FuturePrint Podcast
FuturePrint is dedicated to and passionate about the power of print technology to enable new opportunities and create new value. This pod features deep-dive discussions with the people behind the tech as well as market analysis, trends, marketing and storytelling!
FuturePrint Podcast
#287 - Rethinking AI: How Manufacturers Can Turn Hype into Real Productivity
In this episode of the FuturePrint Podcast, Marcus Timson speaks with David Rochholz, consultant at Netlight and keynote speaker at the AI for Industrial Print Conference at FuturePrint Industrial Print in Munich. With a background in computer science and machine learning, Roccholz helps some of Europe’s largest organisations turn AI from a vague concept into practical business value.
David explains how the evolution of AI — from early “big data” to today’s generative tools — has created both excitement and confusion. His core message is clear: before thinking about AI technology, companies must first understand their value chain. Only by examining the specific steps that create value can organisations identify where AI can augment, automate or accelerate their workflows.
He shares real-world examples, including a case in aviation where AI dramatically improved aircraft ground-time planning. The human stayed in control, while AI acted as an “exoskeleton” that amplified human performance rather than replacing it.
For industrial print, David argues the applications are vast: job classification, production scheduling, workflow optimisation, data clean-up, error detection and more. But AI also reveals uncomfortable truths about digital maturity. If data is inconsistent or fragmented, AI will expose the flaws instantly.
He also tackles the common misconception that AI simply removes jobs. Instead, he says, AI removes repetitive tasks — freeing people to focus on creativity, innovation and problem-solving.
For those beginning their AI journey, David recommends something simple: start experimenting. Play with ChatGPT, try tools like Lovable, observe the possibilities, and then begin imagining what happens when such tools are fed your company’s data automatically.
This is a grounded, practical, and inspiring conversation that sets the stage for David’s talk in Munich — and a must-listen for anyone serious about AI’s impact on manufacturing.
Listen on:
Apple Podcast
Google Podcast
Spotify
What is FuturePrint?
FuturePrint is a digital and in person platform and community dedicated to future print technology. Over 20,000 people per month read our articles, listen to our podcasts, view our TV features, click on our e-newsletters and attend our in-person and virtual events.
We hope to see you at one of our future in-person events:
FuturePrint TECH: Industrial Print: 21-22 January '26, Munich, Germany
Welcome to the latest episode of the Future Print podcast. I'm really excited to have with me today a gentleman who is going to be speaking at our AI for industrial print conference at FuturePrint, Industrial Print in Munich. Um, welcome to the podcast, David Rocholz from NetLight. Welcome, David.
SPEAKER_00:Thanks a lot. Hi, Marcos. Thanks for having me.
SPEAKER_01:And we're gonna have a really interesting conversation, not least because there is a relevance to printing, but the focus of the conference and also this discussion will be around AI and its exciting development and contribution to the world. Um before we get into the discussion around technology and and everything else that goes with it, David, would you mind giving us uh an introduction about yourself, your background, and the kind of work that you do with NetLite?
SPEAKER_00:Okay, okay, this will be will be interesting. I try. So I would say in the work context, uh, like I always start with, hey, I'm David, I'm a consultant. Um, and this is already very hard, uh I guess I think to understand what this actually means. But then I start, I have a computer science background since I can remember. I started programming like when I was 10 or 11 with websites, and somehow did go into it, started computer science then in uh bachelor and masters, and during my masters, I uh found my interest for machine learning, basically. Like now we would call it AI mostly and in this hype cycle. But back in the day it was mostly data science, machine learning. And uh there it was mostly about I started my first job in finding out whether doctors would repay their debt and then make a scoring about them. So you can pack them into a debt package and see how much of this debt package I would get back or not. So this was, I would say, my first encounter with probabilistic thinking and not deterministic thinking. Like probably we will talk a bit about this later. And uh in the in the connection to NetLite, which uh where I started later as a tech consultant, I did first the same: like data engineering, data science, building predictions for the aviation industry, building um like ground time models, like when an airplane lands. Like, what do I have to do on this airplane? And what can I do with the tools available, with the personal available at a certain airpoint if I have this window of time where the where the airplane is on ground? And I would say also like the consulting NetLet does go into this direction. We would we would classify ourselves as an innovation consultancy, going more in the classical IT direction. So meaning all of the systems you have in your in your company who are running your operations, but we're coming from a space that is where I really like okay, I want to have a new product, new digital product, how can this help me to get new customers and how do we realize it in strategy, but also in tech?
SPEAKER_01:Right. Yeah, so you've come from a tech technical way through, like you say, the development with machine learning, the understanding, and so on. And now really, are you the I I would imagine the intersection between helping manufacturers understand how to integrate AI and and new technology for improving productivity and so on? So we'll get into a bit more about that now, uh I assume. And w we've obviously talked a little around the the um the integration and adoption of of AI. You mentioned there in your introduction the hype cycle. And and and that's I think fascinating. So it's is it Gartner's hype cycle uh that is used as a sort of reference point for the adoption of new technologies. Explain a bit about the hype cycle, perhaps, and where you see AI being on that. That that would be a nice context to start, I think.
SPEAKER_00:Yeah, there's uh it's a nice context. Uh I didn't mean the Gartner hype cycle, even though I really like the Garden Hype Cycle. It's uh it's a very, very nice, uh, nice tool. Um, when I think about AI as far as I can remember, at least, or when it became relevant for me, like we had it in 2009-2010, like the first wave in my head, where it was mainly about big data. So everything was big data, and companies started to collect data because this is the new gold, the oil. And then when I graduated in 2015, it was more like data science, like there was this idea, at least in my industry, like the data scientist is the most sexiest job of the future. And I thought, okay, let's do that. That sounds good. Yeah. Um, and there it was really about companies stacking up data scientists to figure out, hey, how can what can we do with them, basically, and how can we improve our business with this machine learning, with this data science. And then it became quiet again uh for a while in 2017, 2018. Like the if you would say the Gartner hype cycle came into the valley, and companies realized, okay, this is really what I can do with it. It's not exciting anymore, but it's already an improvement. And then you didn't hear anything in 2022. OpenAI released ChatGPT, and then it was again where we are now in this big rise up. Everybody can see the potential. You have OpenAI being downloaded or ChatGPT being downloaded so many million times, and people see the value, but they can't yet really map it on the business. Like, okay, I can prompt this thing and it tells me something, but what do I actually do with that? And maybe to come back to my background, obviously I started as a machine learning engineer, but what I do now with my clients is to find out how to take this new technology and apply it to the business domain of the clients. So, what does it mean in aviation? What does it mean in real estate? What does it mean in finance, for example, and what does it mean in the in the shop floor, basically?
SPEAKER_01:Yeah, and uh it's helping them fathom and understand the strategic point of AI rather than just using it on a personal or individual level, how it can affect and improve things across the business.
SPEAKER_00:Yes. It's uh when you think about bringing AI into the business, you you decouple a bit from the idea, like I said, uh you have this chat window, and then you prompt it something and gives you something beautiful back, and you can use it, and you you at some point elevate to the idea, okay, how can now everybody in the organization use this? This is usually where it starts. Do we get Microsoft Copilot in, or do we have other solutions? And how do we do this? And then how do we go away from I always have to prompt it, like exactly what I mean, and go really into the direction of automation, and from there into the direction of agents. So when I when I look at our client base, for example, I would say at this at this agent game, there's not a lot of companies yet who really run many things automated, but everybody I think now understands it a bit more and and gets what the what the potential is, and then asks questions, but how do we do it? And this is where I come in with my consulting. It's um it's really more like more than saying I want to have open AI in my company, it's really the question uh how will people start using this? Because now they already do stuff. Like if you go into an organization today, usually it's more like everybody has already 120% of stuff to do. And the question is how do you get them to experiment with something new that's a paradigm shifter?
SPEAKER_01:Yeah, yeah, interesting. And so you will work with a variety of clients. Um thinking of manufacturing now, um, the title of your talk really is around how to even think about AI, isn't it? Because I think like you've just said, a lot of us are playing with it, we're using it, we're whatever. Tell us a bit about that, that that how to think about AI before you even go near any platforms or near any technology. What advice would you give perhaps people listening from a strategic viewpoint? Is how do you think about AI?
SPEAKER_00:Maybe to go into that, because what I see with my clients happening is they're getting approached by all of these solutions. So I think it's more than actually going all that into AI, it's actually a bit of managing chaos because every every unit in your business probably at the moment gets um uh gets contacted by somebody who has an AI solution, wants to bring it, and sees how it could bring value to a certain part of your business. And I think in the first step to organize this a bit, like is one of the most important steps, and then really how to think about it. Like how I love to think about AI when I when I think about a holistic organization is everything an organization does should, or everything an organization does to create value, to specify a bit more, should happen inside of the organization's value chain. So in the best case, when I go to a client, they have a very, very good understanding already of we do these things, these things support our main primary primary things that create value in order to bring our client something or our customer. And when you have this view, you can start to zoom in and really hypothesize of okay, now inside of this value chains, we have different value streams. And if you zoom in more, like Google Maps, for example, we have different zoom levels. And if I go into that, then I could bring in a specific AI solution, for example, for this piece of the business. And this is maybe on the shop floor, but this is maybe also on classifying emails, because a lot of data is getting shoved around in emails and Exils that are attached to emails. And then it's really about making a pre-analysis, for example. For example, if you get an order or anything, like a qualitative assessment. And then you collect these hypotheses and say, okay, now assume I level out again. What does this mean for the whole value stream? And now I assume out again and say, what does this mean for the part of the value chain? So this is the strategic approach, um, how I like to go at it, because then it becomes very clear, okay, this is what it really means. AI can be super fluffy. Like this is one of the problems. Like you can solve everything with AI. And usually when stuff is fluffy, you need to break it down. And broken down things, they don't sound fluffy anymore. They sound very banal, I would even say, you know, they'll sound really boring in some cases. But this is the this is the level I think, how companies should start to think about it.
SPEAKER_01:Yeah, and and the more you're saying that, the more it makes complete sense. It's having that human thought at the beginning. Yeah. Because many people have said to me that AI is fantastic, but it is only fantastic if it has quality input. Human input at the beginning. Otherwise, like you say, it can do everything and anything, but that becomes a a potential giant mess as well, doesn't it?
SPEAKER_00:Yes, and these are two different things. Like human input and qualitative input are different in the way, not of saying like a human can good give good quality or not. It's usually what uh what I see with with AI currently happening is like a lot of organizations getting aware of, like zooming back to what happened in 2009-2010 to this big data, the data we're actually having is not that good. Look, you can't really drive automation on top of that, because for example, we use our CIM system for our leads and our customers, but we don't really fill out like all of the fields we would need to automate the things our people are doing, or I do, so I need to do less administrative stuff and can like focus on creating more value or better value. It's then really about okay, for example, now we have to uh like close the the holes, yeah. And AI is like brutally honest at showing you this when you when you have not good data.
SPEAKER_01:Yeah, yeah, yeah. And that that's that's critical, isn't it? Um have you done sort of obviously I know there's sensitivity because you're working with clients, and I I imagine many don't allow you to to give specific examples, but is there any sort of examples of where you've or projects that you've worked on in perhaps more on the manufacturing side of things that have delivered some good results?
SPEAKER_00:Yeah, like I would say, like in in 100% of the cases, I would say like every every project I did delivered uh good results or solid results. Um It depends a lot on where you look, for example, for applying AI, because like this could be an assessment, for example. Like a lot of clients at the moment are saying, Hey, I really want to get used to it. I really want uh to look into into my factories and make an assessment of where AI can help us. And then there's the ones who are saying, Okay, can you can you provide already a solution? And here's this one project, for example, when we spoke about the ground times. And I think this is a very good example.
SPEAKER_01:This is an aviation example, yeah?
SPEAKER_00:This is an aviation example. Um, but when we talk there about the ground times, like obviously the help these tools can bring, like can cut down like planning times, for example, of workers easily by 50% or more. You know? But what's very, very crucial is and how AI is approached a lot of times, is I see CEOs, CTOs, CEOs going in and saying, hey, if we bring this AI solution, we don't need any humans anymore. And since I work at least with AI, I've never seen this being the case. Like I always, or what I always see is the humans, you you can really like it's an exoskeleton, it amplifies what you're doing. It's like a battery for your bicycle, you know. You still have to uh to to spin the the wheels, let's say, you still have to apply force, but it gives like an extra kick of making you faster and maybe reaching your your goal a bit faster. And this is something that usually happens is okay, we don't need people anymore. Now we go in and say we built the solution that doesn't need people anymore, and then like nobody will will start to help you to build the solution internally, you know, like um at these businesses. So, yeah, with the groundtime planning, like um it took a while because you're also not kicking off an AI project from like zero to 120%. Like I think the development time in the end was far beyond two years. Um, but the the goals were like planners could see, okay, this is the specific things I need to focus on. For example, this is the specific tools I have at an airport now, and this is like what this thing would suggest to me. From my experience, does this still make sense? Like the human in the loop component. Yeah.
SPEAKER_01:And I I I like that because that's the immediate uh threat that uh people will lose jobs and and so on and so forth. And and I think to some extent that's happening, but that's equally uh an economic fact, not just AI taking roles, is it? But I like what you're saying there is about supercharging performance as opposed to replacing people.
SPEAKER_00:Yeah. And it's also thinking about jobs like is this a programmable task? Is this thing I'm doing every day basically something where I don't I don't know how to express best, but maybe where I don't have to think a lot, where I take one thing and I put it here. And then uh do X, Y, Z in the in the uh in the in between. Like I would classify this without so uh without sounding too arrogant as something that's probably I don't want to do. Like I want to do something more impactful, something that creates like maybe more value for somebody else, you know. Uh and with this I don't always mean money, because for example, with AI, we also work with the World Food Program. Uh that's not about money, that's really about helping people, about scheduling tasks. And like yeah, obviously, if you're if your work is highly programmable, then you can be replaced. And then the question is how many of these tasks are actually 100% uh programmable and how many are not.
SPEAKER_01:Yeah, and it's liberating people to do the more creative or more human things that um will then improve productivity and and general um general well-being as well. I mean, so I think one of one of the classic problems we all have is the fact we have to multitask across so many different things. Yes. Because the working world has changed, and if that can liberate some of the more menial tasks, then surely that's a positive. Um, yeah, so look looking looking ahead really with with Munich in mind, I I I like the theme about how to think about AI and going through those processes and that that aviation example is superb, and um it's just thinking perhaps around printing, David, and and and it whether it's printing and packaging or automotive or any of the industrial areas, um, I know it's not the printing, and it's part of the reason we're really happy to have you among a couple a few of our other speakers, not from the printing industry per se, but actually from outside. How might you envisage AI playing a role perhaps in in the world of printing and and and given given what you know around it? It's a manufacturing sector. How how might you see that perhaps play out?
SPEAKER_00:Very good question. Like also in in uh preparing for my talk and preparing also for this podcast, uh I spoke with friends actually from mine in the printing industry. So and like very unsatisfying answer, I think also for the listeners, but we can apply it everywhere. That's a bit like the case. Like you can apply it in I'm getting an order, you can apply it in okay, now how to really play this out to the print machines on my shop floor, yeah, and everything in between. Um I think how it will play out, because this is what I'm seeing everywhere. Again, decoupling from printing, but this is also highly relevant for printing. What you see when you start to apply AI is you might have holes in your already like progress digital transformation. And when I speak about digital transformation, I talk about or what I have in my head is you have an organization, and then you have a reflection of this organization in the digital space, like a reflection in the water or a reflection in the mirror. But you look at this reflection and you see, okay, there's certain parts not there, you know, and the digital transformation of an organization always does what you do, but in a digital way. This means, for example, ERP system and connecting it to a CIM system. So data can flow between these two systems. And for example, you're getting an order, and now you can play it out into the warehouse or into the into the shop floor to do printing of t-shirts, of shields, of paper. Yeah. And you will see that you have holes there. So one step will be to close the holes while you already sink, because like uh printing, as I understand, is a very low margin business. So you want to go in and really, really already apply AI because you want to increase the margins a bit. And this is something probably companies have to do at the same time. And then what I see in the digital transformation space, and probably also highly relevant in printing, either you form alliances, because also AI requires a lot of investment. It's not something that comes for to you for free. Yes, sure, if you use ChatGPT, one license,$20, yes. But usually to uh understand the value of a technology, you need to invest. To understand the value of digital transformation, you need to invest. And then the idea is to increase the productivity more than increasing the efficiency. Because if your assa cake already has a or has a certain size, then seeing that you eat more of that cake, okay, can work for a while. But I think companies should think about how to make the cake for themselves a bit bigger and maybe the overall, if this if this uh makes sense. No.
SPEAKER_01:Yeah. I think I think what you're saying there, productivity is also about creating value. Productivity is creating value, but we're thinking about productivity.
SPEAKER_00:Yeah.
SPEAKER_01:And you're right, they often get the word efficiency or productivity are often used interchangeably, but actually they're fundamentally efficiently. But it isn't it's almost internal, isn't it? Productivity is external and and vacuum. Is that right?
SPEAKER_00:Efficiency when I talk about it, is about I have something today. Like I have a revenue, for example. Let's break it down to the revenue layer. I have a revenue today, and there's a profit inside of that revenue. And I look that I make everything in my company better in the way that the profit gets higher. So this means like actually this is that this is preserving the status quo and making the status quo better. And then productivity is increasing the revenue, increasing the status quo because you do new stuff. You find new ways to create value, you have new business models. Think about like digitalization again, like digital transformation, like the things that are possible today because the internet are like arrived. Like Netflix, Spotify, all of these things would not have been possible without the internet if we would have stood on LPs and CDs and so on. So the internet was a huge driver of digitalization. And now you have AI being again a driver, not necessarily digital transformation, but being a driver for digital transformation, because now things are possible that were not possible without having AI.
SPEAKER_01:Yeah, really interesting um point there about the fact it's about it's also about unlocking new potential, new revenue, and also a huge way of um helping people innovate, right? Yes, fail faster, get success quicker, and understand things um and just as a basic research tool as well. Really interesting stuff. Just just uh a final um uh question for you, David. Any advice sort of gif if you're gonna leave someone with a listening with a key bit of advice about AI and strategic integration, what what might that be for somebody listening that knows they want to get more into AI and perhaps use AI more? What what key piece of advice would you give them?
SPEAKER_00:Very good question. And it's so different from where you are standing, I would say. If you're standing like right at the beginning and you have no idea you just see it on your side.
SPEAKER_01:Yeah, maybe for that person, the person at the person at the beginning.
SPEAKER_00:And you're like, oh my god, what what do I do now? Like go now while you're listening to this to ChatGPT or to Claude and just try what it does, you know. And after that, maybe just for the fun of it, go to Lovable. Like it's uh it's a it's an app from Stockholm, I think, um, where you can create websites and do these two things, you know. Just by prompting, you can say, I want to have a website that looks like this, probably also highly relevant when you think about uh design for print. Uh I can do XYZ, uh print on demand, for example. Uh I can do uh like give me a website that does XYZ and then do a few prompts with OpenAI. Okay, probably you will be impressed if you never did it before. And then second step, think about okay, what does this mean if I don't have to sit there to type in the words? You know, because then we talk about automation. Then we talk about now these things getting fed by the data inside of your organization to do a bit of it automatically. Problem is the result is not deterministic, it's probabilistic, so it has a probability of being right or wrong, like put very, very like easy. It's a bit more complex than that, but it has a probability of um being right or wrong. And then just think about that. I think this is very important. Think about your prompt window, saying I do this as one person, now go away from this and and think about what this means for my organization if I don't want to type in the prompt anymore. I think this already helps tremendously. Just grasping of what is possible.
SPEAKER_01:Yeah, yeah, brilliant. And I like that. It's about starting, isn't it? Playing, seeing, experimenting. Exploration. Exploration. Fantastic. Well, listen, thanks so much for joining us, David. And and uh you know, David will be speaking in Munich on January the 27th at our AI for industrial print conference as part of FuturePrint Industrial Print. Thanks so much for joining us today, David, and looking forward to hearing your talk in a few weeks' time.
SPEAKER_00:Thanks for having me.
SPEAKER_01:A lot of fun.
