still think that people buy clothes because the weather changes. And in some cases that's true. Certainly the weather outside today indicates that. But there are so many more influences today about what product changes. So this, this model, this vestige, this anachronism of a fashion season continues to cause problems in our industry because it assumes that we can predict what the customer's gonna want when the product hits the store, and it also assumes that the brands lead fashion. Back in the day, in the '60s, when one year there were miniskirts and the next year there were maxi skirts, that was driven by the brands. It wasn't driven by social influencers. So think of once Ralph Lauren, Tommy Hilfiger, these guys, the icons of the last century. When's the last time you heard of a major designer driving the industry? So I suggest that it's time for a new model because this is a typical calendar. I'm not gonna, there's no name names, names named here, but this is not unusual having seen how different calendars work. You start the product development season with this executive concept alignment and 16 months later you issue production POs. And with all these phases, all these meetings and all these deadlines. After you issue the PO, 6 months later, there's product in the store. So it's not really a 16-month calendar, it's a 22-month calendar. And literally you could have 2 kids join the family in the time you get product to market. And if your calendar is that long, When it's time for the next fall season, you can't get feedback from the previous fall season because it hadn't hit the store yet. So we're starting with a process problem. And you can flow it out in a diagram however you want. Here's one of two seasons per year. Oh, but let's improve our calendar. It's basically the same thing. You're changing the sequence of meetings, but the length of the process is still that long. Oh, let's go to 4 seasons per year. Well, here again, it's not much change. The outcome hasn't changed, but all of the meetings and design 1, design 2 meetings that you seem to forget what you did last time, they continue to be in flux. So this is basically a plot of when lab dips were added at a particular brand. That red line in the center, that's the deadlines for when all lab dips should have been stopped. And you can see they continue after the fact. Now that's not because the merchants or the PLMs are undisciplined. It's because they know this— the closer they get to the market, the better their choices will be. Same with fabric. This is a plot of when fabric development requests were added. That yellow line in the center, that's a deadline. No new fabrics developed, but the vast majority of them happen after the fact. And well, how much does development cost? Well, we've built a calculator that is basically a spreadsheet with macros. That's nothing super tech, not AI. But we basically gather some relevant information out of if the brand has a PLM system or a PDM system or even master spreadsheets, whatever they have to record certain information, we gather that and then we basically then calculate some metrics around the cost of, in this case, color. So we have Over on the left side, we want to know how many solid color lab dips were requested in a year. So we went in, we went in, in this case we went into the FLEX and we pulled all that data. And then we looked at numbers, you know, how many rounds of lab dips does it take to get to an approval? Who's involved in, what roles are involved in lab dip approval? And then we actually, we pulled in their salaries too so we could calculate, a cost factor. And then since when colors are approved at a brand, a vendor is involved as well, and there are vendors doing the same thing because ultimately a vendor is responsible for color quality of the finished goods. If they ship off-shade goods, guess who gets charged back? Not the LabDip people, but the vendor. So vendor has the tools and the people to do the same thing, and they can do it cheaper because they're not in New York or places like this. And then we got some factors from the mill. How much does it cost them to produce a round of lab dips when you factor in time and equipment and all that kind of stuff? And then we also got the cost of running a production lot because we're not just looking at lab dip approval, we're looking at first bulk approvals. And what we found was in the case of a brand in the EU that's relatively They had the smallest number. But basically, when you factor in the cost to the brand, the cost to the vendor, all the courier costs, but most importantly, the rejected first bulk, they're spending almost $2.5 million a year on color development before a single yard is cut into a garment. So we can talk about sustainability and wasted garments and all that stuff, but what's going on behind the scenes is a bigger sin, and it's the outcome of processes. Literally every brand in the world has some type of lab dip process that generates this type of behavior. A bigger brand, a much bigger brand, they're looking at a lot more colors, but they're spending $14 million or actually they're not spending it. Their supply chain is absorbing $14 million and yet they're still delivering goods on time and they're meeting their cost margins. So it's a miracle. A brand with a very good digital process of evaluating color with spectrophotometers, all that kind of stuff, they're still, forcing nearly $7 million of cost on their supply chain. So what about fabric? Well, this chart shows that the fabrics developed at one brand, 30% of the fabric developments were canceled. Fabric development is a lot more expensive than lab dipping because when you develop a fabric, you're running it on a production machine. And if you're developing a fabric on a circular knitting machine, that means you're not running production. It means you're not generating income on that machine. So 30% of their developments are dropped. But fabric's just fabric. What's the big deal? Well, there's 4 classifications of fabric. Your carryover fabrics that you run season after season, you have to keep track of those, make sure they're still to spec. Whenever we move a sourcing region from Asia to Central America or wherever, not only does it— we move, change factories, but we have to redevelop that fabric in region. And replicating existing fabric is harder than developing a new one because you actually have a target that you have to hit. And you know if you're wrong, the customer says, hey, what happened to my Polo shirt? It pills now. So resourcing is a problem. New development. Sometimes new development is just copying a fabric that you found. And sometimes it's developing a fabric with certain performance attributes to make all athletes better or something like that. But it's an iterative process. It's not done. It's not driven by the brand. It's basically send suggestions to your development mills. They send swatches back. And then you tell them if you like it or not. So it's this guessing game type of process. A lot of times there's no true specifications for the mill to know when they've succeeded. And then finally, all the new bio-based fabrics, hemp, all this kind of stuff that we're seeing a lot recently. Well, the marketing people love that stuff because it helps them tell a story. It helps them promote the brand story that we care about things. But the fabric development people oftentimes are saying, are you kidding me? Hemp? I mean, hemp, it's a great story, but it's itchy scratchy, and fabric is next to body. And so you have to not only comply with the business requirements of sustainability, but you have to consider the customer and these requirements— comfort, durability, how well does it dye, What about fading or, or crocking? All the kind of stuff that tells our customer that we know what we're doing. So fabric is a tough development process and it's what garments are made of. So it's super important. And then there's 3D. Well, 3D's been around for over 20 years and there was a signature project done in the EU in 2004 called the LeapFrog Paradigm. And they basically in that year, 20 years ago, they were basically saying, what if we could design a product in 3D and then we could transmit those specifications to machinery and robots and make stuff? And they actually made, they made a blazer with robots for Hugo Boss. But one of the founders of Well, a founder of one of the 3D companies that everybody knows wrote— he wrote a long piece about the requirements for 3D software. And what he said was, a 3D tool is not a design tool. Once a design is approved, then it is transferred using this 3D tool to eliminate the need to create a final proto. And it made sense, but they pointed out that the requirements for operators of this software are much different than the requirements of a designer. It's an engineering tool. And in the intervening 20 years, the companies, all of whom their software is all based on this, they've made it better, but it's still a problem. Because the main use of 3D today is to help the buyer make a decision of which designs to put in the line. Because it's a great— I mean, it's a beautiful fashion illustration. It's better than an old 2D croquis made in Illustrator or a hand-rendered croquis. I mean, back in the day, a buyer understood. They could interpret in their head. What a 2D cartoon looked like in a product. But we're constantly bombarded with realistic images in everything we do. So there's this need to see what the product's actually going to look like. It's great. But when the brand decides, yep, that's going in the line, that's when the actual product development at the factory begins. These files that are intended to be transferred to the supply chain often aren't. In fact, even if they were, they'd be of little use because they're not actually building the garment, they're faking the garment. There may be two layers of fabric here that they— forget about that, I'll just change the physics and make it look like there are. So it helps. Improved the beginning of the process, but we're still hamstrung at the, at the end. And when you think of the revenue that our industry generates, you know, what is it, uh, $1.8 trillion in sales for the apparel brands, but you double that when you factor in the revenue from the garment factories and the mills. This is big and it gets bigger every year and it's never gonna go down. Because unlike the newspaper industry where when they went online, basically newspapers kind of went away. Well, clothes won't go away, at least not while there are public decency laws around. So this just a— it's a continuation. So our industry is big and has the potential to grow not only just incrementally, but in the offshoots of all the technology that we're developing. So the problem is— And all that revenue, 75% goes to cover operating expenses, not just buildings and people, but PLM systems. And I got to give credit to PLM folks, best salesmen in the world, best salesmen in the world. And I won't say why, but 3D, it's an operating expense. The goal of technology at the brand should be to lower operating expenses and lower cost of goods sold costs. Right now, if cotton prices go up, the brand goes to the mills and bangs on them and say, "Cut the cost of fabric. We don't want to pass on that cost increase to our customers." And when you think about the cost of apparel over the past 20 years, it's amazing that it's still super affordable because the supply chain has figured out a way to absorb a lot of that pain. So how do we actually connect development with production? Well, we're gonna go off script here a little bit because, well, actually we'll wait a second. We can show a couple of these technologies. These are just a few technologies that are in development right now. And I think are important. First, LED Simulator. LED Simulator, LED Simulator, uses new technology to specify colors. Currently, when designers are choosing colors, they're going through a Pantone book or an Arkoma book, or they're walking around found objects and they're cutting garments. Oh, I love this color. So they have to find a physical, an analog product before they can get started with color. Well, this basically, it starts in a laptop using a simple color picker scheme that everybody has used, Creative Suite is used to. And they're choosing color, they're choosing a hue, and then they get selections, and then they get to the color they want. And then what happens is the signal is transmitted to this device. Instead of the color going on screen, what's actually happening is this is a light booth that everybody's seen. The only difference is there's this hole in the back, and you can see what's behind the unit here. And what's behind it is you have these LED panels jutting off to the back illuminating this board that's about a foot behind the unit. And on that board, there is— you can place any fabric you want. In this case, we're looking at terrycloth because it's heavily textured. And when you have textured fabrics, it's not just the color that's important, it's the total appearance. How do the shadows impact that color? So basically what happens is the user is sitting in front of the light box. They're choosing their color on the laptop. When they get it, they click a button and the LED lights illuminate the fabric in back. So it's— and actually these are 12-channel LEDs, are super precise, and the gamut is— way bigger than the gamut of monitors. We could actually simulate 1.5 million distinct colors that are achievable on fabric, in which case we'd need 1.5 million and 1 for a designer, right? There's always that one. But it's a better tool to start with because it doesn't require a designer to go hunting for colors. They invent their colors, and then it goes to a digital and LAB value, then it goes to spectral data, which can be sent directly to the mills who do their dyeing by the numbers just like they have been for 10 years. So this is one cool thing. It's being developed by a company called Thouars Light, and there's a guy named Dr. Ronnie Luo who invented most of the color metrics that are used in color management today. The next thing, the Natific Color Warehouse. So when I was at JCPenney, we're doing these stupid lab dips. And I'm thinking, no, I don't care what lab dip's like. We don't sell lab dips. What's production fabric look like? So we worked with a company called Natific. And we came up with a way to go to the mill. Every mill that dyes fabric, before they ship it, it goes to a place called load release. They swatch it, they do some testing to make sure that shrinkage is right and all this kind of stuff, and then do a shade evaluation, make sure it's on shade. And they're usually doing it with spectrophotometers. So spectrophotometers, digital color management is in use in the industry around the world. You'd be hard-pressed to find a good mill that didn't have one. So what they do is they upload all of their production lot measurements to this cloud system. So there's all your mills are listed in here, and it tells the number of dye lots they run in this time period. It tells the average color difference of all those dye lots, and then it tells the precision color difference, which instead of comparing the color to the standard is comparing all the batches to each other, because that means you can cut these fabrics together without any shading from panel to panel. So it tells all the things that are important to a dye house to know if they're producing right first time. And then based on what the mill uploads, it automatically assigns them a score, a competency score. In this case, level 4 means that everything they're dyeing, every production lot is under a Delta E of 1, which is the standard tolerance that the industry follows. And if If you're better than that, then you're level 5, and that means every production lot is under 0.6. And we dealt with Masood, a mill in Faisalabad, Pakistan, and these guys were so good that for 3 months they were at level 6, dying nits. And it wasn't because they were sorting stuff out, it's just they had that much control over their process. So they weren't taking any more time than anybody else was, they just they knew what they were doing. So this is a view of— I'm looking at the production lots that were dyed at Swiss Tex in LA, and that oval represents the tolerance. So if a dot is inside the oval, that means it's on shade. If it's outside, it's off shade. In this case, they're all in tolerance. It's telling me that their average Delta E is 0.78 or whatever it is. And interestingly, you look at the top, go-go in-flight. So I was on my way to China, I was in the clouds looking at the cloud, looking at what a mill in California was doing and 10 and a half years ago. So this is not novel technology. It's actually, Natific has several million measurements in their database right now. Another brand, Under Armour, adopted this. There's several brands in the industry that have adopted this. But this process basically takes 3 months out of your calendar because you can add— literally, a buyer can add a color the day that POs are issued and the mills can handle it because it doesn't focus on lab dips. It focuses on production. Now is where we're going to have to— actually, we're going to have to go back. Going a little off script. I'm going to go back. We're going to go live to a couple of websites to see some new technology that's just starting up right now. OK, so we have basically a Mill fabric library here. Every Mill— that we deal with, regardless of where they are, they have fabric libraries. And fabric libraries, they can run from 5,000 to 10,000, 20,000 fabrics. And they represent a lot of work that a mill has done over the year when they're doing fabric developments for their customers, when they're doing fabric developments to go to fabric fairs and show off their goods. The problem is fabric has not been organized in such a way to be able to find this stuff. So what mill after mill tells me is they will get a fabric development request from a brand. The fabric technician looks at it and says, you know, this looks familiar. I'm sure we've done this. So they send somebody back into their fabric library, which is literally a room with header cards of fabric. And after about 20 minutes, the poor person gives up. You know, I know it's in there somewhere, but we got 10,000 fabrics. So what they do is they basically say, okay, let's start over. So what happens then is a knitting technician, in the case of a knit, he pulls a yarn, he unravels it, or she unravels it, looks at the shape of the yarn. They can then determine what stitches were used. There's 3 stitches in single knitting: knit, tuck, and miss. And then they look at it under a pick glass, or a microscope to count the courses and wales per inch, the rows and columns per inch. And then they set up a machine because those are the settings on a knitting machine, mechanical knitting machine. And then they proceed to knit a fabric sample. Well, what this does— we're going to wing it. So I'm going to create this stitch pattern repeat. The sequence of stitches that the knitter is trying to find by unraveling the yarn. When he figures out what that stitch pattern is, basically, he creates a grid. That's the repeat. That's the stitch repeat. And then he basically— let's try that again. Reset. OK, we're going to do a 4 by 4. Creating the grid again. Now this time we're just going to go and I'm going to enter some, some stitches. Now that's a 4x4 grid. I remember back from my school days that there's 16 possibilities If I had started with TTK as the first and then added the K to the other end, it's the same stitch repeat. When you think about that block, all I'm doing is repeating that square over and over again. So there's actually 16 variations that represent the same exact pattern. So we're going to search all possible variations. And what it's going to do is it's going to show me that here's all the possible variations. Okay, bingo. So here's a fabric with that stitch pattern, and this is what the front looks like. This is what the back looks like. We know that it is— we can record whatever the mill wants to record as far as how they identify their yarns. We know that it's 25 wales per inch by 32 courses per inch, and it uses this particular stitch pattern. Now every mill that produces fabric, regardless, for their customer has, oh, he has to test the fabric to these specifications. So some guys use ISO, some guys use ASTM. So the mill is just going to, they're going to test their fabric and they record those test results somewhere. So what we're doing is we're grouping all the test results that a mill has regardless of what system it is. And we're basically saying, okay, the ASTM test method for shrinkage is roughly equivalent to the ISO test method. So we're grouping them into a bucket for shrinkage. And so we're predicting what end uses a fabric may be suitable for other than what it was developed for. So we're leveraging all of this data from the past and to give a customer or the mill a better idea of not only is the construction right, but the end-use requirements are good as well. We're also partnering with a company called MTAC out of Germany to measure the hand of a fabric, because after color, it's probably the most important subjective aesthetic property. But color has terms like red, green, yellow, blue. Hand has whatever word you come up with. I mean, you've probably heard the craziest words to describe a way a fabric feels. So we are measuring it, we're converting it to numbers, we're creating groups out of these numbers that basically fabrics within this group feel the same. This gives the mill a better chance of actually knowing when they're successful at meeting the hand specification. It allows a brand to create a a signature hand so that if they want all their fabrics to feel the same regardless of whether it's polyester or cashmere, this is a method at least of grouping them and giving the mill some idea of how to do it because hand can be achieved either by the yarn, by the construction method, by softeners and chemicals, by brushing and mechanical efforts. So it allows the mill to use their knowledge how they've gotten finish is done in the past, so how they can achieve it in the present. The process of that KT, that grid thing that I had so much trouble screwing up on, we've invented a device called the Black Ops Fabric Decoder, or BFD, because it's a cool acronym. And what it does, I take the piece of unknown fabric, I scan it on a conventional scanner. I run it through some algorithms in the cloud and it outputs that grid. Then that can go directly into the system and search it out. So right now, this model is good for a mill, but very few people at brands would be able to define their fabric this way. So the BFD device would then be a way for brands to access that the library, not only their own library, but if mills want to post their libraries online so that brands can search them, they can do that as well. And of course, what the designers want is, can that be an app so I go into Bergdorf Goodman and scan a $1,000 blouse and knock it off? Well, I suppose it'll eventually be, but that's not really the purpose. We'll skip the— Tailor Essentials, without going into a demo, it is a way of collecting these digital assets. I helped a brand implement 3D, and one of the things we discovered, it wasn't enough that the software is fine, but you're supposed to have pattern sets to load into 3D to wrap around the body. Fortunately, this brand, in their vendor contract, it said, had told the factories, all technical work you do belongs to us, the brand. But they're one of the few brands out there that have done it. So if I'm at another brand without that contract in place, I contract the factory, hey, can you send me the blocks or the pattern sets for this? They'll say, nope, you don't own it. So, but factories do own this. So whether you're a factory or a brand, What Tailor Essentials is basically a way— it's kind of like a digital asset management system. It's collecting all of those pieces into a folder so they can be reused in the future. And in a sense, since most brands don't have— they've not organized this, many mills haven't organized their own fabric definitions into any online system. It's basically data acquisition that needs to be tackled first. There's a lot of talk about integrating design with manufacturing. Here's what happens when you do that. Because in the old days we were vertically integrated, the brand owned their factories, it was mostly domestic, but it went away because of labor costs and, and various issues. It went away. And it became this current model, this fragmented supply chain of contract manufacturing. And dotted lines mean there's no ownership. Each of those little buildings represents a type of a factory in the supply chain. And for every building that you see, there's probably at least 10 versions of it out there. So it's a web of— it's a messy web. So the promise of 3D was, hey, we start with this beautiful garment and we have this digital fabric and digital color in this one pattern set. How do we— we want to transfer it with a counterpart in production. So in the case of— it's actually— there's great mature technology to take a pattern set and turn it into a marker. And then the factory takes care of fabric yield and utilization, all that kind of stuff. Color, well, most systems are using RGB, which is unfortunate. But if they embed spectral data, then we can go direct to dyeing. And if we use this new method of digitizing fabric, then we can go straight to knitting. And then with a digital spec for knitting, you have a digital spec for dyeing, cutting, sewing, and voilà, you have a garment. But that doesn't work because we're not set up to do that in the current model. We don't have the required expertise, nor do we have the willingness to do it. But that, you know, all is not lost. Let's start with manufacturing. The factory for that garment, they know exactly what's necessary to make the fabric, to finish the fabric, to cut the fabric and sew it together. They have real bills of material, real bills of process, order of operations, all the kind of stuff. If you placed another order with them in a month, they'd know exactly how to make it. They know how to set up the sewing line to get optimum efficiency when sewers, they could do this all day long. But, so in essence, this is a key next step for the supply chain. If you talk to a factory, a lot of them wouldn't be able to find this stuff because they're not set up to do reorders, they're set up to do new orders. However, if they aggregated these assets and had some type of automated 3D system that they could basically say, here's a folder, here's my avatar, here's my pattern pieces, here's my fabric, and boom, it creates this nice little visual 3D that is a physical representation of that real product, plus it's a digital representation that if I'm using Google Lens and I see a garment that Macy's is selling, I can search this and say, oh, It's that garment. So this idea that we need months and months and months to predict what the customer wants is ridiculous. I mean, all the new startups know that there are other ways to figure out what the product is to put in the market. When you look at the styles out there, there's not a whole lot of variation. So if you tweak, add a pocket, or you tweak the silhouette a little bit, you're still going back to square one. You're going through fit development and all this kind of junk. You only do product development when— for a new, significantly new product. The factories that are producing this stuff, guess who owns all those patterns? Let's say a factory in the past 10 years have than 10,000 styles. Their inventory, ready-to-manufacture inventory, is right there. So if you developed a new tool to determine what your product is, like Amazon does when they tell you to buy this instead of that, you literally can send an order to a factory, and depending on their planning, they can have the garment ready 2 to 3 weeks. So you go from 22 months to 2 to 3 weeks. And these factories are all over the world. They're in Asia, they're in Latin America, they're in Europe, not so much in the US for other reasons. But the next brand isn't Ralph Lauren. The next brand is Factory A, who is vertically integrated, who knows how to make stuff, who has a proven track record of making stuff that sells, They could even go into their ERP systems and track, well, when we made this style for Brand A, what was the quantity? How many did it sell? So they have their own internal metrics that they could be running AI to drive revenue instead of AI to do silly things. So this whole notion of a fashion season, it goes away. Open to buy. Makes sense because in open-to-buy, basically if you're making a product and it's selling, you keep making it. You're not making huge amounts up front that you're going to ship to a warehouse and sell them out over the course of a season. You're going to make it in a small quantity. If it sells, you keep making it. If it sells, you keep making it. If it stops selling, you stop making it. And then you look on to what, to what else is in the market and you quickly add a new product in. So that's basically, this is the new sequence. You come up with a new way to identify the product, whether it's Amazon or any of these, the AI stuff that's driving that. And then you go straight to production. You go to, no, there's no product development. You go straight to production with factories who have made this stuff. So they're reusing the assets that they've done before. And then depending on where the region is, you basically aggregate your logistics. So you're, you're shipping containers or whatever, or trains or, or whatever. And, and then basically you have data about the whole process and you, you constantly improve the process, what works, what doesn't. So you have a closed circular system of data not fabric, and it drives more efficiency. So this is operational excellence. The outcome of operational excellence is sustainability. I mean, you can't help it. If you do things right and efficiently and you're not wasting and you're using the right materials, you are by definition sustainable. So let me just finish this thing off. There's one view of the apparel industry by the people who make the clothes, by the people in the factory. There's another view of the apparel industry by the people who read social media. So I've taken a soundtrack that represents social media, and I've put it on top of a video that shows a really cool factory and the great stuff they're doing.

But the life and joy I know is scared beyond belief way down in the shadows. And the perverted fear of violence chokes the smile on every face. Common sense is ringing out the bell. This This ain't no technological breakdown. Oh no, this is the road to hell. You must learn this lesson fast, learn it well. This ain't no upwardly mobile freeway. Oh no, this is the road. This is the road. This is the road to hell.

[FOREIGN] So you think of the brilliant engineering that went into the equipment, the processes, the sequence, and the skilled workers that are basically choreographed with this whole process. That's brilliant. And you compare it with a process that we take 22 months to go from concept to in-store. Which should we be listening to? And I'm saying, we have a great industry. We have all kinds of opportunity to improve what we've done in the past, and there's a new model coming down the road we better be ready to take because people want clothes. And it's just how are we going to make it, how are we going to provide it, who are the best parties. So that's it. That's my presentation.