Codex Futures
In this debut season, our global team at TRIPTK sits down with leaders from some of the world’s most dynamic and influential companies to explore what the future holds for how we live, learn, and play.
We’ll be sharing insights from executives at several iconic global brands who are category‑defining innovators in technology, hospitality, entertainment, and consumer culture, to name a few.
TRIPTK is your partner for brand-led transformation. We put Conviction into Action, aligning your organization around a new future. Our multidisciplinary team recodes brands into a unified system for growth, transforming brands inside and out to create enduring business value.
Visit www.TRIPTK.com for more information on our global studio.
Codex Futures
Exploring the Digital Twin Revolution with Dirk Hartmaan
Use Left/Right to seek, Home/End to jump to start or end. Hold shift to jump forward or backward.
In this episode of Codex Futures, Dirk Hartmann, an expert in digital twins and technology innovation at Siemens, discusses the transformative potential of digital twins across various industries. He emphasises the importance of curiosity in driving innovation, the role of prototyping, and the intersection of AI and digital twins. Hartmann also highlights the cultural conditions necessary for fostering innovation, such as trust and critical thinking, while envisioning a future where digital twins enhance everyday life without being overtly visible.
Disclaimer:
The views and opinions expressed by the guest on this podcast are solely their own and do not necessarily reflect the official policy or position of Siemens.
TRIPTK is your partner for brand-led transformation.
So I think the perfect example of a digital twin is actually what we saw in the Matrix movie trilogy. Yes, where there was a complete digital copy of the real world, and you could not really distinguish whether you're in the digital channel. There's probably something not very realistic in the near future. But perfectly sketches what digital twins are all about. Experience, opportunity upfront in the digital world, where they are cheap, where you can explore many, many different options to then realize them in the real world.
SPEAKER_01Hello and welcome to Codex Futures, a triptych podcast decoding how we'll live, learn, and play in the future. Each episode, we dive into a different industry to uncover the trends, cultural shifts, and powerful forces reshaping it, and what those changes mean for the people driving innovation forward. In today's episode, we're thrilled to be joined by Dirk Hartman, a university professor and expert in digital twins and head of SIM Center Technology Innovation at Siemens. Together we'll explore how these technologies are changing the way things are designed, tested, and run across factories, energy systems, transportation, and beyond. Welcome, Dirk. Wonderful to have you.
SPEAKER_00Thanks a lot for having me. Looking forward to the conversation.
SPEAKER_01So you have a fantastic title and deep expertise. Would love if you could share more about, I guess, what that means in your own words and your journey into that role today.
SPEAKER_00What that means is I think technology innovation, okay, you innovate tech technology. Maybe to give it a specific twist there, because you know, many people thinking about Siemens, it's the fridges, et cetera. But as you pointed out, yes, these days there has been quite a transformation. And actually, during that transformation, we built up quite a software portfolio. So I'm leading the technology innovation team, part of the software portfolio, which is mimicking things in the real world, in the digital world, and trying to predict the behavior, how it would react in the real world. So this is a bit kind of the portfolio and innovation part I'm responsible for. And innovation sounds like, okay, we build stuff, yes, but that's only one part because obviously you need to know what stuff to build. So a large part of this is understanding what is feasible from a technology point of view, in a way, also connecting to my role at the university. Yes, once you figure out a unique opportunity, let's say some machine learning algorithm you could use to accelerate the capability of our current portfolio, accelerate that by, let's say, two or three orders of magnitude, meaning 100,000 times faster predictions, then typically it's all about figuring out where this is really useful. Yes. Obviously, I do have my ideas where this is useful. So that means speaking to tons of customers, trying to understand where are the gaps, where are the opportunities. Then it comes to the realization, and in the end, well, it's also very much about marketing the stuff, uh, what we do so that people learn from this, and on the other side, also really to underline the broad technical capabilities of Siemens. So, from all this, you know, a bit of sales, a bit of a researcher, a bit of marketing, a bit of everything.
SPEAKER_01Oh, interesting. No, no, I think that really brings it to life. So, I mean, it makes me think as you as the conduit. Like, so you understand the tech and the research, but you also have to connect that to the kind of the human problem and the customer and identify the use case, and then I guess communicate it in a way that lands with impact to them. Where do you start then? Is your starting point figuring out or thinking about what's the technology first and then you find the use case? Or is it customer first? Because I guess there's three things in play. Where do you start and what's that toggle between them like?
SPEAKER_00Well, I think the key starting point is always curiosity. And then, you know, you could be curious about a specific customer problem because this is really a pain for that very specific customer. If you would resolve that, that would unlock enormous value for the customer. It's figuring that out. Yes. Maybe at another opportunity, I get excited about a technology I pick up at a conference and then just look a bit, you know, where would this fit? So I truly think there is no golden rule how to approach innovation except driven by curiosity.
SPEAKER_01I love that. The starting point is curiosity. No matter how, I guess, you know, you obviously have a deep technical expertise, no matter how far you go into that, you know, you need to keep curiosity at the heart of it. Could you bring that to life with, I don't know, if something that you've worked on previously. I think like just connecting that technology to the customer problem would love to, you know, whether there's an example.
SPEAKER_00So there was a point in time, actually these days, many people talk about this, that Nvidia introduced something called CUDA. So CUDA is a way that you could program graphic processing units, so GPUs, not to be used for graphics, but to use the machine learning for any other kind of algorithm you might have have in mind. And obviously, when the when that came out, we saw the massive opportunities, capabilities, and started tinkering around a bit what could be done. At that point, I think it was not really clear how this would make its natural way into our software technology portfolio back back then. But then something happened early 2010, 15, something like this, additive manufacturing came up, 3D printing. All of a sudden, customers had a problem, how to exploit this massive set of opportunities you can create, yes, because 3D printing has no limits in in terms of your imagination in its shape, whatever, to produce. So they had the challenge, um, how to produce optimal parts beyond human imagination, beyond the stuff that we're used to. So this was kind of the match we made that we said, look, at the one inside we have this massive compute capabilities. We can compute things close to real time. So you bring in the thing you would like to imagine, roughly where do you have boundary forces applied, how that fits. And then the computer imagines basically the design, feeds that back to the designer, and in this loop comes up with much, much more creative designs. And so that was really, I think, one example where just get curious about the technology, started tinkering around, not finding good fit, then you know, found the right fit on the additive manufacturing side. When then came the long journey, all the way, you know, getting this to a product, and a couple of years later, it's actually in the products.
SPEAKER_01What's the product at the end of it?
SPEAKER_00Then it's like 3D printing. It's a software which helps you to imagine the design you would like to print. If you'd like to print something, first of all, you need to bring it into a digital shape so that you can actually then you're like a Word document you would send to the printer. So say, you know, bring something into a 3D shape, send that to the 3D printer. And our software innovation really helps you to come to this optimal shape, you can then send to the printer. And a specific element there was in this part of the software, so it's known as topology optimization, that you can print something uh with a minimal weight, and in 3D printing, that's quite interesting because printing price scales was weight, and you get wonderful bionic shapes uh in the end.
SPEAKER_01Oh, interesting. Wow. So it's the software, and and apologies if I um dumb it down slightly, but it's the software that works with kind of, I guess, the the human prompting and kind of the design that people have in their head and translates it into something that can be used in a 3D printer.
SPEAKER_00So one part of the software really helps the person to create that 3D shape. Uh, in this case, there is not much compute involved because the human basically interacts with the computer with the computer-aided design system to create that shape. But then that is the shape imagined by the human, and then our software kicks in afterwards trying to see how this can be further optimized. Yes. How to bring in bionic type of structures to reduce the weight, to reduce the cost for printing and the like.
SPEAKER_01Okay. Oh, wow. Amazing. And how did you go about, I guess, fine-tuning what the use case is for that software? Were you was it kind of user testing with people and and um exploring how people might use it? What was the journey into kind of evolving and creating it?
SPEAKER_00Indeed, true. So um at Siemens compared to many other software companies, we have the unique setup that at the same time we are a big hardware company. And so we do have many, many people using our own tools inside to create other products of Siemens. So we produced a series of prototypes, gave them away for test users within Siemens, uh, collected that feedback, you know, then you get geometries, challenges you cannot address. We found the software, et cetera. That was really a very, very big asset to address this.
SPEAKER_01Could you tell us a little bit more about the setup of I guess, you know, this of how this happens? And so you mentioned there's different people with different expertise within the innovation center. How does innovation happen at Siemens?
SPEAKER_00I think key driver is their curiosity. So obviously, we do have all in the team very, very different expertise, but but in the end, it it all starts with, you know, what if uh I could do this, or you know, what if I would take this technology, bring, bring it somewhere else. Typically it's it's very much a teamwork because everybody comes out of his or her domain. And in the end, I think innovation comes together. Innovation happens, takes place when you bring together different ingredients, the right technology, maybe another piece of technology, the customer challenge. And in this way, typically um things are are shaped. And then another main point or important point for us is bringing in prototypes as early as possible, trying it out first internally, then with with other uh test customers, etc. And evolve.
SPEAKER_01And when you say prototypes, do you mean a physical prototype?
SPEAKER_00Prototypes in our case is always computer-based. Yes, so our main product is software. So it depends a bit. Yes, a prototype could be as much as used to be as as little as something mocked up in PowerPoint, you know, how the user interface would look like. Okay, these days with large language models, the whole capability of machine learning, prototyping is much easier. So it's really very often functional prototypes and then refining that uh over over iterations.
SPEAKER_01Okay, very much so that kind of experimental loop of kind of testing something out, getting some user feedback and iterating. Okay. And what's, I guess, you know, thinking about this is the future of live, learn and play, what what's exciting you about the future at the moment? Like are there some you know, forces or kind of technological advancements that are particularly exciting you at the moment?
SPEAKER_00Going back to that example, Jess, yes. So before we be put small POCs in in PowerPoint, you know, this is how the user interface would look like, and then you know, go from one slide to the next slide, visualizing a bit the flow. But now with this agentic AI, the copilots coding gets so much easier. So that basically our jobs before have been 5% of creativity, 20% of talking to people, and then really 75% probably of the work, putting things into code. And that's not really the value general. Well, in the end, it generates value because this is what's what's used in the end. But that's so Thai consuming. With the large language models, the co-pilots, you can automate that to a very high degree and realize innovations specifically on the software side at a much, much higher pace.
SPEAKER_01Okay, interesting. So actually I'm curious about because I guess that's what a lot lots of companies are thinking about at the moment is you know, bringing agentic workflow into how you do things. Like so it sounds like is it mostly in the coding phase? Like where are you bringing in agents or AI across that innovation process?
SPEAKER_00On the coding phase, it helps a lot slot us as a team. But agents have have a much, much, much bigger capability. So I I work in in digital twins and simulation. And since it's very much out driven out of an expert community, yes, I mean it's built by experts for experts. So you had here another menu, there another button, uh, there another parameter to add. So it gets super, super complex software. This high complexity limits actually the adoption in a broad scale. So there was a recent study on digital twins. You know, everybody loves digital twins, finds the idea great to mimic what you see in the real world, in the digital world, try stuff out first, but it does not really scale. And here, really, agentic AI agents, I think, will really boost uh usability of these systems in the coming years. So that's why I'm getting super, super excited about agentic AI.
SPEAKER_01Would love if you could share a little bit more. I guess maybe for someone who's quite new to digital twins and and the world around it, if you could give us a little bit of uh uh overview and then we'd love to dig into some of the details.
SPEAKER_00So I think the perfect example of a digital twin is actually what we could saw in the Matrix movie trilogy. Yes, where there was a complete digital copy of the real world, and you could not really distinguish whether you're in the digital design or not. That's probably something not very realistic in the near future, but perfectly sketches what what digital twins are are all about. Experience opportunities, experience decisions upfront in in the digital world, where they're cheap, where you can explore many, many different options, or you then realize them in the real world. So the question is why is it actually called a twin? Yes, I mean on the one hand side that that makes sense. But on the other hand, um, again, going back to the movies, there's a famous movie Apollo 13, probably also 10, 10 years old by by now, but is about uh landing on the moon, and then you know something happened on the spaceship, on the moon lander, and it was about troubleshooting. And for each of these spaceships or landers, NASA had a twin in their basement. And so, really, in that movie, you can see people going back to the twin, doing the troubleshooting, figuring out how to resolve that issue, get the crew safely back to Earth. And at some point, NASA said, Well, can't we do that in the digital version? And that's where digital twins, in a way, the concept was born around 2010. The ultimate goal would be something like the matrix. Cannot distinguish really the digital from the real. That's probably not realistic at all. So at the moment, I think that is kind of the uh goal. I'm I'm I'm trying to achieve, you know, have something sufficiently realistic, you would accept to the reality to take some decisions.
SPEAKER_01Is that the main use case for them? It's to, I guess, simulate and test and and before you actually launch something in the in the real world.
SPEAKER_00The term digital twin is is new. Again, NASAR sending sending rockets to the moon, trying to troubleshoot, trying to understand if something went wrong. Some of the stuff you cannot really experiment small scale under these conditions. So they went back to computer models, implemented the physics equation on the computer, trying to simulate what would happen. So increasing this understanding, understanding what would cause the trouble during the start, how to make things more reliable helped them a lot in the design and engineering process. And then, you know, it was only really for the very, very deep experts at NASA, and then this made then a kind of success story: computer-aided engineering, computer-aided design. So using simulation tools, digital tools to support de-risk decisions in engineering. Testing virtual is now probably adopted for any kind of product in our daily life. So this is more or less state of the art in engineering. And then the additional thing from the digital twins on top is you know, we have these wonderful engineering models we use to base our engineering decisions on, is how thing is a certain metal piece in a car so that the metal supports a potential crash, the driver still stays safe. And how can we translate these wonderful digital models to the operations and figure out how we can operate certain elements more and more effective? Let's take again a car or a truck example. Yes, you could run a digital twin of a truck while the truck is driving through this digital information, optimize how that truck is operated, meaning you know what gear to choose, uh at which speed to drive, so to still keep your timetable. And because the digital model can try out different scenarios, it can really choose the optimal one. And there's nice cases where these digital models then really help trucks to save five or six percent of fuel just by appropriate use of models. And this opens, I would say, a completely new new new space because trucks are there, but putting just an additional piece of software on them as retrofitting is rather easy, rather cheap, doesn't produce any CO2, but at the same time you can realize massive sustainability benefits as CO2 reduction. And that's a bit the journey, yes, understanding in the engineering, all the way something running in parallel while you use um different devices, cars, trucks, planes.
SPEAKER_01So you talked about transportation. Are there how is it being used in other industries? Was transportation the main one?
SPEAKER_00So it really kicks in, I would say, in all industries. You see, you see many, many different people speaking about it. I think while we see most probably in aerospace transportation at at the moment, is because they are quite far advanced in digitalization using these computer tools since ages to design cars, to design planes, as we spoke about the rocket. So that's why it's it's really dominant there. Very much used and probably also there quite quite a long history in the process industry. Yes, in those big process plants, you cannot measure everything, so digital models can help you to generate additional insight, run things more optimally. Robots is a very, very big field these days. Everybody speaks about physical AI, yes, robots, physical understanding, play in a way upfront, what happens if I do this decision. And even and that I personally found quite surprising. Yes, even in the food industry, this this can be used. Yes, how to optimally process your food to make sure it's really heated appropriately, not getting too hot. So there's fantastic works about, you know, how can you use AI to optimize cooking your turkey as an example? So there's probably no field really where this is not not entering.
SPEAKER_01Again, apologies if this is a basic question, but I don't know, when is it just AI and when is it a digital twin? Or is it both kind of interrelated?
SPEAKER_00I would argue that that the digital twin is more a view on how we use this stuff. Yes. So we use digital twin to do something, to take de-risk these decisions, play, play, play through different scenarios. And then technology inside is also to a large part AI, machine learning. And the uniqueness there is very often that is combined with classical physics models. Because many of those industrial decisions where you run digital twins might have severe implications if you take the wrong decisions. Yes, I mean, if a big machine truck is in trouble, that that can really harm other people. So you have a completely different requirements in terms of reliability than you know, just if you order the book, get a book recommended. And so that's what it would make it exciting. This combination of you bring in a bit of machine learning, need to think about how can I make this more reliable, how can I fuse in physics. Um it's really bringing all these different domains. So to put it short, yes, AI machine learning is a part of it.
SPEAKER_01And how do you see that technology or digital twins evolving in the, you know, with AI just you know accelerating, both in terms of its ability and adoption curve? What do you see the future for digital twins to be?
SPEAKER_00I mean, on the one hand side, there's a there's this massive acceleration in compute power. There's this massive acceleration in data. So you so you could do this fantastic training of machine learning. Um, there's another element in there, because I'm actually by training a mathematician. If you put those algorithms designed by mathematicians to train machine learning models to solve these physics equations, they also grow in their capabilities with such an exponential curve. Yes. So you everywhere you see exponential curves, curve, and it is very hard to predict. At the same time, if you would have asked me, you know, back in 22, you know, would there be something like ChatGPT? I would have said, yes, very likely, in 10 years. So I really stopped doing predictions about timeline.
SPEAKER_01Because I mean this exponential evolution is so hard to when you say, for example, you think with AI enabling digital twin to be more mainstream and kind of more people can use it, what do you think that might look like for, I guess, the everyday like because at the moment it does seem it's obviously there, but it's within certain industries. What do you see it look like for I guess a mass consumer audience?
SPEAKER_00So I would say in the from point of a mass consumer user, you would not see it at all. There's a few people like me, probably, who laugh tinkering around with digital twins, you know, configuring these super complex simulation models. So ideally, it's like in the case in a truck, yes, some miracle happening somewhere in the computer, just making sure you drive more efficiently, your house is heated more, more efficiently. So I would rather envision, yes, we interact daily in many different situations. Cook our turkey, uh sorry, digital twin in the background, make sure it gets the right crispiness, etc. So everybody will use it, but but not necessarily see it.
SPEAKER_01So you don't see a world where I guess in like how ChatGBT, everyone is experimenting with that in their own way. Do you think there will be a version of that where I don't know, everyone is creating their own digital twin in certain aspects of their life?
SPEAKER_00It's a good question, yes. And if you look at Siemens, we we we speak a lot about the industrial metaverse. And then I think if you speak about the industrial metaverse, there's one issue that that's very often understood as what people and connotate with the consumer metaverse, yes, meaning 3D virtual worlds where you hang out with your friends, etc. And I mean this has been demonstrated not really to be taken off as expected, living and continuously interacting with these 3D elements. Now, on the industrial side, this is really more you know the thing running in the background in in the shadow you don't see but which constantly is de-risking your or optimizing your decisions. So that's why I would say from from the ideal point of view you would not see this. I very much find that perspective of of the metaverse quite interesting. During COVID we experimented a lot with that gave it a completely different flavor of social interactions, etc. But yeah, hasn't taken off so probably still something for for the nerdy side of users. No clue where that is going towards too.
SPEAKER_01I'm curious actually your kind of personal journey because I think you're obviously a deep deep expert in this and you you lecture in it as well. What excites you the most about this this area and this technology? Like why this and kind of why have you dedicated your career to this?
SPEAKER_00So when I when I speak to to friends or whatever, you know, in in the retro perspective it all looks like like like a very straight straight career. But it was was a big zigzag move. So so you know I think the the constant aspect is is there really on the one hand side the the curiosity and working on on on challenging mathematics problems. But but at one point you know I I wanted to become an aerospace engineer then I want to become a physicist. So went to to Cambridge to study with Stephen Hawking. Then you know somehow found okay cannot really grasp this quantum mechanics or the cosmology part of it. Then I went to biology actually because there were some very interesting problems on on the biology side of of things. Then you know after my PhD in a postdoc I decided well maybe it's time you know to to go to industry to create a much bigger impact. And through this it all miraculously I I would say happened. But the constant is big interest in solving hard math problems and generating impact in the end.
SPEAKER_01Something I've been asking quite a lot of people I guess I'm curious about because I guess innovation is a big word and it shows up in different contexts. It's slightly there's some things that are the same but it's different in different industries and I think particularly in the context of where you sit kind of deep tech innovation. We talked about curiosity but like what are some of the capabilities and the culture that you look in to create in your teams to drive this type of work? Like what are some of the cultural conditions for innovation to thrive?
SPEAKER_00The biggest cultural aspect is really trust. So it's really to create an atmosphere where people can you know express ideas in certain contexts maybe start to laugh about classify as absolutely crazy never ever realistic. And I think this kind of atmosphere is super important because in the end to realize true innovation you need to think out of the box. So from this point of view trust safety in in a way to come up with these crazy ideas is the most important element in in teams plus a good mixture of different backgrounds.
SPEAKER_01Absolutely okay so kind of diversity trust in your team is it trust in the team among others you know that that nobody is going to to to laugh about you.
SPEAKER_00Well so in the end you know creating safety is a personal safety feeling free to express these crazy ideas that I find super super important.
SPEAKER_01I'm curious and I think we've naturally talked a lot about you know the acceleration in technology and AI as as you know we think about the future are there any other big forces maybe outside of technology that you're seeing are going to heavily impact you know the the future or the work that you do?
SPEAKER_00Yeah I think technology is probably really really the main one because it enables so so many other things. Because I truly believe that that technology in the end is a key lever to address all the other problems we we we do have. Yes, sustainability, wealth, etc um feeding the world so that that makes me actually super optimistic that that we'll be able to address many many of these these problems. On the other hand I see there a bit risk and that goes again back to agents, large language models that our capability of critical thinking is being reduced. And I think really what is important to have on the one hand side you know the these crazy ideas, the machinery to implement help you a lot to accelerate the past, realize things, but then it requires also the critical step where you sort out the good things and then throw away the bad things. And that is something where I'm a bit afraid that with all these developments not really sure whether we can maintain in in general this this kind of critical thinking. But in general I'm super optimistic that technology is really a key leader to address many of these things.
SPEAKER_01Yeah no that I mean that's interesting the critical thinking point as well that I think absolutely I think same as you on the on the positive side but I think it's good to think about the the watch outs and the things that we want to maintain and that that don't get lost. And I think critical thinking we talk about this idea of conviction actually actually believing in you know when there's so many choices and you can get the answers so quickly but actually having real belief and conviction in the one that you're you're driving forward is going to be crucial.
SPEAKER_00That then requires good gut feeling etc and some of the good gut feeling you only get if you work yourself hard through through some of these challenges and obviously naturally having copiles etc will not force you anymore to to go through through this hard experience.
SPEAKER_01Yeah. I'm curious actually do you have guardrails or principles to make sure that we keep some of those things that are getting streamlined but making sure that we do apply you know critic critical thinking and you know I have absolutely no no no golden way how how to do this.
SPEAKER_00Yes and go a bit back to to my university role it's it's it's a fundamental question I'm I'm asking myself yes how would the education of over the next generation of students need to look like with having all these these fantastic tools and how to manage a good balance of of this critical thinking you know working yourself through through some of the hard problems but at the same time getting more fluent how to also exploit explore the the fantastic opportunities offered by co-pilots, large language models etc I'm really experimenting trying to understand uh but but I don't have have a final answer yet.
SPEAKER_01I don't think anyone does and I think that word experimenting is is key isn't it I think we're only going to move forward by experimenting and trying things out and figuring out what works. I'm curious actually you know university I think that's a huge and all just the education system. Like what's your perception of I guess how it's different when you're within that context and kind of lecturing and with students in today's age like are they is AI very much part of how they're learning.
SPEAKER_00It's quite interesting yes because on the on the one hand side you have the younger generation like my children kind of in a during school time are now growing up with this and trying to find very creative ways I would say how you know how this helps you with your homeworks etc specifically for for the subjects you you don't like. And then students at the moment they did not really you know grew up uh with was with large language models. So so there's somehow in between I sometimes have have the feeling at least for some of the students I am encouraging them more to use large large language models than they would do on on on their own. So I think all this we we still need need or also you know next generation will need to figure out.
SPEAKER_01Well thank you so much for your time really great conversation I feel like I've learned loads and thank you for helping to describe you know very complex things in a way that feels easier to understand. I appreciate that. If there's one idea or kind of takeout from this conversation that you'd love our listeners to take forward into their day to day what would that be?
SPEAKER_00Curiosity and never give up if you have a dream if you have an idea what works. And then whether it's technology and politics education in whatever system I think those two are are are the two most important elements.
SPEAKER_01Fantastic love that love that as curiosity as the constant in this changing world that's what's going to help people navigate through. Amazing well fantastic thank you so much Dirk be really great conversation really appreciate your time and your thoughts as well thanks a lot and I really enjoyed our chat thanks for listening to Codex Futures. If today's episode sparked new ideas we'd love to hear from you. Email me at louise.newson at triptych.com or click on the link below. Follow Triptych on LinkedIn and subscribe to our Codex news