How H&M Embark Towards Scalable Digital (3D/AI) Fashion | H&M | PI Apparel

Thanks for the Ram. He was kind of confused if he should call me Bhargav or Ram. I was like, yeah, I go with Ram. And for a lot of people who have seen my face, yes, I did grow my hair quite a lot compared to my photo. I did get feedback from my brother that I should update my photo. It's been like 2 years I've kept that photo for like there. Yeah, my name is Ram, and the topic that I'll be kind of focusing today is around the area of how we embark towards digital scalable fashion. But actually, I was reflecting in the past couple of days about, is this the real topic what I'm talking about? But I felt like maybe it's more about how did we mess up? That's majorly where my focus is and how we were trying to fix within H&M the landscape. So this is my different topics, and the time I kept on the top so I don't forget the time slot. So the numbers that you see, don't get baffled by it. It is just that I have to keep my 30 minutes time slot. So I start with a hello, a quick introduction. Then I go into the topic of this is chaos. Then we go into let's tidy up a bit. Then we go into the filtering of new ideas and A for any curious Qs, hopefully for any questions towards the end that I'll be focusing on. So we're starting with a hello. Usually I like to do two introductions. And I apologize to my team who's sitting in the crowd because they have heard me doing the same introduction for too many times. But yeah, myself as an individual, I've started my career in 3D and tech and AI in architecture. Later on to work for Adidas for 3, almost 3 years, both in apparel and footwear. And as I say, a lot of familiar souls that I worked in my past experience. And what is my basic background is that I like to connect the world of business, financials, tech, 3D, AI, like all the worlds together to kind of, and always my favorite topic is scalability. I am not happy if 5 products work. I want to make sure that everything is scalable from end-to-end perspective. I do have backgrounds in 3D and hands-on, but now I'm more the Miro boards and PowerPoints guy working on Excel files. From H&M perspective, I'm a product manager leading a team of around 20+ individuals, plus we also work with a lot of external partners with a mix of 3D kind of people with a mix of talents of 3D, tech, UI/UX, business skills, and everything. Our responsibility is from the idea you make a sketch of design to all the way you want to publish it to different downstream consumers to activate it is our responsibility within the brand. Yeah, we are responsible for like, yeah, business processes, asset creation, libraries, tech pipelines, POC to rollouts. The whole DPC shebang is kind of where it's a headache of me and my team to look into the space. And lastly, we also manage all the DPC vendors that we work with, all the including financials and everything. So that's where we are kind of located within the brand of H&M. I like to always start with this slide. I— DPC is digital product creation, not 3D product creation, not AI product creation, not 2D product creation. It's digital. And we use maybe these tools to accelerate. But one of the things I like to say is, with the support of DPC, create fantastic physical products, at the same time leverage lifecycle of digital asset for different use cases. So this is something that we believe, and I like to remind that people often feel that when you talk about DPC, we talk about 3D. I'm like, no, we talk about digital product creation. So the chaos. One of the things I always like to talk about is it's always good to recognize a problem before you actually understand solution. And one of the things that we started doing was when I joined H&M 3 years back into the brand is with an experience of myself from the fashion industry as well as what we did within H&M, the first step we started doing was let's understand the chaos that we are sitting in. Before we can start changing the organizational setup of what we want to achieve within the company. So I'll be going through different topics of what is chaos, starting off with the reality check. Often one of the topics that I kind of drew to my heart and a lot of people that I work with is the area of DPC versus pre-production, which is actually factories. We always run in the hypothesis of sitting around in Europe or in US that this is how suppliers work and this is how actually factories work, but that's not the reality. The production lines are different. How we make a small tweak and a small change to a design could be much done easily on a 3D, but that actually influences the production lines that they have to completely change the way they create a product. So we make a lot of hypotheses sitting down here thinking this is how DPC can be easily, we can make changes, but we don't think about what effects actually it has on the suppliers on pre-production processes. Fit decisions. We always make decisions by saying that, oh, we can do a lot of fit decisions in 3D. You can't, not in the physical world. That's reality. If you don't have the right fabrics, if you don't have the right physical, like the particle distances in CLO or whatever, you cannot do it. It's a reality. But we maintain expectations that, oh, we can do actually like a physical decisions and digital decisions. No, digital decisions are different. They're not the same as physical decisions. Sustainability and other claims. With a lot of stuff in this area, with consumption efficiencies and all of it, we have been making a lot of claims sitting in our places about, how we can actually do this to improve the sustainability of products. But in reality, about what consumption developments happen in suppliers are a whole different ballgame. We make a lot of assumptions on this also, make a lot of claims. I'm not saying we in the sense we means brands, including H&M. We learned that in a hard way. We have mixed DPC standards. We started creating 3D standards. We started creating pattern standards. We started creating print standards. We started creating supplier standards. They don't mix altogether. Everybody is creating individually all these DPC standards and make— hopefully it will all work together. And my favorite, the 3D button. I'm still searching for it. Hopefully one day I'll find it. Is when we create an expectation within the brands that there is a magical 3D button that somebody can press. And because we have oversold the idea of 3D, that people think that we can do high fidelity without any kind of work that is put behind it. So that brings us to the main area is design versus pattern versus visual versus production are for me two different things. Making visual decision versus making production decision, making a design decision versus making a pattern decision are very two different aspects that we have to understand when you talk about DPC as per se. Design tech. I promised Yuri that I'm going to touch this area, so I'm going to go there. Interoperability. We are one of the largest fashion brands in the world, and we work with— we have issues of interoperability that design tools don't talk to each other. Different formats, different exports, different imports, different developments. How do we actually handle it? And then we have this amazing landscape of, as you see, both 3D is there, patterns are there. I'm not saying we're using all of them. Hopefully, luckily we are not using all the solutions. I would've been in more chaos than I would imagine. But this explains that this is a landscape. It's not just about Pattern N2D. There is Illustrator, there is Photoshop, and now there is Midjourney, DALL-E, Firefly, and MADNESS coming and hitting us. How does interoperability work across all of them? Most of these tools, they don't even have a lack of any APIs and SDKs. So how do we communicate with each other? How can we do automations on top of any of them? Business versus tech alignment and maturity. Business units have different expectations, whereas technology and IT teams have different things. I've seen organizations where two different teams in two different organizations, and we— business thinks that we can do amazing 3D, and the IT guys is saying that I don't even know what 3D is. And we expect everything to work fluently over there. We have disconnected processes, how they don't connect together. So connected to all the things that I showed. And lastly, we made an exception on how our conventional software setup are. We don't run software companies in our design tools the way we actually run other companies. We don't have proper requirements work. We don't have roadmaps. We don't have developments about when it's going to be activated, how it's going to be activated. We are kind of running in black box model here. The next area, narrow versus extended. Now we talk about design tool. Design tools are one specific thing, but then we have the entire landscape that we have on, on Hedeck on top of us. We have scattered libraries. Companies start talking about there is a 2D library, there is a 3D library for fabrics, go here. For products, go there. For prints, go here. How does it work? Imagine if you do it in the physical world. I go on saying that if you want to find a trim, go to this building across the table. If you find a fill of some couple of silhouettes and blocks, go down to that table. Our designers will hate it. And we have completely scattered libraries of what we are dealing with. Connected to most stuff I put on the bottom. And then we're kind of going to the era of golden cages. Every companies are trying to build up their own application landscapes. Everybody, every company is trying to say that we can do design, we can do PLM, we can do DAM, we can do 100 other things. But as a brand who works with home, footwear, accessories, so I should have a golden cage for apparel? Should I have a golden cage for home? Should I have a golden cage for footwear? How does it work? How do they talk to each other? It's chaos. We have chaos in terms of our maturity on downstream, upstream. Our downstream application doesn't even understand formats and makes a claim that, oh, give us an amazing FBX or an OBJ and we'll make it work. But maybe the design tool doesn't have the functionality. And how do you actually convert all these applications? Promises to reality. We make a lot of promises and claims that I've seen many companies making claims of saying that within the next 1 year, we're going to do this amazing 3D for you. And I've seen the roadmap after 1 year, didn't change. Still the same. Because we oversimplified the complications of a 3D in fashion. And then we are entering a new era with AI. Where we are entering the world of from under-content production to over-content production. I remember one of the companies reaching out to us a while back and saying that in 15 minutes we can create 1 million assets. I was like, what do you expect my designers to do? Like literally like go through a Tinder option of select right and left, right and left for 1 million assets? Doesn't work for us. So we are going to an era of over-productions with AI where we can have so much of content production that we don't know how to validate it. And we don't know how to actually figure out which one is valid for us. I do have to wear my financial hat. I'm a product manager and purchase orders are my kind of a thing right now. Now we combine all of this together. You'll have DPC design tools for 3D, 2D, AIs. We have all these libraries. We have all these experiences we need to build for reviewing annotations, decisions, or experiences of e-commerce or whatever. Then we have DPC custom integrations that we need to build between all these applications. Then we need to digitize, and then they'll change the formats and standards, and we need to re-digitize everything all over again. And what's the total cost for all of this? How much is it going to cost us for buy versus build? How much is the ROI? And plus the people cost, investment costs. And the best part of it, we often use the word as sample reduction, and it's like, does it actually cover the overall cost of the DPC that we're investing into this space? Are we being very objective in looking at what's best to be done here? So this is the chaos that we observed within, from H&M as an enterprise application around enterprise brand, like being one of the largest brands, how to actually solve this problem. I know, I love to see a lot of serious faces going into it, like this kind of a ball drop moment. I don't know what I did here, but I do want to have a light moment. This, I'm a big fan of Batman. So this is actually Heath Ledger doing skateboarding and Batman during the Batman scene. So just to have a bit of a break before I go to the next topic. So that people feel like some cool stuff is also, we can speak about once in a while. So how did we start tidying it up? One of the rules that we set within the brand is it starts with the what and the why before we can go into the topic of when, how, and the who. The brand DNA, we start setting towards that. The first step is we said, understand your brand. And this is where I'm putting from understanding of brand as H&M. As H&M, we started realizing that as a brand, we are a large-scale affordable fashion brand, meaning that we create a lot of products. Scale is a big topic for us. We have a large supplier base. We work with factories across Bangladesh, you know, China, and different locations. So it means we have to think about region, localizations, and developments. That's our chaos that we have to handle. Oh, sorry. Luckily, we don't have a lot of legacy tech. We didn't build a lot of libraries. We didn't build a lot of those things. So we had a kind of an open playground to play with. We had a large design team, so it basically allowed us to say that because you have a large design team, we had to understand how to onboard everybody. So it's not that we have to onboard like 20 users or 30 users, we had to onboard hundreds of users. So it also takes efforts and the time to do all the work behind it. Then we started understanding that how kind of an organization setup we are. From the last year organization, what we set up is we unified our business and tech team together. So as I mentioned previously, we don't have an IT and developer team separate team, and we don't have a 3D team as separate. We combine the team together who's focusing on improving our design processes. So we are not the ones designing it, but every team, everybody's under one team. So we can do development, we can do standardizations, we can do vendor management and everything under one space. That's an advantage we felt. And as a brand, we have our own stores portfolio. So we don't need to sell it to like Zalando or Foot Locker or something. So there's our own landscape. So our first step, what we started to understand is, who are we? Who are we as a brand? So that's my first step, is to understand before looking at outside, what kind of portfolio of stores do you have? Do you are selling out in e-commerce only? Is it virtual brands? Is it— are you selling to— because if you wanted to Zalando and Foot Locker and those things, you need to have sell-in meetings. The process is different compared to having your own store portfolio. Expectations are different, deliverables are different. So we first started understanding our brand of what we want to actually do. Then we started looking at objectively towards our tech. I'm a guy who's from 3D background. I love 3D, but I have— we have to look objectively of whether we need 3D and where do we need 3D. So we started thinking about funding on foundations was our experiences. Where do we need to focus this funding on to building libraries versus building a cool render pipeline? So we start figuring out where to put our major focus on and building towards it. Once we have that, we start looking at what are service layer applications. Service layer applications basically means they don't need a UI. They're just like something that is behind the screen, something is running. So if a company comes and says that we will give you UI, we say thank you but no thank you because you don't need your UI. We only need APIs to call you and work with you. We start consolidating our user-facing applications. We don't want our users to work with 20 or 30 different applications and expect them to do auto format conversions and everything. Integrations and pipelines and also modular adaptable architecture, we started building towards this. Once we go into that, we enter tip 3. It's just business. It's just sometimes we have to think of numbers. Otherwise, I've seen enough reorganizations. This is my third reorganization in my 6 years in fashion brands, and I don't want to see it all over again. So we set the goals about we have to look at costs, terms and conditions of these deals that they make. Often companies come and say, "This is a good idea. I'll give you a 3-year agreement." I'm like, seriously? Like, without even me piloting and poking, you expect me to sign a 3-year contract with you? Like, what's happening here? Incremental investments with parallel iteration. We started looking at like, let's not do any big shabangs. Let's start slow. Even if you do a simple library, it could actually reduce the amount of lead time for product creation. That itself is a great number. So incremental investments. We don't need to do one big investment to change the overall organizational setup of what we want to do. We start accepting that we are now in a volatile industry with the wars in Russia, with everything that's changing the world dynamics right now. It's not the same as before. And also with evolution of AI and everything, we have to be kind of smart in how we play this game. So we are looking at not putting all eggs in the basket, but smartly play about who should be doing what work. We started looking at where we should do in-house versus external versus supplier. There is a cost aspect if you do it in Europe versus if you do it in Asia, if it's to do it with suppliers. We started objectively looking at where it actually makes sense to doing it, not from cost perspective, but also from speed perspective. Doing it in Asia means it's closer to suppliers, it's lower lead time, lower carbon emissions, makes more sense. So we started looking objective towards where we want to work towards. And we start embracing that we are entering a new world of 3D and AI. We have to accept it. AI does also do cost reductions massively compared to anything that you have to do with digital fabric development. Through an AI fabric development will be a whole different ball game. From lead time point of view and everything. Tip 4, we start looking from mindset point of view. The rule that we set within H&M space we start putting is called give something to take something. We cannot go to the design community and say, start creating high level of standards, but I'm going to give you scattered libraries. Doesn't work. We cannot give them overarching workload of doing 20 more tasks without actually taking away 5 more tasks from them. So we want to give something before we can say improve your standards, improve your developments, improve your timelines. Partnerships and collaborations. I've met many of you guys today and I've been like discussing about different topics about how we can collaborate more. We started opening the boundaries and I can say this is one of the first times in DPC space in the past 6, 7 years H&M has done that H&M is on the stage here talking about it. And the reason being is we want to be more open. We want to discuss, share what we have, discuss about what you guys are up to, so that we can actually look at it from a close collaboration point of view. We don't want to do it all by ourselves because I think it's— we can all as brands come together and build a better future over there. Be curious and onboarding leadership on reality. As I mentioned in the ones before, fit decisions and all of it. One of the things as we started doing was we started making our leadership sit down and say, this you cannot do in 3D. This is the reality. Accept the reality. Instead of— from the time of overselling it, we started now underselling it, making the brand realize that these are limitations in 3D. These are limitations in EPC. You cannot do these things. It's OK. It's still better than having to do a physical sketch and putting it in— sending it to suppliers or something. It's still better. Accept the reality. And one thing that we have started working also is on core versus side value that drives us. I remember when I came to H&M, I asked a question about why are we doing DPC? And often the question was like, they gave me a list of like 20 topics of why DPC can do, like it can reduce samples, we can do e-commerce, we can do 100 other things. I'm like, I'm not asking the value of DPC, I'm asking us why in H&M we're doing DPC. Because if you want to do DPC for production, your tech stacks are different. If you want to do DPC for visualization metaverse, your tech stacks are different, your processes are different. It's important to understand what's your core space of value relation that you want to work on, and rest of it could be side benefits. Maybe while doing this core part, you have other side factors that can help you to accelerate within the company. Then we start looking from— look towards brand's ecosystem. So when we start talking to companies like the ones that we work with, design tools, we stop thinking about what— how does it— how does this specific requirement or feature help the brand? But how does it help brands? Because that also means that there's a better chance that these design software companies will actually implement the solution because it doesn't help H&M, but it also helps different other brands because it's the same problem all of are facing within the plan landscape. So you start seeing the mindset of opening the doors as well as cross-collaborating with different other companies. It's easy to set a strategy. It's very easy to put a plan, put in execution methods and everything. But it gets harder when you start executing it because it's like the sand can go through your fingers. We can start going back to old habits and start losing our sense into it. So for that part, we started setting the fact about filtering ideas. How do we filter new ideas and work towards it? In order to do that, one of my, like David Warren, who actually reports to the CEO, he asked me this question like, in order for us to be successful, what do I need? And I asked him, "Sir, I need you to be disciplined. And I need you to not come to me with new ideas every day, but ask about where the idea fits into it." If the manager is not disciplined, it's very hard for us to say no to new ideas. And then we started creating a balance between what is cool versus what is boring. Building a library is boring. Doing high fidelity, amazing render shots is cool. But sometimes you have to create a balance on where to focus on. Exploring versus exploiting. Sometimes we'd exploit the knowledge that we have instead of exploring new venture revenue streams that we have, something that we have not understood about, creating the right balance between those two ecosystems. So for that, we started building a concept architecture. It's a very simple thing, nothing fancy. We said, let's start with digital asset producers. This could be different things. This could be you want to create a fabric, you want to create a product, you want to create a pattern. I don't know, different asset producers. We want to put them all into a single digital asset management solution. One library, not 2D, not 3D, one library. We want to connect it to our lifecycle management tool. So PLM applications connected towards it, make it happen. Underneath it, we'll have different services, maybe 3D rendering, maybe fabric rendering, maybe AI-generated fabrics. These are different services that we will utilize accelerate our process. Top of it, we will build collective experience. So the things that leverage the asset producers, design asset management lifecycle, and there's a different collective experience. It could be you want to make an annotation on a digital product, or you want to send it to e-commerce, you want to make different decisions. How do we leverage these different experiences and have a collective experience on top of it? And then we'll have different upstream and downstream spaces. The different downstream, upstream spaces is different applications because big brands, we do have different players in the company itself working towards who are the ones we have to get information from, who are the informa— places we have to send information to, because not always we are the ones sending information to. Maybe you need a composition score of a fabric done by supplier that comes from the physical world first. Maybe that's an upstream that we need to connect towards. So once we set up this one, it became very clear whenever a new company or a new idea came where it starts fitting into this landscape. Second, we introduce business capability model. Business capability model made us to show that anything that we introduce, we talk about people, process, data, tech. Any new idea, any implementation should be able to address all these four dimensions. Without that, we cannot go ahead. Then we started building different chapters. So we first said chapter one could be the foundational chapter. So we start with the foundational chapter now of what we want to do. Chapter 2 is where, once we have the foundations, we'll build integrated experiences where we'll start connecting all the foundation of integration, application landscape, build them together. Then Chapter 3, we'll have a curated experience. Now we have this amazing integrations, pipelines, everything working, right? How can you curate something new on top of all of this? How can you leverage AI, generative AI, or whatever? The big data topic in the morning in keynote, these things become a leveraging aspect in terms of how we want to leverage. Chapter 4, who knows? I don't even know what happens at that time. Maybe there's a new geopolitical chaos will come that we don't know about. But we set different chapters. Then we set a heavy business value framework. We looked at our brand plan of H&M. We looked at what are different chapters that we have. We looked at different outcomes that we need to do in order to reach our chapters. We looked at different KPIs that we need to adhere to those chapters. We looked at different features that we need to implement to make those key results happen. That made it clear for us to understand where to focus on and where's the focus energy we need to put ourselves from towards. That we used it for filtering and allocation. Whenever a new idea comes, we put into governance model. A governance model is where we set the rules. How does it fit into a model architecture? Do we have an understanding of business capability? What chapter is this connected to? Do we have foundations to scale it? What is the timeframe and size of this business validation? So we set real boundary rules about how to validate this DPC idea that is coming into the space. And then we start allocating it to different streams. Is it an improvement to already what we've released? Is it completely new thing that we want to do? Is it a here and now kind of a cool thing that we just do for like a very short period of time that can work towards it? or is an explorational phase that we know eventually it could come in maybe a year or two that we need to invest towards and to see what happens. This helped us to understand which bucket are we putting and what is an expected results out of this bucket. Improvements are new and here and now has an immediate effect. Exploration doesn't have an immediate effect. Maybe it'll have an immediate effect after 6 months or 1 year, but we can kind of keep ourselves up to speed on the market analysis and what we want to do. This gave us insights about how we want to develop towards. So to conclude, from all this, we created a vision for the next years to come on how we plan to embark on building a scalable digital fashion. And the last part, not much is there because I do want to keep time. I'm looking at the time there. I'm actually on time with 5 minutes there for A for KineQs. I want to make— basically, I put some couple of faces here. We have JD in the room with me here. We have Moa in the room with me here, but we also other people. So I have a team of people with different leads focusing on different areas. So myself as a product manager, JD handles everything on digital libraries and developments about how we're going to do. Fritz, who couldn't be here, handles everything on tech mesh and pipelines. Madeline, who couldn't be here, handles anything on products and pattern design. Cora handles everything on asset digitalization. And Moai is also in the crowd here, handles everything on avatars and bodies and everything. And as I put over there, for all intentional reasons, I kept it on a high level. We are presently implementing Chapter 1 and 2. Maybe the next year or the years to come, we will show more about it. Because we're in implementation, I cannot talk to you about what exact solution we're dividing. We don't have high-level plan. We have a detailed plan, which solutions which providers and everything, but we cannot disclose it till we actually implement it. So for all intentional reasons, I didn't disclose that, but hopefully in the years to come, we will also share that openly within, in this kind of forums. Thank you.

Okay, so go ahead and scan the QR code and enter your questions. If anyone has any questions in the audience, I can also give you just the mic. Any questions for Ram that anyone wants to ask directly? So actually, while we're waiting for the questions to come up on the screen, Rama, I have a question for you. So what— and maybe you can't disclose this— what specific benefits, if any, were you looking to achieve with your digitization efforts? Like, maybe even in terms of priority, like 1, 2, and 3.

I mean, that was the battle that we had when I started in H&M. But right now, our focus is on half-time to market and production is our primary focus. Visual experience and everything we see as side benefits. Our core area is going to be how we can activate suppliers and make product decisions much faster with a shorter lead time. And make more relevant decisions towards it. So that's our primary focus towards why we want to work on DPC, at least for the next couple of years. Okay.

And has that— how would you contrast what your process looked like before you did your digital implementation to what decision-making and your speed of product development looks like now?

I mean, what's one thing that we did in H&M is we have developed a process very fast. So our 3D adaptation is very high. At least that headache I don't have, that I need to onboard the users saying that believe in 3D. They're already there. But when it comes to actual work on it, the issue that we had is it was open for interpretation of why individual people had their own interpretation of why the value of 3D is. But that is basically now what we're working on is to kind of bring us into one single idea of a belief. But the level at which we did 3D implementation, I have— it's been a while I've seen any other brands actually at that acceleration made it. I would say like almost like 70 to 80% of our user base is already working in digital product creation, which includes like pattern designs and 3D designs and everything. Yeah.

Great. We have time for just one or two questions. The first one is, how is the DPC between in-house and suppliers? So what's your mix there?

When we started a few years back, the focus was heavily on in-house, but then we have been kind of shifting our focus towards both external partners because I do believe there is a third one. For me, there is in-house development, there is digitalization partners who are there in the industry who we can outsource it to, and then the third is actually suppliers. So for me, there are three levels there. So we majorly right now work more from in-house as well as external agencies that we work with rather than within our own place. But actually, I was just also in Dhaka last week and we are having discussions about how we could take it from supplier point of view.

And when you say suppliers, does that include both your factories and your material vendors or suppliers?

That's what we are looking into it because we do have a huge supply database, so So the idea is to activate them, but we are still discussing on who comes into the first priority list, who should be focused on. The advantage of tier 2 could be that fabric mills directly there, but it means we are over-digitizing. But if you do with tier 1, maybe the advantage is that whatever digitize, we will use more often. So there are these pros and cons in terms of what we're looking there in terms of who should be activating from an overarching strategy point of view.

Fantastic. I think, yes, we are out of time. Just to say, though, before we go, one of the questions— one of the questions was that there was no question, but there was a great amount of thanks for you being so open and honest with presenting the chaos. And I think, to me, I find that very valuable in terms of being able to see inside the process at a big brand and what they actually have to do to go through to implement these digital changes. So thank you very much for everything.

Great.

Thank you.

Thank you.