AI in Operations for Hardware Companies: Lessons from the VanMoof Resurrection

I am here today with Elliot Dertheimer, who's the CEO of Vanmoof. Amazing bike brand that has become a household name across Europe. But, unfortunately, they grew a little bit too fast in a complex environment post COVID and had to fight for bankruptcy. And around that time, Elliot, who had just joined McLaren, applied to grow their micro mobility brand. Saw an opportunity that he couldn't miss. In this episode, I'm gonna unpack with Elliot how he's thinking about rebuilding the trust within VanMoof and how he's been growing the brand since. We're gonna think about how technology and especially AI can help him accelerate and improve the economics of the business. So thanks, Elliot, for being here.

And Thank you. Thank you for having me.

Perfect. So to set a bit the stage, Elliot, we've known each other for a bit of time. We studied together at Imperial in The UK. And back in the days, you were already building in the micro mobility space, with this company called Fiorosystems. So my first question for you is what originally got you excited about micromobility,

and would you have imagined that ten years later, we would you would be still speaking about being in the same space? So, look, I think at first, I was I was an aerospace engineer working between space agencies. So really looking at, like, deep tech, but in terms of, like, you know, complex advanced electronics that can do things that are relevant in space and in those kinds of environments and for those types of applications. And so a very intense research slash academic environment. And at the same time, within that context, we saw, like, batteries really improve in terms of energy density, in terms of pricing, in terms of accessibility. Same with, you know, control electronics. That's especially relevant for motors in the sense that now you could have much more compact motors that are efficient, that are silent, that deliver a lot more power than before. That's also how you saw the advent of drones. Right? Suddenly, motors were miniaturized and still able to output a lot of power. And so, you know, within that within that context at the time around 2015, it became very relevant that all these improvements had huge potential for applications in mobility in urban environments. Right? In the sense that you could just really take a lot of energy and pack it in a smaller package with, like, smaller motors that even a lot of performance in really informed factors that were optimal. So as a good gig, I initially approached it as, you know, what is you take a battery, you take a motor, like, what is the form factor that's, you know, the best in terms of efficiency and delivering moving from a to b. Actually, you cannot get to a skateboard, to be honest, which is basically what cars are built are built on today. Right? And then you get all the concerns about, you know, security and safety and usability and approachability and rider features. Right? And so eventually eventually, we converged on electric scooters and electric bikes actually just because, you know, in terms of stability, it's great. So that's how I ended up there. Right time, really interesting industry, that was just, you know, kind of fledging, like, beginning to grow and so on. So, that's how we started our first company, Fiora Systems. And we did I expect to still be there ten years later? I think that's a great question. I actually honestly, I don't know. You know? It's, it's been a journey, like, you know, super interesting one, like, in terms of progress, in terms of where the products are today, in terms of, you know, my own career and experiences. So, know, I would say I'm happy to still be here. Like, really, really interesting stuff happening right now. One thing that I noticed with you in the past ten years is from Fiori Systems when you were asking for feedback on the branding

to taking on Vanmoof. It seems that one thing that stayed consistent is the importance of brand and design in everything that you've worked on. And so what's your view on the importance of of design and branding in everything that you've worked on so far?

You know, on Vamove specifically, I think one of the things we saw with with Vamove that was extremely attractive to us and that we thought held a ton of value, much more than anyone realized from far away was actually I mean, I can cut it in two different areas around the brand. Right? The first one was the fan base and their resilience and their passion and retention around the Van Moof ecosystem. Right? So people absolutely loved Van Moof when they own their bike, and they tend to and they tended to just buy more Van Moofs. Right? They didn't wanna change brand, etcetera. That's, of course, part of the product as well, but, you know, the product builds up what the brands become. They were very resilient towards the issues that were happening. So meaning that, you know, despite all the noise out there and fair enough, like around repairs, access to spare parts, reliability issues, most of the rider base wanted to see a restart, wanted to see the mission fulfilled, wanted people to come in and fix the company and make it what it could have been, right, what it should have been. The last thing is, you know, and that was even more impressive, like, you know, VanMoof became a little bit of a of a cultural brand in the sense that, you know, it really in a way penetrated culture. So people are aware in VanMoof in many countries, although they've never really wanted to buy any bike. Like, they've heard of it. They kind of know what they do. Somehow, they relate to it aesthetically in terms of what VanMoof represents, a little bit the future, like mic mobility, like this new thing coming up that's quite cool that sometimes pushed against, commonly accepted, realities, you know, around the use of cars and, in cities and so on. So that's part of, you know, why, you know, we went for Van Wolf, and that's to say that actually it's extremely important. Right? Because a brand is many things, but mainly at a high level, it's how people, your customers, and beyond your customers. Right? The people that, you know, will interact with your name on your world in some way relate to you. Right? And of course, it's based tremendously around trust. Right? It's like, what is the amount of trust you project to those people interacting with you? And it's also very emotional. Like, what do we make them feel? Like, how do they feel when they see us? Right? And of course, you know, despite all the troubles it went through, Vamov retained a lot of, like, oh, and, you know, the product carried that kind of wow, that's the coolest bike out there. Like, wow, the features, you know, it's incredible what that bike does. You know? And so that that to us was hugely important. Mhmm. And to a wider extent, you can just observe that as an as an entrepreneur, you know, like as you start a project and it's awareness and the and you create a brand around it, you can see how people evolve around how they interact with that with that brand. Right? And how they might just commit to spending more with you or just want to be interacting with you more in whatever way that is. Right? So I think it's extremely interest important, especially in a world where now, you know, complex products are getting easier and easier to build. Mhmm. Right? So meaning that differentiation is harder and harder to to actually deliver. And so a big part of what you do around your product and how you make people feel and how they trust you is absolutely, like, super, super important in just achieving your goals and and getting to your targets and so what you can achieve with your business. Yeah.

This touches on a bunch of different things I wanna talk about later on, especially around the cost of development that is going down to zero. But just out of curiosity, do you still remember the first time you rode a van move, or what was your first experience riding a VanMoof or the first time you you actually came across that brand?

Yeah. Actually, the first time I rode a VanMoof was around the time we started to look at them in September 2023. You know, first, I was following them a lot from design, from an from an engineering perspective, from, like, patents and what they were doing in the industry and so on. And the first time I tried one, I was super impressed. And, actually, you could tangibly see, like, you know, the impact on feeling and performance that owning so much of their own development had. Like, it was great. Like, it felt amazing. It felt intuitive. It almost felt like what you'd expect something like that to feel, which is smooth, you know, with very low latency, very quick, very powerful. Basically, a product that delivers what it's supposed to. Right? Like, going through a to b at any point with a lot with a lot of fun no matter the environment, the context, and so on. So that's just like the the I remember being particularly

impressed by the feeling of the ride, which I didn't, but the way the ride felt was too And there's a lot that goes into it. I I remember listening to to one of your podcasts talking about the chain. Could you elaborate elaborate probably just a little bit more? Like, experience, why do you it's it's it makes it so special? Yeah. Because,

like, VanMoof is basically at the core. A bunch of very talented, like, you know, user like, for user experience, a user interface designers, but also, you know, industrial designers and, you know, writers themselves that spend a lot of time thinking about what is the best feeling and experience, like, I can deliver through that product in the context of where it's used. Right? And so, you know, that's backed up then by an incredible team of engineers that create electronics and software that deliver that experience. And this works in a system altogether, like iterating between ideas and experimentations and, you know, first prototypes, and has then fed into a great, supply chain team that then delivers that product, you know, as close to the intent as possible. Right? Mhmm. So that's how it's so special. It's because, you know, rather than going and selecting prigs, how the sensors interact with the motor and all that, that's the core of why riding a Vambov is so special. That's why we have the boost button that is so great when you use it. That's why it feels the way it does. And and just it's just a huge part of the product development that goes in there that a lot of our competitors competitors will just not do. They will not do that. They will just buy an existing system from somewhere, slightly tune it, and use that. You know? So I think that's how we can get so deep. And, of course, there's other things like choices we make around mechanical components we'll use. You mentioned the chain. You mentioned you know, that can talk the same about gear shifting and all that stuff, which participates to building that that experience in the end. But it's also true of electronics, firmware, software, like a lot of things you won't see as well.

So it it sounds like the basically, the product was amazing, but, unfortunately, the the the return rates were a little bit too high, which made it a bit difficult to keep the the company afloat. And something that you've talked a lot about is how to rebuild trust. So how did that actually affect how you thought about the go to market and the release of your of the latest model of your of your bike? Yeah.

So I think the the first thing is that, you know, you you mentioned something. Right? Like, you know, first and foremost, like, Vambov products were always, like, great from the way they felt and how people use them. And, basically, when they worked, they've always been amazing and ahead of the rest of the industry. I mean, you know, historically, they were pioneering what they were doing. Right? So some of the issues were in parts reliability being lower than expected, not catastrophic either, just a bit too like, a bit too many issues on certain products. But then that was enhanced by very low margins, so the the the bikes were sold quite cheaply for what they were. And one of the core points that's very that's not discussed a lot out there is the fact that they were also very difficult to repair. So the things that had the tendency to break were tough to access or tough to replace. And then, you know, when when you're a d two c brand, you have to do everything yourself, you should margins, and you have too many issues, and those issues are difficult to deal with. And you're the only one that can support yourself. You don't have partners effectively. Unless you get access to a lot more funding, you can't make it. So that's what happened. But what it also meant is that a lot of people, like, never had an issue. A lot of people had a great experience from the get go, are still fans. They've had many VainMoves. Or they had some issues, it wasn't catastrophic, and then the experience of the product was so great, they just kept going for it and bought new bikes. Right? You know, a big part of the audience was very happy with what we're doing, and a big part of it was just really pissed off. Right? But somehow, they still, you know, wanted the product in the first place for what it was supposed to be. And so we just realized that if we come and we had the talents and expertise to do that, if we come and we fix the issues, you know, because the first company already made a lot of effort to somewhat bridge the gap, they just had a lot more to do and they run out of time. If we came in and finished the job and then we're able to deliver that, you know, people would eventually kind of realize that that's done. That's fixed. It's now a van of products that deliver everything it shall be. I mean, it's supposed to be. But at the same time, it works perfectly. And if I have an issue, which, you know, will happen with any vehicle, I can just go somewhere not too far from my place and have it fixed very quickly and under warranty when it's relevant or, you know, at a relatively affordable cost when it's not under warranty. So that's basically what we've done. Right? Like, when I say it like that, it sounds easy. It's a tremendous endeavor. Like, it involves everything around the business, from all the development aspects, like, from all the engineering aspects, whether it's hardware or software, the supply chain, you know, fixing all that, the logistics, creating a retail network, creating an after sales network, increasing our prices in a way that still makes sense against the features we deliver. Right? Like, you know, we're too cheap, but we still have to be competitive. Right? So we did all that, and then it's kind of finally today, as I'm speaking to you, I would say that we cut there probably a year, year and a half ago. That's when everything was fixed. And then through 2025, we delivered on all these expectations, and now people were able to see it. And so today, I'm talking to you from a position where, you know, whether that's the press or our or, like, different communities, you can start to hear, okay. Vamov is back. It's now the bikes are reliable. The products are great. Like, finally. You know? If I understand correctly,

basically, overdelivering is the fastest way to build up trust. As part of your marketing, there seems to be also some emphasis on working with creators. So to what extent do you think it's important to have plan where it's unscripted and you wanna build this authenticity authenticity to build up trust, but at the same time, you want to to make sure that you hit your sales target? Well, yeah. But, look, you know, they don't necessarily go against each other. Right? Because

from from you know, you can only overdeliver if you're on if you're underpromised in the first place. Right? It's like overdelivering by default is giving someone more than they expect. So that's the first bit. It's like setting expectations to something realistic and then, you know, really beating those expectations. Every time there is a milestone or something that offers you the opportunity to do that, you hammer it down. Like, I'm doing better than you guys thought. I'm we're coming back. We're sorry. We're gonna be your best friend again, and this time you'll be able to trust us. You know? That's one thing. And effectively, as you call it, less script scripted content. I mean, you assume like media that's with proper production and stuff. Right? That's what you mean. Yeah. So that's, you know, imperative to build your brand world, in a way. Right? It allows you it gives us the control to pass around, like, the emotions we want our writers to feel when they use our products, which obviously is something that's also in the product itself. Right? It's designing the product, and we create that world around it so that people understand what we're about. And then what you call the unscripted content, which will be influencers, which will be writers, which will be, as you call it, user generated content, that's more of a trust that's more of a trust, exercise. Right? Like, can go around and create these beautiful worlds, but if it feels distant to you and you can't relate to it, like, it doesn't matter that much. It's a nice story, but it's not me. By involving our writers and people that love our products, whether they're influencers or not, they sell our writers. Right? You kind of show your audience that, you know, there's real people going through that world right now, and it exists and it's out there and it's accessible to them, and they can trust it because now it's here. Mhmm. So that's how you balance it, you know, and the user stuff is obviously a lot more opportunistic. Like, you need the people to be willing to do that, and you need the opportunities to to be able to capture those, right, to be able to create the content around that. And that's one side of it. And then, of course, the scripted stuff, as you said, is is much more of a huge creative effort and, and takes a lot of work, but brands and different companies even in the same segments. Right? Like, you'll have very successful e bike businesses,

which will have very different approaches to us just because they have a very different product and target, and they deal with a different audience, you know, that won't react the same. You were probably, like, very tempted to to use AI in different places. So first question is, do you can you share, like, some areas where you did actually use AI to, like, create more content? And also, was the this kind of trade off between, you know, the the imperative to keep it authentic, to rebuild that trust versus being able to just throw a bunch of content with AI across all platforms?

Yeah. That's that's very interesting right now. Right? Because authenticity is absolutely key to what we're doing. Right? We're restarting an incredible company and incredible brands, and it's all based on trust. Like, the only thing that broke down was trust. Right? So authenticity and transparency is one of the most important thing Vanmoof can do today. Right? Mhmm. So obviously, with the state of where AI is, which is already super impressive and advanced, like, we still run the risk of people seeing that it's not a real bike being ridden, that's not a real actor, etcetera. Right? Which in some cases might be okay, in some cases won't. And for us right now in the current state where the company is, it's not really what we want. We want real people on real bikes to make people feel like this is really it's out there. It's for you. It's achieved. Like, it's reachable. Now what's super interesting, like, few anecdotes of how we used it. I have a co cofounder called Albert Nassau, which focuses on product. He, for example, like, rare like, I would say a year ago, we had open source image, you know, generation and video generation AI running on, our own machines, and we're training that on all the historical campaigns of Amov. And we actually got to a place where that sometimes produce, like, beautiful stuff, like people riding bikes on the sea with incredible imagery, etcetera. We used you know, we we kept training it, and it's available to us. We haven't fully deployed it to the market because whatever we do with it will be something that, you know, inherently people will know it's AI, not because of the quality, just because of the context, like, what's happening with that. And we don't feel ready for this because, know, we just launched our new product, and it's about trust, again, authenticity. But that's one thing we've done that's still experimental, but, like, it's definitely getting to a place where, honestly, if if you you put the effort behind it, you can actually create some, you know, high quality, beautiful contents with, current image generators, like open source stuff that you can run on your past campaigns and so on. The second thing is we'll be much more basic, especially to a guy like you. And so some text generation that we then modify, like, just to accelerate what our people do. Right? Copy. That it's the most relevant right now. Also in research, of course, although that's pretty standard. Research and strategy and so on, it's a great support. Like, it's basically an enhancer for, like, for, like, thoughts, you know, for, like, the way you think. There's other areas of the business where it's much more active, like customer support, like many companies, you know, they have a very robust strong layer that solves, like, 95% of our cases, which is huge, especially for a technical product like us. For 5%, of course, it's escalated to an agent and we help them. But off the top of my head, you know, company wise company wise, that's probably the two most relevant ones, and then we'll have you know, I can tell you about how I use it personally and how I push it to a different place in the company, but we can talk about it being more and more relevant. Yeah. So my follow-up question was gonna be around your take on AI from a productivity standpoint and marketing standpoint, and I think you've answered both.

So, like, image generation as part of the marketing campaign, but also, like, with customer support and so on. But maybe just to dive a little bit deeper into the product, do you see AI as and GenAI more specifically as a tool that could potentially help as part of the core experience with the product?

Yeah. Look. I mean, you know, it's becoming so rich and complex. Like, it's it's it's it's nuanced, the answer to that question. Right? In a way, like, I personally use AI. Like, I have an assistant with me that's, you know, very well educated and helping you with a lot of things, whether it's finance analysis, deep research, online, reflecting on strategy, like, coming up with a lot of ideas, like, also finding people and reaching out to the right people to do stuff. You know, it's becoming so wide. So if you apply that to, like, a product development program and a product development team, there's so many points of impact where at least it enhances productivity. Sometimes it will lead to more creativity that, like, it's getting more and more important, the impact. Right? So something more tangible that actually might be relevant is now you will have some supply chain tooling and planning AI. You will have some electronics design AI that, you know, links up with, like, suppliers and components. You will have, obviously, on software, like, we don't need to talk about it. It's becoming, you know, for one developer, you get three in the sense that one of your developers gets to the output of three just because of, the stack, sorry, the support of different, of different tools. There's still limitations there. Mainly, the limitations will be that when you have an existing complex sophisticated stack with its own rules, then season integrations, it's much harder than when you start from zero to have, like, support from those kind of tools, but it still helps you, like, accelerate and ideate. So there is no exact tool from the top of my head that I can tell you yet that one, like, made product development 20 times this year, but that's out there. It's just feeding into, like, increasing improving productivity,

design velocity, and all that. Coming back to what you mentioned earlier around the competitive advantage where with AI, basically, everything becomes more accessible. How do you think about your competitive advantage,

you know, in with with within the context of AI? Yeah. So, look, I think, like, the the first thing is, you know, there's like a ladder of of complexity you have to look at. Right? So, you know, for any for brands and companies, whatever the industry, whatever the product like that, competing on the lower end, I think that's gonna be harder. You know, it's probably gonna be a lot more based on brands, feelings, specific use cases, and things like that. Just because at lower levels of complexity, it's easier to outsource most of the work to, like, a supportive AI in a small team. Right? Mhmm. In our case, like, you know, we've got deeply complex, deeply sophisticated integrated products. You know? Like, if if you draw a circle in your head, it starts from, like, industrial design that integrates electromechanical components that then connect to, like, you know, precision and power electronics that then connect to back end services that then feeds many different tools, whether it's, you know, front end on web, like an app, another app for quality control for our retailers and app for our writers, and then all the things that stem out of this. Like, it's such a profound, rich world of technology that, you know, I think at the higher level of complexity, like, you know, you can still with a great talented engineering team, like, really differentiates. Right? And so us today, we're delivering products and features that are really, really good, that feel really, really great to our users, that deliver new features, like, every month that are really impressive. There's very few companies in our industry that are able to deliver to the same level in terms of writing performance and tech. And I think that advantage is here to stay for a while because, like, when integration and, like, play such a role into delivering your output, like Mhmm. Obviously, it's more challenging for for, you know, like, it's more challenging to find ways to do that very simply, you know, or very quickly or or leveraging other tools to go faster. Right? Like Mhmm. The step of creativity, and system engineering and integration, all that stuff takes a lot more time. Right? I think in the end, at least the state of it today is getting to fast outputs, fast ideas where you don't need that much integration and it's much more flat as a product rather than layered in terms of what works inside of it, then, you know, AI can have a tremendous impact, and it's tougher to differentiate. But when you have these super deep layered stacks of technologies working on top of each other to deliver complex features, like, we still have a few years before, you know Yeah.

The the technical advantage gets erased a little bit. Yeah. In terms of layers, you recently added a new layer where you used to sell directly with VanMoof owned stores, and now you're moving away from that using indirect channels. What's your what's the biggest challenge in your opinion when you onboard new partners who are so essential to to be an extension of your brand, an extension of the after service, and so on? So how important it is for you to find the right partners to work with, to train them, to make sure that they are communicating in the right way about your product. Well,

I I could write a book on that, lad. So I'm gonna try to I'm gonna try to sum it up and and brief it up and contextualize it to us. You know? So first of all, we were the, like, child image of the d two c brand that doesn't like retailers. And vice versa, retailers didn't like us at all. Right? Because we were, in parenthesis, the guys trying to prove that, you know, they're not essential in the industry. Obviously, we failed for the right reasons because you need those guys. You need the local bike dealers to sell locally, to support your customers locally. One thing that's very important is, like, an e bike or an electric scooter as well is much more akin to a car than a normal bicycle, Right? In terms of complexity and systems and after sales and maintenance and the way it's used, like any bike needs the touch points and support and after sales than a car or motorcycle might get. Right? So the local dealer network is essential to delivering that basically around your product ecosystem. So that's something the company realized, in the most difficult way. So to so we had to come back to market and be like, guys, we saw we were wrong. We wanna work with you. We had to engineer contracts and financial agreements that were very advantageous for them. I mean, against our competitors that obviously might have put more constraints on those guys because they were bigger and did more volume, etcetera. So we did great contracts that are easy to work with, easy to understand. That's one way that, I would say, the entry door. The second thing, our product is great. It's obviously one of the best in the market, if not the best for our specific segment. So, you know, you come in, you're like, you know, I'm giving you this also. You know, you're gonna significantly enhance your sales. We brought new customers that they didn't necessarily have access to, customers that were not necessarily the type to go to a a bike store. They were gonna go online. They were looking at tech companies. It was a different approach. So we kind of, you know, sent a lot of those people their way. So they were happy about that. Mhmm. And then we created an entire ecosystem behind the retailers, are training apps, training documentation, diagnostics app that actually also allows them to, you know, all the parts, components, gives them guidance on replacing them, give them guidance on how to identify issue issues, etcetera. So we really made it easier to deal with what makes our product special, which is their technical sophistication, which is what delivers a lot of the value to customers. So we bridge the gap of complexity to make it easy for them to adopt it and to go into that ecosystem. And I think, you know, all in all, it's a lot about the narrative. In the end, narrative is where you're gonna 70% of this because, you know, the the a retailer is a customer. Right? It's a guy in a workshop or that owns a store that will have the same feelings as, you know, what other people might have about the brand. So a huge because we do a lot of studies around that. We go around our stores to understand the ones that sell amazing, why do they sell amazing compared to the ones that don't sell so well. And the biggest single thing that matters is how much do they believe where we're at. Right? When we tell them guys, the bikes are the most reliable they've ever been. We're, like, above industry standards. We can support you. We've got spare parts. We have all these features that you're not used to sell, but, you know, learn about that. Like, all these anti theft tech, the boost button, you know, the integration, the aesthetics of our products, all stuff that your customers want. But maybe you're not used to selling to them because you didn't have access to that before, but now we can give it to you. And so when that whole narrative comes together and they stop believing that, then they stop, you know, transmitting that to the end customers, and that's when things really shoot up. Right? Mhmm. So in general, for me, the core core challenge is that one. It's having the mechanics, the sales guys, the store owner to believe in our story so much that they then become your ambassadors. They will then shout it out to everyone that comes in the store. Right? That takes a bit of time. As you've measured it, it's two to three quarters for them to to start to to stuff with us and then, you know, actually, believe in us and really increase their performance. But, it's it's yeah. We have a full team on this. Most of the company's strategy is based on enhancing that. So it's not something easy that you just start and it works. You know? It's very difficult, actually.

Yeah. So it's interesting. So you were saying 70% is around basically marketing building that story and narrative for them, but there is also something around making their life simpler Yeah. Around AI or sorry. AI. I'm a bit biased, but around technology. To Yeah.

But you know what? It's everything you're asking me, which, by the way, is is a nice discussion because everything is connecting. Right? Like, I mean, everything I'm telling you about narrative and so on is basically the same as over delivering. Right? I'm I'm telling them I'm good at this and that and that. They're like, okay. Yeah. Why would you not say that makes sense? And I'm coming and through this ecosystem that you just mentioned that I created and all the things behind it, we're then showing, yeah. Actually, I am the way I'm telling you I am. And we can really help you sell 10 times better now. And we can give you access to something you've never had the opportunity to have before. Look, and that's what's happening. Right? So it's all working together. You know? You the product, the marketing, the branding, the operations,

everything has to help each other. Yep. And and maybe the full circle moment is going back to your team within VanMoof. How do you think your team is gonna evolve considering, you know, everything that is happening right now in terms of productivity? Do you have a vision already about where you want to like, how you want your team to look like in terms of, you know, skill sets, tools that they should use? Do you think about these topics? So a big part of what I'm doing with Vamos, like, kind of forced

me and us actually to, like, think about it very early on. Right? So when you restructure a business, the first thing you do is, okay. What is the best possible structure for that company to be successful? Mhmm. And then you have to constantly think about that, right, all the time. Because it's easy to, like, reset to a place and then let things creep. Right? So once you've made all the work to be in the right place, you wanna make sure, like, you're on it so that it doesn't creep away, so you have to do again things, like resizing and so on, which you don't wanna do. Mhmm. So, actually, I think today, like, our team is extremely talent dense. Like, everyone in every, like, position is someone that's very good at what they do and very talented at what they do. Mhmm. Enough so that it's kind of a company trend that people, for their own self development and to enhance their own productivity, are embracing whatever new technologies are available to support their work and growth. So today, like, the the way the company is for what it does today is, I think, quite optimal. Mhmm. Whether it's talent or number of people, whatever locations, etcetera. I think, you know, in time as we're growing quite fast in terms of volumes and deliveries, we're gonna have to, of course, adapt some sectors, like, of course, you know, technical technical warranty services and a lot of the operations side, like, that's linked to real world deliveries and real world interactions. Like that, you have to scale effectively to make sure that your service level stays the same for your users. But then there's a lot of things that actually, you know, you you don't have to scale them as much as you used to to keep the level of delivery you currently have. Right? Mhmm. And that's the interesting thing. Like, I I can't really I mean, I'm I'm not sure if you can, but I can't really say yet, you know, in a year or two where we're gonna be because things are progressing so fast. Like, you know, it's it's tough to know, how you can adapt to this kind of fast, fast, fast progress. But for now, you know, I would say looking at the things the way they are for now to the next six months, I go in a very good place.

Yep. Yeah. It's funny because the in AI, the closer you are to innovation, the more difficult it feels to have an informed opinion about it because it's all progressing so fast. And people who are the most opinionated are those who are actually not using AI.

That's kind of the the dichotomy, I think, about it. Which is very common in, like, you know, in everything.

Yeah. Yeah. No. That's true. That's true. So but but so maybe if we were assuming that tech's about to remain the same, where do you think today it's kind of falling short? I think today, as a research tool, it's incredible.

You know? We've never had anything like that before. Mhmm. It's accelerated it like crazy. It's really good around, you know, copywriting, copy things that are linked around language. It can be quite good at image generation and everything that's media related, even sound noises, voices, stuff like that today. Like, probably good enough to deliver professional outputs, like, if you really train it for it and, you know, use it in that direction. But you've gotta be ready to deal with the fact that people will know. Right? At some point, they will know that you're using AI. And in some cases, it'll be great. Some cases, like, they'll be like, dude, I don't wanna see that. I wanna see the real stuff. Yeah. Where I think it's not good enough yet is I don't trust it yet 100%, not because of me, because you still get random errors. Right? Mhmm. You still get little mistakes, little inaccuracies. It will feel that you know, it'll be very good at doing a very thorough analysis of an Excel document and then creating you, like, some kind of other model that satisfies really well what you wanted it to do. But, also, you might fill in an equally complex document right after that, and then it will all all look amazing, but two, three numbers, two, three calculations are really not done properly. You don't really see it, and it's can have dramatic impact on your plans if, you know, you're not you're not working with the right data. So I think on accuracy and the certainty of the output, it's still not there. So much so that you still kind of have to review everything, especially around complex decisions and complex data that you really need to help yourself. Like, for example, it's difficult to trust a fully a full AI flow on anything that has to deliver, like, operational outputs or financial outputs or take really strategic decisions that have to be right every time or as right every time as possible against the data we currently have. Right? It will still hallucinate. It will still do some errors, some mistakes, and sometimes it'll be perfect, and that's the trick. Like, it can be perfect 80% of the time. And then the, you know, eighty first time, it's gonna give you a bug that then screws everything off. So that's where I don't think it's good enough yet. It's like I can't trust it yet like you could a human. Right? That is trained and and can double check stuff. It's not great yet at the double checking and verifying its output and making sure it's is there and, actually, you can see a lot of companies are starting to be born around that. Right? Error correction and quality assurance and all the stuff around AI. Right? I think. Yeah. Yeah. I think there are, like, two tracks happening at the same time. One is about

reducing the likelihood of the of AI to fail. And then there is the second, which is more about organizational changes on how can you create a company that is stochastic first, where basically, like, you factor in the fact that AI isn't predictable, and you have to, like Yeah. Shape your operations accordingly.

Exactly. And I think, you know, there's probably a lot of business that can do that, by the way. A lot of businesses, sorry, that can do that, and there's some that can't. You know? If we have let's say, if we integrate AI in important design decisions, design and validation decisions, and it's right 86% of the time Mhmm. On complex design or mathematical or whatever models, like, this will kill us. You know what I mean? Like, in the final output, I need my bike to be consistent within very tight tolerances and a lot of certainty, like, 99.9% of the time. Right? Mhmm. And the 0.1%, I need to be able to catch it either at my factory or before I deliver it to my consumers. Because in my case, it can be a life or death situation. Right? Mhmm. So some businesses, I would say real world hardware ones, or those that have an impact on the real world, like financial decisions, like, you know, accounting, taxes, financial modeling, you know, all that stuff, like, you can't really do mistakes. You know? Yeah. It has to be the truth. And for that, it's tough to to adapt. Other businesses, which I'm sure you'll know many more than me, I'm sure can can work on the stochastic Yeah. Approach.

Yeah. For for me, when I think about AGI, I'm like, okay. Am I ready to let AI fill out fill out the taxes for me? And as long as I don't feel comfortable with that, I'm like, I don't get a direct is view. Literally.

You know? Because it's all about your downside. You know? Mhmm. If I let you do it, what's the worst that's gonna happen? You know? Mhmm. Frauds, someone will die on the products. I'll kill the company. You won't do it. Yep. The upside is always gonna be amazing. You know, I save hours and hours, like, on some projects and stuff Mhmm. Leveraging AI, but I always have to double check it. I always have to make sure, like, the output's consistent, things like this. We're not in a place yet where I'm like, oh, great. Done. Like, take it and go. Yeah. You

mentioned at the beginning that you're basically using it for, like, deep research and so on. Are you excited about being, like, an IC again in the sense that you're actually involved in, like, producing, contributing to to things and not only managing? Because I guess when you were a CEO, you're probably involved in steering, but now you can actually go back to actually doing,

which as an entrepreneur is probably quite exciting. So how excited are you about this, and how do you think about those topics? Yeah. I mean, like, Luke, that's a great question. I think I'm a bit of a of a specific case in in in that I'm very technical to start with. You know? And in the company, you know, I'm kind of CEO, CTO as well. Right? So I am still quite involved in, like, strategy, you know, decisions around tech and specific product development things and so on. However, you're right. Like, I don't build much. Right? I don't I haven't built in a long time because, you know, I've got much much better people than me to do it. Why would I do it? Right? But I don't think it's allowing me to do things I wouldn't be able to do otherwise, but it's really tangibly accelerating, like, you know, the strategic side of things, the ability to review stuff quickly, to be a little bit more involved in the review thing without being overbearing, like, just double checking things, running different things at the same time. Like, an example now, like, you know, there's a I mean, Clotbot or OpenClot OpenClot. Sorry. Yeah. Like, I'm currently integrating in my own workflow now to, like, accelerate a lot of things that otherwise it's not even that they would take time because I couldn't do them. I just couldn't do them. I don't have time to do them. It was something I had to push back in time. Right? Whereas now I'm gonna be able to link different things to deliver, like, different types of information, recurringly much better. Right? So Mhmm. Mhmm. At that level, it has an impact. You know? I think, interestingly, you know, for very strategic roles and so on, like, it's just helping you go faster and deeper. Mhmm. Like, and just be a lot more efficient. But I don't think it's necessarily helping you build anyways. Right? Because if you've got an existing you know, if I was starting from zero tomorrow, 100%, like, I'd be able to do everything much better. But today, I don't feel like it's in a space where it's allowing me to do that at the new level yet. Yeah. Interesting. That's why we talk about AI native companies

where, basically, when you build from a blank slate, it's a lot easier to integrate AI. Whereas for larger companies, there's more legacy, more complex stack systems,

and more complex technology. So it makes it harder. The main limiter to us adopting AI to the full extent we could is literally legacy systems. Mhmm. Right? We can't, especially on the software sides. It's very difficult because we have so many active systems in production, delivering stuff all day, running all day. It's very tough to, like, fully leverage, what's available now to just create code and replace modules and, you know, integrate what we're doing just because, like, you know, there's an existing integration and an existing, you know, get get a certain way that it's very difficult to plug into without a human. Like, you know, so for us, it's not relevant yet in that. I just have one last question for you. Is there any specific conviction that you've understood in the past ten years that helps you in your decision making every day? Yeah. Yeah. So I think there's two or three. Like, first one is thinking from fundamental principles. Mhmm. Like, it's very important because it allows you not to be overwhelmed by problem and complexity by problems complexity. So when something comes at you, like building the reflex to, like, cut it down in smaller approachable problems, thinking where they come from, and looking at the fundamentals also allows you to see things you wouldn't see otherwise. Right? So Mhmm. Thinking from fundamental principles is, I think, paramount. The second thing is resilience. You know? Like, it never goes well in parenthesis. Like, it's always up and down, up and down, up and down, up and down. And seeing through this and remaining nonemotional, actually, by seeing through that, it really helps, is probably one of the most profound things for longevity and taking the right decisions and going the right way and just keeping on going. You know? Mhmm. Being able to adapt and react quickly, etcetera, and being resilient and not stopping, not abandoning unless you really have no choice. So those are probably the two biggest takeaways. Can people find you somewhere? I know I can share your LinkedIn, or is there any place where you talk and share some thoughts that people can LinkedIn is the best for me personally. Like, as well post, like, interesting learnings and everything on the, you know, on the company, what's happening, where we add school development, sometimes business insights, and how we've done something that people might be interested in. So I think LinkedIn is for sure the the best platform.

Okay. Perfect. Thanks, Elliot. Thanks so much for for listening, everyone, and I'll see you next week for more content about how to run companies with AI.