Is AI Changing the World of Fashion? with Omnya Elshafey

0:01 Artificial intelligence is changing the world of fashion and manufacturing with today's amazing special guest, Omnya Elshafey. Now, Now, Omnya, I'm so excited to have you here because this is something near and dear to me. Fashion textiles, we often think about AI as something that just lives inside computers, but we're seeing more and more effects on manufacturing in the real world. And I guess I wanted to see from your perspective as someone who's worked around both of those things, have you noticed how

1:22 some things were already changing in the world of in this kind of world of the physical? Jonathan, it's a pleasure being with you today. Absolutely. The AI has a huge impact on the fashion industry ~ nowadays, ~ on on many levels, starting from designing and ending with even ~ yeah, go ahead. So ~ let's let's talk about the manufacturing. Can you start from there? I'd love to know what kind of changes you've already seen. So ~ like i fashion industry it's all about the details.

2:02 You can imagine like one shirt, the thread, the fabric, the bottles, like many, many accessories going into one piece. And keeping the inventory and creating the production plan for such a thing was really a hassle ~ years ago before even ERP system was something. And then, of course, with ERP system, things started getting better. But then with the AI today, like creating the design with all the details and optimizing the inventory and optimizing the ~ production planning and build the right line.

2:39 For each item. This is very, very crucial in the fashion industry. Like it's not like you can have the same setup for every single piece you're you're producing. Like even if you are only a shirt manufacturer per se, for every shirt you have to change your the layout of your production line. And now, and this would used to take sometimes weeks. to come up with the best production layout to get the best productivity. Now AI make this happen in a few minutes

3:09 once you feed it with the right data. So, one of the challenges a lot of people don't know about is that Every piece of clothing has to be made in different sizes. And predicting, are we going to sell a bunch of size eights or a bunch of size tens is really important because it doesn't matter if you have a bunch of size 12s if the person who wants to buy it is a size eight. I know this has happened to me before.

3:35 There's something I want to buy and they don't have it in my size. And I'm not you can't take a different size. You can't take a smaller size and won't fit. You can't take a bigger size, it will be loose. And I know that this has often been a challenge in the fast industry, trying to predict. Not only which items will sell, but which sizes of which items. Do you think AI can help people to start to predict or even do dynamic predicting

3:56 if it's a shorter cycle from selling a batch and then creating the second round or the second run? Absolutely, especially that ~ like years ago, I'm gonna talk about like almost 25 years ago, there was like a basic ratio, how you create the sizes, like how many each batch, how many smalls versus large versus extra large. Now a lot of things changed in the world around us, and if the AI can take those data, which is pretty much the sizes itself of the people,

4:28 like the percentage of like like how many people are now into. gym and taking care of themselves versus like all these weight ~ management like this is has a huge impact believe it or not on the fashion protections for for sizing as you said. So now that AI can very quick do these analyses and come up with the best size ratio for each batch, once you but here's the thing. What are you gonna feed the AI to give you this piece of information?

5:00 You need to give him who are you ~ targeting with this item, the size range, the the the location, the ~ like ~ the market, which market. Once you give AI these three things, it will be able to predict the best size ratio for each batch you're producing. So in the past, the manufacturer would know how many units had been ordered by each different storefront or Who was selling directly to the customers, but they wouldn't know until for quite a

5:35 while until the end of the season, right? Which one sold and which one didn't. Is there now the possibility to get that information with a shorter cycle so they can say, ~ how many units are selling each week or even each day? So that the manufacturer can actually do more dynamic creation to kind of man like load manage? On daily basis. You can get this information now on daily basis and there is tools in the A and I that can right away tell you that

6:01 within two weeks you're gonna have reorder from these colors, these size, so you can be ~ ready for your next batch of production. Do you think that we'll get to the point where the manufacturing is so dialed in that like outlet stores disappear because they no longer have all this leftover inventory that was the wrong size or the wrong style that that kind of they get that dialed in. This will ever happen. Honestly, we're talking about when you talk about fashion,

6:37 it's a human taste. Like in my own experience, we would sit in the designing room, create this collection, and one of the designs, everyone in the room will be like, my god, this will be our best seller. And then we pick another design just because we have to complete the and then You take it to the market, surprise, the item that you thought it was gonna be the best seller is barely selling and the item you thought that you were

7:04 gonna just put it there to fill in the collection is the best seller. Like human taste is different and it's it's it's more now especially nowadays, it's more about trends. Like a trend will gonna come up one day and your collection will go completely a different direction. Okay. So it's it will never happen, in my opinion, that we can get to the point where there will be no outlets or everything will made will be sold out. It it it's really very

7:31 even for AI to predict this, it's it's almost impossible. It can make the invi the the the the items that's not sold out way less. Yes. ~ it can predict about these trends in future. way more, absolutely. And and this will take me back to the point is like what of data are you feeding your system? So what kind of output you're gonna get out of the system? When I was young, ~ my underwear and socks would always come with a piece

8:07 of paper that said inspected by number six or inspected by number seven. And I always wondered if there were just only ten people inspecting all the socks in the world. But I know that manufacturing defects, that's a really big part of the fashion industry, trying to minimize the number of defects and catch them before they get to the store. Is it possible for AI to assist or even to catch defects or track these things? better than a human can, or does it still

8:31 kind of require the human touch? ~ definitely it it still required the human touch, but what can AI do in this specific incident that the analysis of why you're getting those quality issues and how to avoid it and make it weigh much less in the future? Is it a fabric issue? So maybe you need to upgrade your ~ fabric inspection system? Is it a ~ labor issue so you need to train your people more often and it can also identify like which step exactly

9:05 in the manufacturing so you can focus in this area and improve it. But the the human output is is still needed like because like ~ you you you do the inspection basically using your hands and your eyes unless if you're using robots which is happening in some manufacturing right now but ~ Even with like fully automated industries, like we don't have in in fashion yet a fully automated ~ like production line. But in some other industries we have,

9:40 even of those one, ~ like ~ there's still quality issues happening. It's it's not like 100% perfect yet. But the anal the root cause analysis for why these problems happening and AI can give the best plan to make them way less. absolutely can can improve this point a lot. In my experience of AI quality control, sometimes I'll show two images and I'll say which one of these is better and it chooses wrong a large percentage of the time. And then I'll go, this one has a

10:11 misspelling, this one looks weird, it's the colors are wrong. And I'll go, you know what? You're right, I agree with you now. So I still think it's very far away from that ability. And if an AI is building something and then the same AI is checking it, there's that danger that it will like protect itself, right? We've already seen that it will take actions Mm-hmm. to defend its own actions, which means it's actually it's kind of like having the same inspector as the person as the

10:36 creation. There's a reason they're two separate roles. We don't have that kind of division, I think, that we're very far away from it. And I think you're exactly right because there's an element of art to fashion that s still there's a randomness there that you don't know. I never know what I'm gonna like until I see it or till I try it on. We have seen at the very front end where people can take a picture of themselves and then

11:01 put it themselves in different outfits to see how they look. And I know that's becoming more and more popular. And so we're seeing that part. But are we also are there places where AI can kind of take a larger role on the factory floor now and in the future? Do you think that's gonna start happening? improving the fitting is already happening. Like fitting was always a challenge for any ~ fashion manufacturer. Like ~ because like again, like even even from race to another,

11:30 the fitting is different. So the biggest challenge was always like, okay, now I'm selling in five different continents, and ~ so how I'm gonna make my sure that my fitting is matching everyone? ~ which is impossible. because we all build differently. But w what I saw that AI is doing recently that improving the fitting way, way better was like saying like, okay, and even ~ most of the brand names now, and and I don't know if you noticed this, if you're traveling per se to Europe,

12:03 like you're gonna see the same brand name in US and in Europe and they have two different collections. And it's be mainly because of the fitting. So now the AI is helping a lot with this, that's telling you like this fit will sell better in this ~ geographic areas and this this collection is better for that. And it's it's it's mainly based on the analysis the AI is running for like what is the percentage of people at that height,

12:29 that weight, and that age, and this is will be like again, like when you feed them with this col with this information that this is our targeted audience with this collection. And one thing I've noticed traveling around the world is that sizes don't mean the same thing. So an extra large in America and an extra large in Asia and X large in Europe are very different. So in America, Absolutely. I'll wear a medium, but in Japan I wear like a 4XL.

12:58 And it's like it's really disconcerting. It's such a jump. You're like, whoa. And Yeah. shopping for clothes mean, even buying a t shirt for me, I'll buy t shirts for four manufacturers that are the same size and they don't all fit the same. And I run into this all the time, especially because sometimes you get something that was manufactured for the region I'm in, and sometimes it was manufactured in America and it's leftovers and sent here. So it's the other sizing system.

13:23 It's also like wild to me when I buy shoes, they'll have like seven sizes inside. They'll go, this is the American size, this is the European size. Like, so you're constantly having to figure out which of the different numbers to use. So it's I'm always surprised that hasn't been universalized that we don't just all have one set, this is what shoe sizes mean. Do you think that, speaking of tr in international transport, supply chains have been dramatically affected since 2020?

13:50 We've seen a lot of changes, supply chains fell apart. Do you think that AI can help to fix that? I know that in America we really like to do that like last minute supply chain we have or just in time. We have just enough supplies for what we need tomorrow. We love to be, as I say, living on the edge. It's kind of like ~ if you only buy groceries for one day. So if there's any surprise, you're suddenly hungry.

14:11 But we really love to do that in the West. And we want to keep as little stuff in our warehouses as possible anymore. Do you think that AI can kind of help that? And I'm I kind of personally wonder, is that gonna actually increase our risk? Because as we've seen, you know, what happens if there's a the AI goes out for a day, right? Suddenly we don't have this backup system, we don't know what to do anymore, and we have like less we have less margin

14:33 to survive on. It's I I we can apply this to everything else. What if the internet is gone for the day? What if the ERP system is gone for the day? So this is why I I I'm a big believer that AI will never replace ~ human. and the best strategy is to have That you have the right leadership who can use AI as a tool. So when when we come to like a crucial point like the supply chain and and in Vintoni in particular,

15:10 I totally agree with you being in the US for 10 years now. I noticed that ~ customers are tend to order last minute. And and when you ask this question, like How come I was in a meeting with you last week and you were telling me you have enough inventory for a month? And a week after, you're like, I'm gonna run out of inventory tomorrow. So this is telling me that there's something broken in the system. And ~ as you said, since 2020,

15:39 we didn't have a normal year. And this is the main reason why the system now is broken in most with most retailers and manufacturers. In all the industries, no exception. Because how people used to do their forecasting and plan their supply chain, it's based on the historical data, basically. So, like in the last five years, this is what we have been doing. Our customers is growing five to seven percent every year. So we're gonna do our supply chain planning based on the 80-20 method,

16:08 and we're we're on the safe side. But what happened since 2020 that we didn't have one normal year to use it as a base. So like the pandemic and when ev and then after the pandemic, everyone like supply chain crisis happened when everyone was overstocking. And then the recession hit when everyone was trying to get rid of his inventory. And then we were thinking that we were finally gonna have a normal year when the Russian-Ukraine conflict happened. Which had a huge impact on supply chain ~

16:37 t ~ freight cost. And then after that, things start like settling down, and we were thinking, okay, now we're gonna have a normal year. Then the Middle East conflict happened. And when this started to ease up, we had the tariff war. So it's like All over the place. Like you cannot use, like looking at the last five years, you cannot use any historical data because it's all over the place. There is no consistency. So now what I have been using,

17:12 and and I and this is where I see the huge leverage of using the AI and supply chain. I I I used to say that if you really wanna make the most profit, if you are in manufacturing or retail business. You need to control your cogs, cost of cogs of goods sold. This is where 60-70% of your cost is there. And to do this, of course, procurement and supply chain is the key. And you need to have the crystal ball to predict where to buy,

17:45 when to buy, and what to buy. So, like in a best case scenario, most of the manufacturers will be like, we we want to keep our inventory at 45 days. But the question is, where is your materials coming from? What kind of weather condition can impact this? What kind of geopolitical conditions can impact this? And I spent hours doing those analyses to come up with the best plan where now AI can do this in five minutes. Once you feed AI with your customer forecast,

18:18 with where you your supplier list, where they are located. It will gonna build that plan for you. You need to buy this at that time. You need to buy this at that time. Be careful because winter is coming and Panama Canal will gonna be dry, and this will gonna raise ~ the supply chain cost, the freight cost was X amount, so maybe you wanna buy more of this product right now. And then do the calculations of cost of finance and how much cash flow.

18:46 Five minutes that used to take weeks of human to do. But this only works if you really have a solid sales and operation ~ process in place. If you don't have this, no one can help you, even AI. Yeah, it seems like you have to have that data and there's always a surprise. Like whether it's a weather surprise or a geopolitical surprise or something. And I live on a small island, so all it needs if there's bad weather,

19:25 there's no boats. So all it can take is these small surprises and things can really change. Do you think that we'll ever get to the point where AI can warn you about a shortage before it hits rather than after rather than just in time? Absolutely, but a hundred percent accuracy will never happen because as you said, there's a lot of things out of our control that no one can predict, even AI. But we can get easily to eighty percent,

19:51 which is huge when it comes to supply chain. So I've been thinking about this a lot and we kind of see AI as this panacea that can solve every problem, but it's also kind of there's so many options there that people don't really what's right fit for them. And so for if you were starting a new company today with no money, what's the first way you would use AI at the beginning of your journey? The question is first, what do I need the AI for?

20:24 And and and based on if if I have the answer for this question, then I can find the right tool. I mean, AI now can really like if if I'm I'm talking generic, I think definitely marketing will be the first thing that I can use AI for. It can save me a lot of money, with ~ setting the marketing strategy and even creating the ads and where I can go and what should I do. and and then after that,

20:57 the question would be like where I want to be. if it's manufacturing, then definitely I would use U AI for ~ inventory optimization, ~ forecasting and production planning per se. so it it like I I don't think there is like one tool or or one plan that can fit every single business. Everything is is very tailor-made when it comes to AI. It really depends on where you are, what you want to do and where you want to be.

21:36 Yeah, I just I feel like ~ the promise of AI is so different from the reality to it that it's still a tool. You still have to use it. It still requires time, it still requires implementation and the data you feed it affects the data that comes out of it. That for a lot of first time users or new users, they kind of go with the expectation that comes from the marketing from AI, which says push one button, it solves all your problems.

22:00 It feels like it's a magic button that then when they try to use it, It doesn't it's not as easy. It's easier than anything else, but not as easy as you were hoping. And I've kind of been thinking like, let's say it's a tiny company. What are things that t smaller companies can do with AI that used to require much larger teams? Definitely in supply chain area, like even even very small company that we're gonna need ~ a production planner, a material handler, a procurement,

22:32 a buyer, like no matter how is your site, you need this team. I think with the right AI tool, you can for a small company, you can handle the entire supply chain and procurement department with one person. 'Cause AI nowadays can definitely do if this person have the right mindset and the right leadership skills, ~ he can have five different employees working for him using the AI with planning and forecasting and optimization. Even procurement intelligence, ~ like I would use it for it.

23:08 running all the reports. Like so all these things that usually takes a team of five to six people you can do now with one person and the right AI tool. That's amazing. I just think that there's so much opportunity, but sometimes opportunity gets lost because the first time you try it, it doesn't do exactly what you think it will do, and then The other problem I encountered a lot is that there's so much FOMO that businesses and individuals as well, but even large

23:38 enterprises are buying an AI tool because it's the coolest one and the most popular one on the market. And they go, Okay, we bought this tool. What should we use it for? I have to justify the purchase. And I always say this is what happens when I go to the hardware store. If I buy a hammer or a tape measure, when I come home, something's getting hammered and something's getting measured. Cause I kind of want to show to my wife that I didn't waste money.

23:59 I bought, I did need this. To add this tool to my collection. So you kind of have this, and sometimes it's these massive expenses for these large companies where they buy something and then they're trying to find the right problem that this particular tool can solve. And I'm seeing this more and more. We've switched the buying cycle, shortened it so much because AI tools changed so quickly that we're not doing the same types of due diligence. Why do companies

24:27 kind of go in this direction, what's the right solution? How can they switch back to problem first purchasing? That's amazing because this is actually a case I have been working on in the last few months when I was working with one of the companies that they they wanted to do exactly what you did. You want to implement AI just because it's everyone is talking about AI and everyone is implementing AI. And my first question, what what what do you what do you need the AI for?

24:53 Like w what the the ultimate goal you want to achieve using the AI? And I think this is the first thing. ~ that any any CEO or leader need to think about what really do I need the AI for? And then they need to do a full gap analysis for their whole process. Like the before you even take the step forward to identify which AI tool you you need to first make a full assessment for your process on the operation side like

25:27 finance, sales, supply chain, and customer service. They are all linked. If you have any broken ~ point here, if you don't have a strong communication between all these departments, if you don't have a very strong sales and operation process set in place, doesn't matter what kind of tool you're gonna use, you're gonna waste your money and AI will be just like garbage in, garbage out. Basically, this is what will gonna happen. But when you, if you really want to use the AI,

25:57 that my advice will be always first thing you need to do full assessment for your process. You need to do gap analysis, you need to know where is the weak points, you need to know that you have a strong, accountable team, and after that you can start the implementing of the AI. If you did this, then AI will gonna take your good. process and make it exceptional. Really. But if the process is not there, nothing we're gonna work. Actually the AI

26:27 in this case we're gonna add more chaos to the chaos. Because like it's gonna get a make very fast decision based on wrong information, basically. So I've been thinking about this a lot. I think you're so right, which is AI is like an accelerant. And if you accelerate before you have a plan and you're just driving in a random direction, you're just gonna crash faster. And yeah, I love that you call it chaos because that's a really I think that's

26:57 better way of saying it than accelerant because it just leads to more confusion I find that it's the tool that leads to the most because you have the highest expectations and everyone says it works so great. And when it doesn't work for you, then you think it's you. And we're single, we've for a long time been seeing a lot of language large companies about, you can replace your entire staff with AI. And a lot of downsizing has been blamed on AI. And I always tell

27:21 my clients is if you have people who are coming into the office and working hard, do not replace them because it more and more people only want to if you can get someone to work from the office, that's now becoming like a s that alone is becoming a rare skill. Because we're seeing all these people go, they don't want to work in offices that are the nicest offices in the world. More and more this shift towards I want to work from home makes it harder

27:44 to get these employees. So if you lose them, you're gonna be hard to replace. And we're already seeing companies that have let people go, then regret it and back and forth, or they're blaming it on AI and really they just overhired a few years ago. ~ do you kind of think that the we're losing the human first approach and that we're gonna shift back to it and kind of realize that the people using the AI are just as important, if not more important, because you still

28:09 need someone to be accelerated, right? Like a company of all AIs is just almost imaginary. Yeah. Yeah, I I I I don't believe so because as we were just saying, like ~ AI cannot make the decision. AI can make the decision more visible. So you still need the human who the leader who can make the decision. Because ~ you you cannot trust the AI because you like again, one one wrong data entered the system can take the decision to completely different direction.

28:43 And you will not think, but the leader who knows exactly what's happening in his area and have the right mindset will be able to take the data from the AI and say, like, no, no, no, no, this is wrong. Like this was like something here doesn't make sense, or so and so. And at one side and and on the on the other side, ~ like Even even when I think about all those people saying that AI will gonna replace

29:10 and like who are gonna feed the system with the data that the AI needs to make the decisions? Who are gonna set like AI is not a strategy, AI is a tool. Who are gonna set the strategy? You need still the leader, the human, who can set the strategy, and then AI turns the strategy into an outcome. So I I I don't believe AI can. ever replace the human touch for any industry. For some jobs, yeah, maybe like for some

29:35 really big corporate or it's like entire team just to make reporting. Yeah, yeah, I can initiate those reports for you for sure. ~ for ~ forecasting and planning, yes. But for leadership, never. So I think this has been really helpful for a lot of people to kind of see how AI can fit into their businesses. I'd love for you to share a little bit about your people and not headcount philosophy and then how you help businesses and people are interested in hearing more

30:09 from you where they can find you online and find your business. Yeah, absolutely. So ~ right now I'm I'm doing ~ this through my my own company, Omnya Advisory, and they can reach me either through my website, www.omneaadvisory.com or through my LinkedIn. And I'm I'm always happy to to talk to startups and CEOs to help them how to make their business more profitable and to grow the business without the chaos. That's amazing. I'll put the links to everything in the show notes

30:48 and below the video. Thank you so much for being here today for another amazing episode of the Artificial Intelligence Podcast. It was my pleasure, thank you so much.