Computational Design X Footwear's Digital Evolution

Hello everyone and welcome to another PIE Spotlight, the penultimate PIE event of the year. Thank you very, very much for joining us today. We're diving into one of the most exciting frontiers in footwear innovation, and that is computational design. As parametric modeling, algorithmic structures, and generative systems gain traction, the question isn't whether computation can transform footwear, but how it actually fits within the realities of design culture, 3D pipelines, manufacturing constraints, and the fast-evolving influence of AI. Because computational design isn't just about automating tasks or producing wild concepts. At its best, it turns sensation into structure, data into decisions, and constraints into creativity, giving designers new ways to think, iterate, and build. And in today's conversation, we'll explore how brands and creators are embedding computational thinking into product creation. We are joined by a brilliant group of speakers who live and breathe this space, and I'll hand over to them in just a moment. But first, a huge thank you, of course, to them, but also to all of you for tuning in today. We are really, really thrilled to have you here.

Here.

A couple of housekeeping rules: if you have quick questions, feel free to pop them in the chat box on your screen. These will be sent directly to our panel chair, who'll do his best to get to all of them. And please don't be shy— this is meant to be as interactive as we can make it, so please feel free to put in any questions or comments that you might have. And if you miss anything or want to revisit the discussion later, this session will be available on demand via Seamless and the Pi Apparel YouTube channel very, very soon. Lastly, I want to flag that today's session is a teaser of what's to come at our 2026 Stride events. Happy to say we'll be returning to Portland in March and we'll be launching a brand new European iteration in Italy in May. So if today sparks some ideas, you'll definitely want to join us there. So without further ado, Onur, my friend, over to you.

Thank you very much, Michael, and thank you very much, PI, for having us today to run this panel. And the subject is pretty interesting, computational design and footwear's digital evolution. And I think what makes this meeting special is we're going to talk about the uncomputable parts of things while we talk also technically about computational design. I'm going to briefly introduce myself. My name is Onur Yücegün. I've been a leading profile in computational design for over 20 years. And I'm currently the director of computational design at New Balance. And I'll pass it to Maya to introduce herself.

Hello everyone, my name is Maya Zelaskova and I am a computational designer in the innovation concept design team at ON, based in Zurich, Switzerland.

I think I'm next. Yes. Hello everyone, I am René Amedeo, sitting here in Germany in Pirmasens. Shoemaking City, working for Framas in innovation. So it's about digital creation. So let's go with Jesus.

Hello everyone, my name is Jesus Marini Parisi. I'm the founder of Moon Ravage Lab. We are a design studio based in Milan that focuses on computational design for performance products.

Hello, everybody. My name is Erik Oeliche and I currently head up the innovation concept design team at On. Prior to this role, I've held a couple of different positions within innovation and performance design at On and prior also at other footwear companies. And yeah, really excited to be talking with this panel here today. So over to you, Onur.

Wonderful. Yeah, this is super exciting. And again, as I just said at the beginning, it's going to be pretty comprehensive. But let's start with the roots. Let's start with the historic, let's say, development because footwear conventionally has been a material-first, let's say, realm, right? So everybody thinks about materials, putting materials together, and so on. Whereas digital world implies other things, right? So you don't touch everything. You don't kind of like get used to kind of putting things together, let's say physically. So footwear design has historically been a material-first discipline. So what does it really mean to embed computational thinking into the process? Do you people think, are we just digitizing the old workflows or is our thinking also evolving as we use digital tools? So I'd like to actually start with Maya on this one.

Thank you, Honor. I think it's really important the way that you are framing your question, or, and the difference between simply digitizing work, existing workflows, versus adopting truly computational processes. This distinction actually mirrors what we've seen already in architecture, where there was a shift from computer-aided tools to computational design, and And this changed in a fundamental way of how buildings are made or how, but most importantly, or how problems are framed and approached. So I see this a very similar transition happening at the moment in footwear. I think it's fit for me to start talking or to mention LightSpray here, which is a technology pioneered by On, the company I work for. Which only because I think it's a very clear example of applying such a, such a computational thinking and workflow into this industry. LightSpray is not strictly material-driven technology, but it derives a material system because there is an interaction of computation, robotics, and material engineering. And in this specific context, the computation acts as a connective tissue, which means that the performance intentions, the automation, and the functional behaviors are all embedded into the making of the upper. Lightspray is an upper technology. Maybe I should have mentioned that. However, because of this process, the aesthetic expression emerges because of this structural and manufacturing intelligence. I can talk about my personal experience at ON and in the innovation team. It shows that computational strategies are actually used a lot within the innovation portfolio and within the processes. I also see it around in other brands. I think that often the issue is that these remain quite hidden or they are occasionally without a shared understanding of of their value, or they are just misunderstood. So how these strategies are applied or they work. And I think that there is a very important opportunity for computational design to consolidate these efforts and to help shape more integrated innovation culture. And yeah, I think fundamentally, maybe this is something we, I think we all of us try to embed in our work, but When we use computational, how we use computational design, at least for me, it helps us shape footwear creation and see footwear creation not as a final object that needs to be derived, but more as a set of relationship that needs, that those are the things that we design for. And yeah, I think computational design allows us to create this space around. Sorry, I think I took a lot of time, but—

No, no, it's great. I think what you said was like, one thing you said was very crucial. I know that like connecting connective tissue. Because which brings us back to maybe it's not something, it's not necessarily a tool that you use, but it's rather a way of thinking, right? So to connect. And I know Jesus has a lot to say about this.

Yes, actually it was a good point. Actually, I wanted to like connect to that point of thinking about more like a systems, no? So basically footwear, as Onur was saying, is like a material-based— I mean, you see the progress of footwear with another main type of products that is different in the panel. So when you start seeing like computational design and somehow it was existing there since many years ago, so, and it was adapted by another type of computing industry, maybe automotive, aerospace, and so on, because it got little by little complicated. I mean, also footwear got complicated in this, in that same time, you know. So you have an upper, you have a midsole, you have different type of foam, then you have different type manufacturing process, you've got carbon fiber plates, so it immediately can get really complex. So the first time that I actually approached Footwear, it was like, okay, this should be approached more like an automotive problem rather than a, I don't know, like an industrial design product, like a classic one. So it was quite interesting to see that progress in in the way of, okay, how can I frame now the process from a completely different perspective? So that is when the system started like came, you know, like to see how to connect the dots between, I don't know, simulations, CAD modeling, using data, creating aesthetics, creating a system that actually makes sense and resonates with the consumers. You need to be close to the athlete, close to the user, So actually, I believe that that's the nice part of computational design, that you need to think about an ecosystem like a complete, let's say, like a living thing that starts connecting different dots together. And then those dots are actually giving feedback to another. So that's why, for me, it started making sense, and I started pushing more and more and more in this section. That's why I think somehow there were difficulties also in how to translate this always into the new technologies, because there have been things that have been done in the same way maybe for 15, 10 years, 20 years that you know that it works. But at the same time, everything is evolving, no? So we are changing the way of seeing things, maybe the logistics are changing. The supply chains are changing, so everything starts like making a little drift. So this is when digital environments, and especially with computational design through data, you can start like connecting those dots. So I think that's— this is a good moment to start like, um, like say thriving in this area. That's my takeaway from this question.

Pretty good one. And I also know René has some, some things to say about this as well.

Yes, of course. It's an interesting thing, interesting topic. I think we are not digitizing the old problems, we're more finding solutions through computational design. But I always think that we need to take all this heritage of shoemaking, traditional shoemaking. We produce here at Framas, we produce a lot of parts or components to make some shoes, mostly high performance. So it's a very specific and very accurate way of dealing with the brands. We as producers, we have to make all this system for production. But when you don't consider all the tradition in having a hammer in your hand and nailing a shoe or stretching a material over the last, you cannot understand what's happening with the footwear industry, since we consider the industry as the real endpoint of what we're doing. Is ending only there, but the real user is who will be benefit of our computational thinking. But therefore, that's why all the attempts to find new ways or these new challenges, maybe they are all workflows, but the solutions are new because now we have new tools. Automation is here, robots are here, and additive manufacturing is something we are hopefully having now than we— before it was something not even possible. So now using all these technologies, we are able to really find solutions through computational design.

That's a good one. And actually, you know, as we are talking about the potentials of computational design, which appear— like, what appears is that it's not only the tools emerging, but also tools, both like software and hardware and thinking, is helping us evolve the process. The process is changing, and with that comes the limitations, right? So, which is, you know, it's a great way to maybe, uh, to my next question, which is about the challenges, right? So if computational design is so great, why it's taking so much time for adoption, right? So what kind of like frictions are you experiencing where you work or in, you know, like where outside as you, you know, talk about computational design and what potentials you see this kind of like transformation and adoption? So Jesus, this time I'm going to start with you. Actually, I will—

I would like to frame this question maybe from a personal point of view, because I think computational design for me was more like an endpoint. But this endpoint for me was like connecting different dots through my career. So I studied mechatronics engineering, I work in automotive, I did design, a product design, then I started working with interaction design, digital fashion, then I did research about parametric modeling and so on. So all of these parts connected today in what's computational design. Why I'm saying this? Because I realized that it's complicated in terms of how many areas of expertise you need to know sometimes. But there is not so many programs yet outside there that you can actually go and study. I remember when I started like maybe even getting into footwear, like looking in YouTube 5 years ago about footwear, there was nothing, no? Or things like that. While you were looking into Grasshopper and things like that, it was like immediately a lot of tutorials and so on. So I think today the main challenge is that how can you create like this new generation of computational designers, but the ones who need to be also informing or teaching, they need to know a little bit of different areas of expertise. So, um, and also see that from, from the internal perspective in my studio, because we need to be multi— we need to have like different disciplines inside. So we have fashion designers, industrial designers, engineers. So in this way, we can actually like connect correctly to the problem. So for me, it's more like, how can we frame it now that we know that there are specifics, or they need to connect different areas of expertise. How can you bring this into, into the teaching, into the schools, into the new generation? So I think that's the bridge that is missing.

Yeah, that's great. Eric, I know we have been keeping you quiet for a while. So adoption problems and potentials, what do you think?

You're mute.

Yeah, Eric, you might be on mute.

This is a really interesting question because it's very near and dear to, I think, what we, what we are currently living within ON as well. I think there's a few different aspects to it. There is obviously a cultural and educational side that Jesus was also mentioning. Maybe the angle that I would like to highlight here is how our design tools and how we design is very closely tied to the manufacturing processes that we use in our industry. And if you look back at, like, how we manufacture shoes today, that has been a pretty static, static or slow evolution, I would say. And I think that has indirectly determined the types of tools, processes, the creative thinking, and also, let's say, the accessibility for computational design to enter that world and has maybe limited this at certain points in time. And we were talking a little bit about this prior, but what I now see when we introduce technologies like Light Spray, all of a sudden the necessity is very clearly there, right? Like, we could not develop a technology like this without the power of computational design or without having Maya being able to think really end-to-end, like both design, manufacturing, material characteristics, all in one package. So I would say if we can unlock both of these things, computational design and manufacturing hand in hand, I think that's where we can really accelerate the adoption of computational design, or at least it's, it seems to be very successful. And I think, Jesus, you also have done some projects recently with, with Puma where I think you're, you're starting to show like the very direct impact that it can have on product development. And we need these success stories, right, for us to be able to talk about, hey, this is the impact of computational design and like how we can expand this. Right now, it's having a very clear impact on the types of projects and technologies we're developing. But the potential is clearly so much larger. And we need these stepping stones to to really build it out, I would say. I lost—

sorry, René, actually, this time I made it. So, René, I'm wondering about your take on adoption of computational design.

Well, the adoption is something that is happening now. It's, as you mentioned before, the generative design. It's something which is the first door we need to cross to enter into this world. But the challenges now are adopted because of now we can produce footwear using technologies like 3D printing. So it's more democratic or accessible and affordable. Now people can even print their own shoes if you know how to make shoes. That's the important thing about footwear design that is not really related to computational design. We can do so many crazy things. And lattices and minimal surfaces in a very more simple way because now we're using the computer to help us. But if you're not clear how you should design a shoe, obviously this will not be a clear adoption. For us, it's in production more than printing the full shoe is about how to produce more parts. So it's also about mold making. And in the mold making, you know, it's a very accurate and the high precision is a must. So that's why computational design is providing the tools to speed up the process. Sometimes it's a bit technical, a bit boring when the product comes to the mold itself, because then the mold is really hard to break. So it's something that you need to use as a tool for making a lot of shoes in production. So I think that's a way for the adoption.

Yeah, this is great. You also said something very crucial, you know, uh, You said if you know how to make shoes, right? So, and this actually, this is a comment that, that applies across the disciplines, right? Outside footwear to any sort of design, any sort of making, manufacturing, you have to know what you're doing. So we are at the, I think, edge of a dilemma, right? So we're talking about automation systems starting with AI, uh, be it like agentic AI in which agents come together and create things, or be be about like physical AI or robotics in which we see an automated future, right? So this is my next question, and I'm gonna actually keep going with you, you, Renee, if you don't mind. So what if we like— if we can automate everything tomorrow, who knows how to make shoes? And what happens with the next generation of like making shoes, right? So could— what are your thoughts about this problem of automation and with that, the authorship? Like owning designs and making designs. What do you think about that?

I think this is the moment that we can do it. Young generations, of course, maybe they are losing interest in traditional handicraft. The footwear industry in general is something moving more into massification, but the Customization is also having not only a niche. Now it's able to— now we are able to create shoes in a personalized way. So scanning a foot is something that now you can do with your smartphone. And directly from there, you can go into the printing stage if you want to do it like that. But on the industrialization, let's say, it's also a good moment because automation now is enabling a process where, well, sadly less people are present in the factories, so it's more automated. 3D printing is also helping to speed up the the production if we can 3D print the shoes. But also some mold makings thinking more in the futuristic way of if a mold can really be adaptive, for example. I know it sounds not so realistic, but last week it was Formnext here in Frankfurt, and you can see there are some ways of making molds that you can program, they can change their shapes, uh, pointing into mass customization. Um, so customizing some production based on what is happening in the market with the real consumer, I think that's the way that we should think about. The times that shoes or collections were made for forecasting hundreds or millions of pairs, maybe they are not adapted to the actual market, which is moving much more faster. Likely there are more brands and there are more people making shoes There was a boom in the past decades, then there was an industrialization moving to Asia, but now I think it's coming back, coming back to the production to Europe, to America. So it's more feasible to produce footwear now. Not only in Asia, which is a bit contradictory for us. We make most of our production in Asia, of course, that's for sure. But I think it will be happening in different locations. And at the end, this new ways of producing the shoes will not be centralized or based on the traditional industry. There's no factories needed, like traditional big factories only for making shoes. It could be something that using the digital tools and the hardware and software, we can democratize this footwear production.

Yeah, that's great. And just talking about like robotics and automation, Jesus, you worked for like automotive, in the automotive industry too, where robot arms, you know, rock and roll. So what is your take on automation for footwear design?

I think this is, this is quite interesting because, um, in somehow automation takes the hard part in somewhere in, in order to give you a little bit of extra time. So this extra time can be used for something else. So automation is not like to to come and replace and is to help us in somehow. So it will help to enhance what we are doing, what we are doing and who we are framing things. So I think this is, this is a process that we need to adapt and then we need to see and we need to understand in order to gain some extra time to maybe design more, to create a new machine, to create a new workflow. So basically we are solving something today thanks to this technology that in the future will create probably another perspective, another point. So we will go into the, into the, into the next level of optimization and probably product development.

Yeah, great points. And, you know, as we go into this subject of computation and its branches, you, you know, be it AI or robotics or automation, I think one interesting thing that I've been talking about is simulations. This is coming back from my architecture years too, just to touch up on that Maya mentioned that 20 years ago, you know, just working on these huge skyscrapers and trying to calculate the solar insulation, right? So we would get enormous amount of data, but then the real lifespan of the buildings, that was always a challenge to match the numbers, right? So the reality and simulations are very different. But as we work computational and as we run simulations, we lose that, you know, let's say acknowledgement or understanding. Sometimes simulations become so real, we got like super invested in it, and then reality may or may not match that. So I'm pretty curious about your take on, uh, on simulations, where they stand, how much we should trust in them. Uh, Jesus, let's keep going with you because you're, you know, talking about it a little bit already.

Yes, I think, I mean, simulations for us, it was the base of what we are creating Yelp also. So when Eric highlighted the latest collaboration, it was really good because it was through computational design and through simulations that actually we got to get a really good product, a really good example on the market. But this is also to take a thought because it's not only that you put something into an FEA simulator and then you have colors and that's it, no. You need to think through, you need to be creative, you need to design, you need to say like, is not— sadly, today is not democratized technology. And also the data that is behind is hard to get, hard to label, hard to process it. So it's a little bit of a certain level of problematic. So when you get data collection, which is the main problem, then you need to understand that if that data is correct at level, is collected, filtered, is in the correct way, is in the correct time frame. So there's so many layers of thought that you need to put, and then you, you need to understand, okay, now I have my data. Then we go into the material. So the material is another layer completely. Is the foam correctly? The stiffness is correct, viscoelasticity, and so on, because we are working with soft materials. Then we need to add a layer of carbon. Then you need to start thinking about the carbon, and so on. Then you can start thinking about simulation. So It's a process that once that you understood all this workflow, actually you can start like seeing the result that gets accurate because you did from the beginning a way to get close to being accurate enough in order to get an okay answer. So as we were saying, I always say like in automotive, you get maybe 80-85% of accuracy because sometimes the material characterizations are really, really well done. But in footwear, we have an extra layer because we are not working in the same kind of range of materials. So probably also some of you are experiencing the same questions, you know, like what is the hyperelasticity, what is the viscosity, what is the reaction over time, and so on. So this, this is when simulation can be really helpful because it could help to predict, but at the same time could be really, let's say, like a deviator or of where are you going because it's not the FEA software or whatever software is going to give you a result, but it's up to you to understand if that result is correct or not. So it's not about like happy colors or like rainbow colors. It's also to understand what that means. And I think that's a big take for me.

Yeah, I love that. And I'll get back to these colors actually. I want everybody to note this, but I don't want to take too much time myself because I'm already asking, taking time with questions. Maya, I talked about architecture, colors, you know, simulations. What's your take on this?

Yes, I mean, I'm not really sure how accurately we can simulate these parameters in footwear. What I hear from Jesus, it's actually we are pretty close, or we could be really close. And I can focus on the second part of the question, which was related to how do we negotiate the subjective and objective parts. And In the innovation team here, we are very lucky to work with people who do not come from the footwear world, okay, myself included in that. And many of the collaborators that I work with are computational engineers with backgrounds in material modeling, composites, numerical optimization, mathematical modeling. And working with them made me realize that there is something that is very clear, and it is this very delicate balance between the purely scientific simulation approach, you know, it's very clearly and objectively defined. And then the reality and the outcome of these simulations that are tied to the athletes and consumers whose experiences are very subjective. And yeah, there is a lot of this tension. And I think this is what I see here as the biggest challenge. I think what I observed working in this way is that the— me or a computational designer is— the role of the computational designer would feel like an electrical transformer sometimes because it, like, in my experience, I would translate the subjective needs, which are the feedback from the athletes or the intuition from an experienced footwear designer that does not work in 3D, for example, or some insight from our empirical observations on prototypes into very objective parameters that can live inside a mathematical model or in a machine learning pipeline. And it's interesting because this translation also works in, in the opposite way, taking a numerical result out of a simulation and being able to turn it into something meaningful in terms of design or in terms of performance. It's also, I think, where the you know, where the interesting part is and where the challenge stays. I think what makes this process even more complex is that we are trying to build this iterative loop, right, between simulation, physical prototypes, and human experience. And I think we all approach it very straightforward, like simulate, then prototype, then test, and then refine, right? This is always the the workflow that is followed. But in reality, each of these steps is— it speaks a very different— I mean, not a very different language, but sometimes slightly different language, which is enough to create a lot of challenges or operates in a different timescale or has its own biases. And to be able to correctly align these mathematical assumptions with the material behavior or robotic constraints and human feedback, which none of which fits into a, into a single model. It's, um, it's the challenge. And I think this is what we are focusing on, to understand what is the link between all these things and how they can tie together. And we don't try to limit ourselves and try to fit everything into the same framework, but, uh, probably highlight or take the things that, that, that are more meaningful and, uh, more important into these workflows and, and focus on them. But I would say, yeah, it's a simulate— simulation and human experience in our industry, it's, it's not straightforward. And, uh, but I think very interesting.

Yeah, I hear you. And what I hear from both of you is that, uh, like reality is so complex and simulations need some sort of like abstraction and reduction. So you have to reduce reality into smaller models so you can run simulations. And that may, it's potentially missing, you know, a lot of bits from reality. That's what I'm hearing, which is exactly about the next thing I wanted to ask. You know, it's kind of cool that, you know, you talk about simulations and you bring the discussion to data, which was the numbers and colors that Jesus mentioned. And you, Maya, you talked about like modeling. So Eric, actually I wanna kind of continue with you on this subject of data, right? So I can talk about data forever, but I won't, right? So just to let you talk, Eric. So the reality is converted again, as we discussed, into numbers and colors and so on. And then this may include information about materials. Some people ask about it, right?

Right?

So can you simulate or use data of the real materials from the real world and get the results in the, you know, in your designs? But it doesn't end there, right? So we can talk about how people move in space, which is another sort of data. And also just to expand it, we can also try to quantify the sensation or the feelings of people. You can do that to athletes, right? Into data and so on. So when we say data, it's a pretty broad subject. So what do you think about this kind of like translation of, reality into numbers and colors and how that really works in your day-in-day-out workflows.

Yeah, I think this is a really great question. Um, let me, let me approach it like this. So I think that there's definite, um, in, in that translation, there's definitely things that, you know, might go missing or might get altered, um, and that might influence, uh, the type of decision-making you do, for example, during a product development. But I think there's important context here to add, which is that previously, like when we would, for example, develop a marathon running shoe or performance bike, the process would be highly empirical, right? Like, so we would basically come up with a concept or a design with some level of hypothesis. We build a prototype. We maybe bring that to a local athlete and try it out. And then based on that subjective feedback, we would try to to, let's say, come up with the next step or develop, develop that, that system further. I think now when we, when we talk about data and like the vast amounts of data that we can now collect and start to understand better and better with, uh, with FEA and other digital models to, to model material behavior or athlete biomechanics and so on, even though we might lose something or we might not 100% be accurate to the reality, we're going to be so much closer. We have so much more information we can now leverage to determine the next steps in the development. And to give you a very seemingly minor example of this, like when I started here, I think it's roughly 4 years ago now, one of the insights we had regarding footwear was, and I think this was kind of across the industry is like how foam behavior can determine performance. And at that time, this was still like a pretty mysterious object, like exactly what is it? Like, is it the energy return? Is it the absorption, the vertical stiffness? Like, what is it exactly? And I did something very simple, actually, which was just taking a variety of materials looking at the layering. Jesus earlier mentioned within a shoe, there is often a layer cake of materials that can either help or inhibit performance. And we, in very close detail, like made all these material stacks, tested them in a lab, and then brought that into our FEA models. And that seemingly small step, like really helped, like dial in that model to become much, much more accurate. Okay, maybe you're still excluding their gait analysis or joint work or whatever have you, but at least the jumping off point is already that much more accurate, right? And I think now as an industry and also heavily powered and kind of accelerated by AI, I think we'll get closer and closer to that reality. And the other aspect here that I would highlight is the quicker you can validate like your virtual model in the real world, like the sooner you'll be on the right track, right? So I think one of the things we do very intensively at ON here is make sure that we have very short like prototyping cycles. Like we can now really go from, let's say, digital model to runnable shoe in actual performing materials in a matter of days or sometimes even hours if we push it. So I think that's another really important aspect for validating that the digital model with the real-world experience.

These are great points. Any takers on data? Okay, Jesus.

Yes, I mean, I always be data. I think, as you say, it is important also to understand where it goes, like you were saying, like having the layerings, understanding the materials and all the interactions that they have, no? So for example, when you see a footwear, there is so many levels of different interactions and And the interaction could be maybe the contact between the ground, the rover, and then you have like a list of rover types, no? And then you have like the contact of the insoles with the foot, with the upper, and so on. So I think the data is also intrinsically related with who is actually— I mean, with what type of data do you have? What's the quality of that data? And I think it's the hardest thing to get anyway. And sometimes even if you try to get it by yourself, we are a really small studio. So we try to, I don't know, try to go for a run, try to go in collaboration with the university, trying to even get the data directly from the clients. And all this analysis is the hardest part because if it's not quality data, then you need to come up with a solution. Like Onur was saying, how can we shrink the solution into the space that they actually— data may give us like enough of a, let's say, like a view of the problem that we are trying to get. So that's why I would say that one of the problems and also one of the perks of computational design is data and how data needs to be related with all our work streams. So that's, let's say, like a well our main point, but also our main weakness, I would say like that. But I will leave it to you.

No, no, that, that, that— this is all great. And actually, it kind of brings me to the maybe one of the core discussions that everybody wants to hear. And instead of, you know, because it starts with data, and then it was big data, and then it was suddenly AI, uh, generative AI exploded actually back in 2022. And today everybody uses generative AI interchangeably with AI, which is a much, you know, broader term, uh, for any sort of like smart technology that you can computationally build. So I just want to kind of clarify that a little bit and just put like— start the discussion. And I want everybody's feedback on this for like 30 seconds each. Okay, so the question is, one, first of all, let's say AI is not only about generative AI. Should we talk about other sorts of AI when we are talking about software design. That's question 1, or the, you know, discussion 1. And the second discussion is about how you position AI for yourself, or how you position yourself with AI. Do you think AI is a collaborator, as a servant, as a tool, as something that's going to lead you in the future, right? So these are kind of like the different roles that you can think about AI. Yeah, I'm just kind of like curious about this landscape because it's approaching— it's it's expanding fast. It's not approaching, it's already here. It's been here for a long time. So, and I'll go with, uh, like Maya, Renee, Jesus, and Eric. So I just want to hear your takes in 30 seconds each.

Thank you, um, I'm not really sure I can answer very deeply into that because I use AI tools that are— I think most designers use them. Um, I'm still a bit skeptical and, uh, you know, looking at Everything with a bit of caution, but I definitely think that they can be applied. I think it's very important you understand how these things are used and exactly where they are applied. I think, I believe they have space in the footwear industry and computational design and in automation specifically. Um, I just feel like, uh, it has to be the, we still need the critical thinking and the critical view of this.

Great. René?

René?

Yes. Also, happy moment to have AI with us. I use it mostly for almost everything as a tool. Of course, I cannot trust everything about AI. But what I like more is happy accidents where I'm researching for development. Sometimes these hallucinations from AI, they are giving you some feedback so way out of the box that you say, okay, this is nonsense, but what if we save it for later? Because you can start a whole new way of thinking or new project from all these hallucinations and this API actions. Back to the data issue for us to feed up the AI, mostly since we're doing almost the same things since 75 years here, we have a lot of data. Now the challenge is to take all this data and, as Jesús said, to filter it and shrink it into something that we can use. To use AI, for example, to predict what will be the new way of making things better. Jesus.

For me, AI, I mean, I've been exploring from the beginning, like, to see reaction diffusion and what can we use it for creating different tools. At the beginning, it was more like image-related, but I think now I'm starting to believe that it could be more like a companion somehow. Like, it can start actually help you to build something when you start, like, getting into the correct, let's say, sintony. Something that I really realized with AI is that actually you need to know what you're asking somehow. So you cannot ask, like, I want you to create me Grasshopper, and then the AI is going to just create you a full back end and all the nodes and so on. So you actually need to know like what's a dictionary, what's a library, what's a tree, what's— I mean, you need to like make some background there. But if you are able to actually ask exactly what you were asking for and then using it like a component to build something, I think it could be a really powerful thing.

And Erik?

Yeah, I think it's— for me, it's extremely exciting. Like, I think we are extremely lucky to be in this time where these tools start to emerge. I don't know how it is for you, for you all, but like, I absolutely enjoy the play that you can have with different LLMs. To give you an example, like, I have no coding or programming experience, but all of a sudden you have the power of an extremely experienced coder at the tip of your hands. And yeah, I think this is such an accelerator. I think it's such an unlock for ideas to flow more easily and faster and get feedback on them more quickly because you can prototype and deploy ideas more, more quickly in a, in a very safe and protected environment. So actually, I'm very— I like that we can be loose with it, and I like that we can be Although understanding the limitations, like I often think of the example, like if a year or a few years back, I needed to render out the design, I would need to commit a couple hours to this, right? And I think now I can just go with a concept on a gut feeling and quickly see where that takes me. And then I don't waste any time basically, right? You're so much more efficient. So you can just, you know, run 10 ideas in parallel. And I think that also really shifts the notion of what a designer does from, let's say, laying down this perfect line and then, you know, making that into a product to, you know, potentially much more of a curator of ideas and concepts and so on. I think, yeah, I think that's something that is very powerful. Let's see how it influences our, our industry and design and computational design and so on. I think we're about to, I think, feel the full impact of this in the coming years. But I think it's very exciting for sure.

Yeah. Now, this is all great. And one thing, one thread that I'm also seeing with including all the questions about data simulations and AI is that the disciplines are kind of like, they were getting together already closer. The proximity was increasing. But it's almost through data we are able to create this mesh of understanding or abstraction across disciplines, right? So I'm curious about— and we don't reinvent everything. So I have, you know, I have an architecture background, computational design and architecture. One of my associates at New Balance, again, she had like 10 years of experience building design before coming to New Balance. And architects are pretty good in systems thinking. That's what I realize and think. And also, Teaching to them also made me realize that too, which pays the dividends in footwear design. So we are learning from other disciplines. So what do you think about architecture, systems engineering, or gaming can teach us in terms of computational design in footwear design? I'll take again 30 seconds from each of your time so we can respond to the questions of the audience. Again, I'll start with Maya.

Thank you. I mean, I can just confirm what you said, really. I, I daily apply what I learned in architecture, or what I, what I studied actually, in my work, because architecture taught me to, to think in systems and, and in connection of elements. And that's exactly how would I approach and how I would build a computational workflow in looking into it as a system. Also, there is this quote that I particularly like from an architect called Helmut Jahn who says that every building is a prototype. And it's very fundamentally different than footwear industry where you need to produce a lot in hundreds and millions of times. But in innovation context, it's actually very applicable because every experiment and every prototype is, uh, it's, it's a new possibility to explore and to see how different elements interact. So I think that this type of thinking is extremely applicable in this context. And I, at least that's how I approach most problems here. And finally, my time in academia, I really value a lot because it taught me to be truly self— not self, but transdisciplinary, which means that I can move through different specialties and talk to different people, which I think is fundamental also in this industry is not really, um, uh, isolated for one thing or another. And I, I see that it's very funny that computational designers, I think we are seen from outside as very specialized, but I, I think all of us define ourselves as generalists. And I think this, this contradiction is super interesting. And I do believe we are mostly generalist focused on something specific, I think it, it requires, uh, understanding on many different topics to be able to do that.

Yeah, my solution to that, Maya, sorry to cut you off, but it was calling ourselves a generalist specialist because it's weird for architecture too. It's like very broad range, but you also get deep, you know, into kind of a couple of subjects. So you can't just say, hey, I'm a general specialist, and then you're done.

I will try that.

Thank you.

Perfect. Renee, what's your take on the subject?

Well, I think many fields are inspirational. I know it sounds cliché, but I think we need to look at nature to understand what we can do on the metamaterial direction. So now we can engineer and create things that artificial, they are artificial, but when you have inspiration from nature, with a deep curiosity on everything what is happening. Nature in a very broad way, because for example biomechanics for Jesús is very important. But also if you— for us, if you have to know how to make a mold construction or how any process in the factory is working, you have to go deeply into the technical part. But in the technical, when you can apply what nature is doing to improve that system or to know how it works, it's already in nature. So I think that's the biggest field we can learn from. And handicrafts, once again, if we look back on the traditional shoemaking, We don't need to find more answers. Most of them, they were used or they were done before. And if you check on the history, there are some answers for most of the problems.

Great. Jesus?

Yeah, I think, I mean, for me this is an interesting question because I was saying coming from different backgrounds is you need to connect all these kind of dots. And it's always— I remember that I've always been recognized as like a 360-something, like design engineer or neither of those. But at the same time, from an innovation perspective, when I start— the first time that I approached innovation was in 2013 with design thinking. So design thinking, somehow, it is to think out of the box. So in that way, it's like you bring knowledge from different areas and try to apply it in the ways in which you probably need it. So that was really interesting to see how actually you can bring ideas from different perspectives and try to find a solution. So, and also, as Renée is saying, maybe the solution is already out there. It's just like the way of framing it and seeing it. So that, that's why I think it's important to have a broad vision and also like a wide perspective so you can have a better understanding of what's happening.

Yeah. Eric?

Yeah, I think from where I sit in ON at the moment and our teams are developing and growing very rapidly, so I'm exposed to a lot of different candidates from, luckily, actually a lot of different fields in my current role when it comes to hiring for the positions that we have. And I think because of some of the technologies that we've launched and those not being your standard footwear technologies, like we're now exposed to a whole scale of like new skill sets people people from different backgrounds, robotics engineers, inkjet experts, that probably if we had stayed within our footwear lane, we would have not been exposed to. So the reason I say that is because like once you have these new perspectives coming in, or like take Maya as an example, bring in an architecture, architect point of view into footwear, I think those are the intersections where interesting things start to happen. And I think it's something that we should embrace more within our industry. I think we tend to become at times quite insular, and I think this is heavily grounded in the fact that our industry is so grounded in the tradition of how footwear is made and it's done in a certain way and the last comes first, and there's all these things. I'm definitely not opposed to those, but I think we should create fertile ground to challenge these things, right? Because they also hold us back in the past in some ways. So I think, yeah, exposing ourselves to other industries, other fields is absolutely critical for our progress.

Wonderful. I think this sums up all the questions in a great way. And I've been tracking also the questions from the audience, and I think we answered all of them. There are questions about data. I think we explained that you can, you can collect data or create data, but there will always be gaps. The other ones are about the tools or where to learn. Some other people responded to that. We can get back to it, and I would suggest everybody in the call watching this to find these names on LinkedIn because pretty active people and also we are trying to create these resources to keep the discussion going forever, right? 24/7, not only for this kind of like discussion live here. But one thing, one question that I would like to take on is, because it's very valuable for the discussion, is this: How far are we, or how far away is the technology from being able to use generative AI algorithms to generate valid and useful technical models of shoes instead of getting just a sketch or raster image. So this question is about can GenAI, and I expand it to AI as well, or machine learning, can create truly usable 3D models and how valid would they be? Any takers on this? Maya, what do you think?

I can take this very fast and maybe Eric, you can add something. And also, I think we spoke about this in the beginning of the panel, and it's really important in this, in this industry specifically that you tie a 3D model to something. And there are very different 3D models for very different purposes. If it's for mold making, René, probably you know that it looks very different from one that you would only need to render, or you would need to maybe use it for spring or or something like that. So I think until the moment that we find a way to link that physical output to that side of the things, maybe not, not really. It's, it's a bit utopic, but yeah, I think definitely the part of making and, and the physical output of whatever it is, it doesn't mean it has to be the complete, the complete show, the complete product. I think until that link is established more clearly, we would be limited with AI. Hope that answers the question.

Yeah, René, any response to this?

Yeah, I think in the technical mode, it's not so easy to get the result. As Maja said, we are a bit far away, sadly for José. I have to be honest. I think it's not that easy to go more or beyond the render or a nice image, which is great. It's a good starting point. But then all the development is something we need to do still with human intelligence, I hope. But now artificial intelligence is helping us, is supporting. But anyway, we need the, our expertise.

Maybe I can add a little bit. The, the sketch or raster image to 3D, I would almost consider that like that's the, the model learning how to crawl or how to maybe start to walk. Um, of course that's still like, uh, quite far away from being able to you know, just, you know, input a, let's say, a treadmill running gait analysis and then a generative model automatically creating a fully optimized shoe. I think that's still, still some ways out, but there's not really any reason why that could not happen in the future, right? Like, we, we do all these things independently. Like, we create the footwear, we know all the manufacturing restraints, we know the material behaviors, we can pretty accurately study the gate analysis. I think it will just take us some time to connect all these things and in a meaningful and accurate way that it will generate like a truly usable output. But I don't really see why that would not be the case in the future. I think AI is moving so extremely quick at this point that if we can collect the data, and it brings us back to the data point, um, and, and truly understand it, I think we can get quite close.

That's great. So, you know, I also get the— people ask this question to me too, right? So I'm like, okay, I touch that to you this time, and I have a very short answer, but it's visual. But before I go into that, actually, I'm going to combine everything again coming from the data, understanding, okay, knowledge. We first talked about converting knowledge into bits, numbers, that you cannot convert all meanings or understanding of real life into numbers. So that's one, which is data. So data is always missing. And the second thing is you can enlarge datasets, but then you have to make insights from it, which is another challenge, cleaning data and building insights on top of it. And then the third thing or the fourth thing, as René said, you are trying to make something even better. Or for creativity, you are trying to come up with the next design, which is another challenge. So here's my answer for that. You know, I usually show this. This is your data, and the distribution can be different, right? And, uh, and AI is built with data, right? But the problem is you want that which lies there. So for all the questions that we received today, I'm gonna, I'm gonna spice it up with my, you know, this last bit, which is You can use AI to generate models or you can use automated systems to make new designs and so on and so forth. The question is, do you— are you— first question is, are you aware of the gaps which all speakers here, everybody started, you know, everybody kept hitting on those gaps, right? So first you have to be aware of the gaps. And the second thing is, because of that, do you really want it? Do you want to really use these tools which are kind of like missing the bits? Which I would conclude with: use computational design to the bones, use AI data to the bones, but at the same time remain aware of the gaps. Thank you. Michael.

Thank you so, so much to all of you. That was amazing. I'll admit a lot of it went over my head, but it was amazing and glad to see so many things coming in on the chat. So thank you so, so, so much for your time today, tonight for some of you, and for all of those of you who tuned in, thank you for joining us. As Onur said, all of our panelists are on LinkedIn. So if you want to know any more, please do reach out to them and connect to them there. And look, we hope that this panel sparks some new ways of thinking about computational design, not as a futuristic add-on, but as a mindset that can can really expand creativity, deepen insight, and open up entirely new possibilities for how we design and build footwear. We will be continuing these conversations, as I mentioned, at our Stride USA and Europe 2026 events. So if today's discussion did resonate with you, then Stride is absolutely the place for you to be. You can find more information about that on our website, or feel free to reach out to me directly at mkr@pi.tv. That's MikeKiloRomeo@pi.tv. Uh, and thanks again. Looking forward to seeing you all soon.