#101 Where AI can help, and where it can’t - Interview w/ Henrik Laxhuber

Hello and welcome back to another episode of the Process and Automation Podcast with the Automation Guys. So as we are going into our next 100 episodes, we recently have reached that 100-episode milestone. Uh, yeah, we are now kicking that kicking off that next era of the podcast. And we're kicking that off with the most popular podcast format we have here, and that's interviews. So yeah, interviews with guests who share the same passion with all things process automation as Arno and I do. So yeah, without further ado, I would like to welcome our next guest on the show. Our next guest says about himself, he would like, or he would like to kill for a good lasagna. So 100%, we will have to hear about that at the end of the show. And yeah, it was great pleasure to have Henrik Laxhuber here with us today. Henrik is based in Zurich and the founder of several AI startups, and I'm very excited to hear his thoughts about how AI can help and where not, and all things around that. So hello, Henrik, nice to meet you. How are you today?

I am great. Pleasure to meet you. I'm super excited to be on the show. It's an honor. I've been a huge fan of the show for a while, so getting to be here myself is really a pleasure.

Cool. So Arno, did you hear that? We have fans.

It's really good. And like you say, we have been inundated by a lot of our sort of regular listeners, and they have been asking a lot about more guest podcast. So we're gonna really make an effort in this next couple of months to get a lot more guests on board and really sort of just diversify the sort of things that we, you know, we talk about in terms of automation. And I think sort of just looking up on the sort of things Henrik gets involved with, it looks like today's gonna be a really, really nice and interesting podcast. So yeah, welcome Henrik, thanks for joining us.

Hi, Henrik.

Yeah. So, yeah, Henrik, we had a chance to have a brief chat beforehand, but for our listeners, yeah, please share maybe initially a bit about your background.

Yeah. So I am essentially a computer scientist. I studied in Switzerland and then moved to the US to found my first AI startup there. What we were doing there is essentially automating workflows of sales teams. We didn't really think of it as process automation, but looking back with the knowledge that I have now, it's really, it was a process automation solution very specifically tailored to the problems that you encounter in B2B sales. And this was like before AI really became a big topic. So this was before ChatGPT launched, before, you know, really all this talk and hype was about. So we thought being able to automate processes of sales teams using AI was pretty cool. But then obviously, you know, November, I think it was November, yeah, 2023, ChatGPT launched and we realized, wait a second, this opportunity is way bigger and just these technical advances are coming way faster than we thought they would. So we started zooming out a bit, sort of thinking more broadly, what can we do to help our customers and started to look at this more through the lens of process automation. So Well, I think a lot of people today, they come at it from a, we already know process automation and now there's AI, how can we bring AI process automation? We sort of came from the other direction where we have AI and then realized, oh, we need to automate processes. And ultimately what ended up happening is we realized that we need to start a new company to really exploit the potential of Lumen Forest. So that's where we are today. We're starting a second company focused on bringing AI to process automation. It's still very early stages. Like, we don't really have a fully built product that we're selling. We're just working with initial pilot customers, sort of testing the landscape and learning what needs to be done.

Okay, that's cool. Yeah. So here on the podcast, of course, everything is about process and automation. And Arno and myself, we have lots of experience in this whole field of process orchestration, workflow, case management, task management, and all that. So obviously it makes sense to, to not just see AI as a, it's just one thing. So it is definitely something we, we always talk about here as well. It's just one, it's just one piece of a puzzle. So combining these technologies together to get the, the biggest, biggest outcome, um, exactly, in the company. So it has to be just, AI is just one actor basically in this whole end-to-end process, like, like a bot is in, in, in a way as well. So, um, Yeah, makes sense to not just focusing on that one. So it's highly, highly busy, that market. I have just looked at that website. I don't want to mention the name if it's still— because it says it's still in stealth mode, which is quite intriguing and all that. So is there sort of a timeline on when you get started with that new business?

Yeah, so it's already started, started a few months ago. But I don't think that we have a timeline for formally launching. We need to tackle quite a big set of challenges. So the vision is really to build a completely new platform for how you automate your processes and to bring that closer to the end users or the ones who are currently manually doing those processes. And to do that, you need to build a whole bunch of separate modules. I can go into like what the vision is later, but it's a, it's a huge task. So we don't really have like a timeline for when we launch. We want to do this one step at a time, very slowly, sort of the way you eat an elephant is one spoon at a time. Yeah.

So, so if I can jump in, so usually a lot of the sort of topics we choose is are sort of derived from practical use cases for automation. This can be, you know, within AI, within traditional workflow, low-code, RPA, chatbot. So, you know, in very simple terms, what is it that the users of your software, your product will see, and how will it make their lives easier? And how would that sort of be— where's the wow factor? If you just sort of described to our listeners exactly to kind of what, what that concept looks like.

So, yeah, so the basic idea is that a lot of software today is kind of suboptimal because it forces you to work in a very rigid way. And that's where Trokens Automation comes in, right? Like you build your own custom processes, but historically that's been very expensive and inaccessible. So maybe zooming out even a little bit further, like every software automates some process, otherwise you wouldn't use software. And famously companies like SAP or whatever, those are like the first to really give you software for your core accounting process, et cetera. And they're very configurable, but it's very expensive to configure them. And so what you see in a lot of smaller companies is people install these systems, maybe not SAP, maybe like slightly cheaper and easier systems, but still very rigid systems. They spend some time configuring them, but really they start building their own process around their systems. So these are things like, well, we always store our files here in this folder and then we always like do this and then we do that, right? Like the classic problem that you have in process automation. And if you wanted to automate that, historically you would like hire an outsourced team somewhere where, you know, cost of labor is cheaper and you can get them to build custom software for you. Eventually we got these low-code platforms that made this easier. So there's a whole bunch of stuff from SAP Fiori, which is super easy to use, to more complicated stuff like UiPath, where you need a set of experts, have a dedicated team to support it. But the actual core systems haven't really changed. SAP is still as hard to use as it was 20 years ago. So the idea really is how do you enable the end user who's actually performing the process to sort of build their own dream system, to build their own user interface, build their own workflows and sort of have their own operating system that is layered on top of all of the underlying systems. So really rethinking like how you think about process automation, because what really matters is the problems you're performing, but you don't really care how that process is implemented. You just want to get it done. And the sort of opportunity that we see is to take a lot of the ideas that have been developed in process automation. So These are things like event storming for how you bring everybody into a room and design what a process is, figure out what a process is. These are things like domain-driven design, et cetera. How do you take all of these things and make them accessible to somebody who's not technical, who doesn't really want to spend the time thinking about how do I automate my process? Because I just want to get it done. I have a quantity of maybe, I don't know, I will spend 4 hours max to make it a little bit more efficient, a little bit. Less cumbersome for me, but I'm not going to learn how to use UiPath or automate this myself. So the approach that we are taking is to take a bunch of these modules that you need to do that. So a way for you to build your own custom interfaces, a way for you to integrate systems, a way for you to do process mining, et cetera. Bundle that all together and make it really, really easy to use through the power of AI, because AI can help you translate between the language that is spoken by the end user, which is very human language, and the very technical language that we need to actually implement those processes.

Yeah, absolutely. So yeah, I saw on the website and I really, really like that AI doesn't have to suck. And I like that really a lot. So it is somewhat true because it is brilliant, but it sucks for the users, for lots of users, for lots of businesses in a way, because they, first of all, they don't know how to use it beyond ChatGPT doing a few prompts, no one really knows how to do it in the wider scheme. And this is where it actually needs improvement. And I heard another term, not term, but sentence recently. I thought that was, I've never thought about it that way. So someone's saying, is AI holding your business back? And I think, no, AI is so forward-thinking, it should not hold you back, but because it is in one way, brilliant, but on the other way, it's not really that usable. And obviously with AI comes all that process automation. And so we obviously think from a technical perspective, it's great. Low-code is great. But from a business perspective, they still don't get it and they still can't use it and they don't have the benefits as promised basically in the news. Say, oh, AI is here, everyone can get a competitive advantage. No, they can't because they don't know how to use it and always need help. And if that's the mission, that's awesome.

Absolutely. And I think there's like sort of multiple vectors here for how to attack this. One of it is that AI will just get better by itself. Like right now ChatGPT is brilliant, but it's still not very good if you actually want to use it to get work done. Like the truth is AI is still very artificial but not very intelligent. And so it's very hard to productionize it, not just because businesses don't get it, but because it just genuinely doesn't work well enough for lots of applications. And finding the applications where it works well enough and sort of evaluating that and de-risking it, et cetera, is hard. But I think we will see that get better. It will just get easier and easier for you to just get work done within ChatGPT or within another vendor who maybe is more deeply integrated into your systems. But that's sort of what we call the interactive on-demand use case. But then there's lots of use cases where you actually want to build a repeatable process. Lots of businesses run on repeatable processes, it's easier to hire, et cetera. And that's where you really need new solutions to come in to bring in all of these technological innovation and problem together.

Yeah.

And is that the point where people was like, no, hold on, hold on. This is Carl from IT compliance, Hendrik. You telling me we're gonna let people loose and start building AI and on our systems. And, you know, we, we need governance and we need security, you know, so, so for, for listeners out there that, that this sounds like an absolutely amazing kind of step, you can to kind of create really intelligent automations, reuse what you have, focus on the sort of task you want to achieve, wrap that up in a really nice package. In terms of the compliance, what is it that you can say to the listeners out there for potentially the products you guys are rolling out? How do we solve that IT compliance problem? Carl Compliance, he's like looking at this and say, we don't want this in our environment. We want to maintain control. We know what you guys are up to.

That's a brilliant question. Yeah, we thought about this a lot, actually. So really, we think about it as two ways of using AI. You can use AI at runtime. So this is when you are actually executing a process for that one specific execution. So let's say you're, I don't know, writing an email to a customer. Like, that's one execution of the process. You can use AI there. Or you can use it at modeling time, which is when you're figuring out what is the process even, like, what do I want the interface to be? What do I want to do when exceptions happen, et cetera? And I think if you're using it at modeling time, it's much easier to sell that compliance-wise because you still have the fully deterministic control of the underlying system. There's a human verifying everything, and once the process is built, like, it's deterministic, it gets executed exactly that way. And you don't need to really evaluate or think about AI risks. But obviously there's huge opportunities also for using AI at runtime. So just think about working with unstructured documents, right? Like historically you would have to build OCR templates that you put into these intelligent document processing engines that turn out to be not very intelligent at all. And now you can just throw them into a big large language model and they can transcribe the documents, it can do vision, etc. And there you really just need to build up a dataset of what you expect to happen versus what actually happens. And there are ways that we can help to make that easier. Like there's techniques where you can sort of write a policy, a governance policy for what the AI should and shouldn't do and what you consider risky, and then have that automatically checked against all the data you have, even if you don't have the labels, like even if you don't have the transcribe documents, but you only have the raw documents, so to say, and then have AI sort of self-improve and help you build out that dataset and help you monitor that over time. But it's still a very nascent area, I would say. I think we will see lots of improvement there over the next few years. So if you are in a very, like, regulatory or compliance-constrained field, I don't know, you want to do processes for accounting or you work at a bank, probably it's best to just stick with using AI at modeling time, which is also our main focus, and to only slowly roll it out at runtime for documents where, you know, it's fine if you make some mistakes or you have a human in the loop to proofread it, to verify that everything is accurate.

Yeah, yeah. I mean, and we, we call that human in the loop as well, for, for some of the components of AI that we use in in, you know, the automations we built. So the AI perhaps make a suggestion, it might flag a document or sections in the document as red, amber, green, and really sort of just present the results, I guess, to a human counterpart just to assist them with decision-making, say, you know, this is what was flagged, but you still have the ultimate control. And that sort of just gives, I guess, peace of mind to people that they are still in the loop. But we do see a future where, you know, the way you sort of described the modeling approach is, you know, where it really can learn from how people do their work. And it is, you know, it's almost like a step up from RPA because obviously RPA is something that you would have to do a time and motion study, understand what somebody does to achieve a job, then you need to go and learn an RPA tool, You need to go and implement the sequences, publish that in a bot. The bot needs to be published. So that's a very, very long cycle to get to that point, you know, and a very kind of a mechanical process as well. You know, if you think about, you know, publishing an RPA bot, although the benefits is really great, but obviously getting to that point where it's actually productive and effective, that can be a, you know, a longer running process than sort of looking at perhaps how somebody can model something, publish it, and with those security guardrails around to say, well, this will only ever execute and mimic anyway what a person would have done, kind of provides that sort of safety net. So yeah, I think that in the future there's probably going to be more of that. And like you say, with the longer running AI, document processing, you know, where there are sort of areas where you can exploit AI's full capabilities to actually surface any problems to you. And that potentially would grow over time, I suppose.

Yeah, it could go, right? Like if you, if, as we get better at solving these challenges and as AI itself gets better, I think there's less and less need to model processes very rigidly. And you're already starting to see some companies sort of do this where you just throw in some documents and you add some API endpoints to ChatGPT and it just sort of does its thing.

Yeah, Henrik, you mentioned one thing beforehand, one thing you're particularly excited about, one application of what you're building on. You mentioned it's in logistics. I guess our listeners would love to hear about one of those applications. Of what you're building and what it could be, how it could look like. Maybe that's something you, you, you can take us through.

100%. So like, let's say you run a small trucking company, you have, I don't know, 10 trucks or something. The way this works today is, even if you have more, if you have like 100 trucks, it's really up to very, very big companies, they all sort of work the same. The way it works is your customers send you an email, they phone call, WhatsApp, anything to say, hey, I have a shipment that you need to pick up at this location, bring it to this location by such and such time. And oh, by the way, there's all of these things that you need to take into account. And how is your grandmother doing? Right. And then somebody sits there and types this off into a system. They prepare a quote and it's very manual process. There's lots of unstructured documents being exchanged back and forth. And now what you can do with AI is you can sort of have AI at runtime to take these orders, automatically transcribe them, put them into the system, get a quote from the system, and then at that point have a human in the loop and say, is this quote fine or not? You want to add anything, modify anything, and then just hit one button, press send, and you send the quote to the customer. But to do that, you really need all of these modules that we talked about before, right? You need a way for somebody to build their own interface to make this really appealing because you don't want to have, you know, click, click the task button to execute that one specific task and it's like open to email browser or whatever. You know, you want this to be a very nice UI where you can sit and do your work and not have your work be in one other system and have the task about with one on the side. So that's like the— you need a nice way to design interfaces. You need a way to integrate with all of the underlying systems. So today you have like some system of record where you store all your orders, your prices, your quotes, invoices, etc. But then you also store any other documentation that you get. So maybe you get a photo of the shipment that tells you roughly how large it is, how much it weighs, etc. And you store that in a different place. Now you need to link those, etc. It's really a mess. So what you can do now, if you just design your own process, you design your interface, bring those together so that with one interface, all of that just happens automatically. And like, none of this is really revolutionary in the sense of you can do this today with any good process automation software, any good interface designer, right? Like if you're in Microsoft Power Automate, you can do this. But what we are doing is we're just making it much, much, much easier to use so that you can set this up yourself as somebody who is in the driver's seat doing those processes. Like you are the one taking orders, or maybe you're the one managing orders or taking, et cetera. What you want technically, and you don't need anybody to coach you on DTMN or any kind of process modeling. You just do it yourself.

Yeah, driving seat so that even truckers can do it, right? That would be good.

Right, right. I think there's some considerations here. You don't want necessarily everybody on the team who's doing the same job to model their job differently and automate it themselves. Like, you want some governance there. So that you have a repeatable and consistent process. But yeah, the idea is definitely to make it very accessible to these mid-market companies where you have lots of people who are interested. Like, these companies are extremely operationally efficient and they're always looking for ways to be more efficient. You know, they're very street smart, but they're not necessarily technical and they won't spend, you know, 3 weeks learning how to use UiPath.

Yeah, midsize company, you mentioned that. Yeah, that's definitely a big category of businesses who didn't use or do not use these technologies these days. It's a huge market, of course, for process automation and for low-code, no-code, and AI.

Absolutely.

Yes. And part of that is also that a lot of companies still, like, you have a core system for your accounting. but you don't necessarily even have a CRM where you manage customer relationships. You probably don't have the CRM for vendors where you manage vendor relationships. Like all of these systems that bigger companies tend to acquire over time, you don't really have like, it's just in the minds of your employees, in the minds of your managers, or in unstructured documents. And really there's no, there used to be no positive return on investment to digitize that. Because you're not doing that much of this type of work. Whereas now it's become so much easier to digitally transform your processes that it's suddenly a positive return on investment to go and do that because it gives you more repeatability, more visibility into your business. Lots of mid-market companies are also struggling to hire good talent. And so by being able to sort of do more work with the same talent, or as people retire, keep doing the same work that you're doing with fewer people, assuming they're part of the value proposition.

And just on the sort of the human capital and the sort of resource and skills part of this, obviously, if you've got somebody with a specific skill set and they jump in and they sort of start modeling processes, is that something that's then easily shareable with other members of the team? Or is this like a sort of a more focused at an individual? How does that work? Because we obviously, in our area of work, there's a lot of process standardization, a lot of sort of pressure to actually make sure that, you know, if one member of the team is absent, then somebody else can pick up the work and process that. So how is it that you guys can solve that challenge or kind of keep that knowledge So it's accessible and usable for, let's say, other members of a team or within the organization.

Exactly. That's a great point. So I think one of the most overlooked factors in modern-day software is sort of the human part of the work, like the work part, essentially, of who knows who and who works together, who doesn't work together, where do approvals flow, etc. So we're very much focused on the team use case, and the way we do it is we capture this entire relationship of who's in your company, who knows who, who needs to approve what. So it's very easy to build processes that are collaborative, that involve multiple stakeholders, need approval from multiple people, et cetera.

Awesome. Very good. So now, where do you see then, with what you're building, where does it head to? So what do you see sort of as a future of AI, maybe in combined with the process automation.

Yeah, so let's maybe take that back to what I was talking about before with SAP, etc., not being very flexible. I think the future of software will increasingly be more human in the way that it will be easier for everybody to tailor their software to their needs. And part of that will be solutions like ours where you build your own processes on your own custom interface. That you define yourself, but we make it very easy for you to go through that. But part of it will also be just new solutions coming on the market, like a new accounting software that is just easier to configure out of, like, from the get-go. So you don't even need an expensive custom workflow automation solution on top of it. You just automate it within that solution itself. So really, the future that we see is you being able to tailor every piece of software that you're using very specifically to your workflow and using a process automation, process orchestration solution like ours to tie everything together and have your own personal operating system, I would call it, for the tasks that you are doing yourself.

Yeah, it sounds very appealing, I guess, for lots of businesses going forward. Yeah. So, yeah, you don't want to commit to a timeline, but it would be great to have it already today. But Yeah, definitely AI is one thing to look at and has to be in this combination, isn't it? So I think for businesses, or why it is not successful, is because it is not the only technology. And it sounds like for lots of businesses and observers of news, it is sort of the big answer, right, to everything. But it is really like Yeah, so what do you want to achieve in your business after all? And then it is that combination of technologies. Maybe it's not even AI solving a business problem, but we, of course, we all believe it will tremendously benefit any business going forward, especially with what you mentioned there. And if that can be realized, that will be awesome. It's probably a combination of all these things, a bit of process automation, a bit of human in the loop, bots in the loop, and AI in the loop, I guess, how you could probably call it these days. And then ultimately businesses have to come up with a good plan. I think that after all needs to be done. So very often it is very reactive. What do you think these days how businesses or people approach AI.

Yeah, I think what you said there is very important, that like you need to figure out what your business wants to do after all. Like that sometimes gets forgotten in all of this hype that really this is not about technology being cool, but it's about how can you help your business get better and focus on the things that actually matter to your business and don't get distracted by the things that maybe don't match to your business, like the actual way of how you execute your operations. Doesn't really matter. You just need to get a job done for your customers, right? And AI is part of an answer there, but I would be very careful about calling it the answer. Uh, I think with all the hype going on, people also starting to realize that maybe this has been overhyped. I think you really need to think carefully about where it's useful and not useful. On the other hand, we will also see this technology just keep getting better every year. It has done so for the past 5 years, and there is no reason to expect that it will stop getting better. So if you can find a way to sort of rebuild your processes in a way that you can leverage AI in small points of it at runtime, not just at modeling time, and then sort of build that out as AI gets better, I think you'll be in a great position. Whereas if you are not able to do that and you need to keep scaling your operations by hiring more humans, then as AI gets better, the delta between you and your competitors will keep increasing.

Yeah. And it's sort of just, I guess, you know, having people understand what the practical benefits of AI are, because I think that we, we sometimes talk to people and it's, you know, it is the hype and it's almost like people people just switch off when you start talking about AI because they just see, oh yeah, ChatGPT, I sort of had a play with that. I don't really see how that's relevant to me. I'm just going to, you know, get back to my day job and solve all of these problems because I've got a ton of work to do. I don't have an interest to actually get involved with this. I think that if that can be flipped on its head and people do see the benefits and it is accessible, easy to use, easy to understand, easy to transition into your day-to-day life, work life, where you can actually use it and see outcomes very quickly. I think that would persuade people to stay focused on it and actually start bringing it in more, you know, as part of their thinking when they look at problems rather than relying on, you know, the IT department to give them solutions because we actually do see that a lot where in one way, the tools we use are very accessible. You know, you can train people to create forms and workflows. But as soon as the complexity increases, then, you know, people start to kind of step back slightly and they look at this and say, well, this is an IT thing anyway, we shouldn't be involved in this, what are we doing here? And I think that is the major problem to solve where you know, IT, you know, you could remove them from being a bottleneck. But what you don't want to do is kind of get people involved in something like this, and then they realize, oh, this is a lot more complicated than I thought, and then they step away. And then there's actually— it compounds the problem to IT because all of a sudden they've got all of these other ideas that people are throwing at them now, or half-baked solutions, and they're like, you know, doubled or tripled their workload to get things done. So it is kind of truly trying to get to that place where, you know, we can genuinely turn people into citizen developers and they are masters of their own destiny and they can actually see the practicalities of this and they actually get excited of using the technology rather than sort of stepping back from it, you know, after a period of time because it's just too difficult, it's not accessible. It's for technical guys. I shouldn't, shouldn't have wasted my time looking at it. It is that. But I think if you get to that point and you can turn that into a positive outcome, we do actually have people's time. People get it, they understand it, and in whatever quantities they use it, they will go back to it because they're like, this genuinely makes my life easy and I can't live without this. And I think that that's actually the key thing is to find that ability to unlock that in people and removing those accessibility barriers, that's, I would say, a lot of it is perception, but I think there is a lot of reality. If when you want to implement something in ChatGPT, that is more than 40, you need to know about APIs. That's kind of, yeah.

And this is the thing, as soon as you talk with the business, obviously business members, they will describe problems on a different level. They say, okay, yeah, then go into the management system and create, create this and that, right? But on the level of actually process automation and AI, it means REST APIs and interfaces and that kind of stuff. So obviously there, they will switch off. And even if you have a nice local tool and then you say like connected systems or whatever you call it, and then easy API integration even for the non-technical, but even that that will make people shut down, right? So, but if it just happens nearly because the AI knows your systems, this is what I understand from it, Henrik, what might be the future from your tool. It will just happen. It knows that when you talk about order management systems or purchase order systems, it will just do it.

Exactly. And I think the key insight here is that as long as AI is the reason why people should use your software, That's not a very compelling value proposition. Like, you don't order from Amazon because it's internet, right? You order because it's convenient and fast and easy, et cetera. And so to get people to use AI, you need to stop telling them that it's AI and instead show them why it's useful to them, why it makes their life better, why it makes things easier, et cetera. That's kind of what we're trying to do. We're making it very easy for you to customize your sort of digital experience, how you get your job done. Without really sort of forcing you to know that it's AI. I mean, you will know because it's an intelligent thing that might feel smart, but it's all like the main logic is all with AI.

Yeah, very, very cool. So yeah, today I was reading an email, um, the experiment AI experimentation race is on, and that describes it very well because we are still on this, on this track for values or finding value. So, and hopefully, um, yeah, your, your implementation on your startup and what you're building will get us out of that experimentation phase because it needs to be that, it needs to be the proper value. And yeah, it is massive experimentation at the moment. No one really is saying this moment experimentation race. I think that described it very well. So yeah, and on that, we come nearly to an end of podcast. Is there anything we missed, Henrik, you would like to mention before we go into our closing session?

Yeah, one thing I would like to mention is that we don't have all the answers. So as you said, it's still very much in the experimentation phase. If we had all the answers, we would already have a lot of the product. So it's still very much in the phase of we have some of the pieces and some of the ways to connect it, and we're working with customers and they kind of like it, but it will take a long time to really roll it out and get it to the point where Everybody can just come in and use it.

Brilliant. Yeah, thanks. Thanks a lot for sharing, sharing your thoughts on, on AI and what AI is and where we can use it and what the future looks like and how that all will be going forward in the future. So I'm looking forward to, to your launch. I will look out for that on LinkedIn and let us know. Maybe we do another session when you are up and running. Very keen on that.

Absolutely, I love that.

And yeah, so we and our listeners here, we like to know a bit more about our interview guests. And sure, we talk about technical things and business things here a lot. So I have a few other questions. And there's this one question in the room I'd like an answer to. So yeah, your LinkedIn profile states that you would like to kill, or you would kill for a good lasagna. How did this come about?

That's a great question. So I'm a very passionate cook, and I remember back at university, like, me and my friends, we would often cook together. And over time, it sort of evolved into an informal competition of who makes the best lasagna. So we would try all kinds of things, like, how do we make it vegetarian but still taste great? How do we do this? How do we do that? Uh, and that kind of birthed my love for lasagna, almost. But kind of an inside joke between me and my friends.

Yeah, I love lasagna. So maybe at some point an in-person event.

Yeah, we can see who makes the better lasagna between the two of us.

Brilliant.

I'll stay away from the cooking. I'll just pour the drinks. That's safer that way. Let's just put it at that. Next question from me, Henrik. Who is your idol and why would you say that?

You know, I've been asked this question a few times and I don't really have a sexy answer because my answer is usually it kind of rubs me the wrong way to idolize people. I think you can get inspired by the great work that individuals are doing, but to then like elevate that person above everybody else, which to me is like the idea of what an idol should be, sort of rubs me the wrong way. So I would say that if you want an example, I think People like Daniel Dines, who founded Wipath, is very impressive how he spent, you know, 10 years like slowly grinding away at that problem. And it didn't really work until, you know, maybe 2017, 2018, when suddenly it turned this huge rocket ship. But before then, he was just like very slowly grinding away at it. I think that's very inspiring.

I think that's quite interesting, Sascha, because that's, I think that's the third time Daniel has made our idol list.

Yes. Yeah, it is.

It's, uh, he seems to be quite an inspiration, inspiring person.

Yeah.

Yeah, I mean, we're UiPath partners as well. I've heard the story and it is just incredible to hear that. And, you know, I think it is truly an inspiration and perhaps for people, you know, in the AI field to say, just keep on going, keep on grinding, you know, this will take off. You know, I think RPA maybe back in the days where sort of AI is now, you know, where it's, you know, people sort of see it and they understand, but then there's this moment in time where if you look at UiPath today, gosh, what a story.

Yeah.

And I guess they sort of the groundbreakers of RPA and what's next. And, you know, so it's kind of a true, true inspirational story from Daniel and what he's done there.

Yeah. So the next question I have is, imagine you could get all the contents and knowledge of a book instantly. Which book would you choose?

I have to cheat here because I'm a computer scientist, and there's this collection of books really which is called The Art of Computer Programming, which contains essentially all the knowledge about computer science that has ever been discovered, because it's still a very young discipline, relatively speaking. Um, but it's cheating because it's a collection of books. So if I have to pick one specific book, like there's one specific volume, it's called— for those who know, it's called Volume 4B. It's about certain draft problems, which I find very interesting. But yeah, that's my kind of very computer scientist answer to that question.

Oh, cool. That's good.

That is a good practical answer. And I think it's probably something that is useful day to day as well.

I have to check that out now, Volume 4B.

I'll check it out as well, actually, especially Volume 4B.

I will put a link into it, so if there's, into the show notes.

I think it's still in draft. So the, it's a professor who is writing the series and he's been doing it for, gosh, I think it's been decades now and he's still drafting these books. So it's not quite finalized yet, but you can already sort of read the—

It will never be finalized, I guess.

We will see. We will see. Right.

Um, I've got another one for you as well. Um, and yeah, I guess this is probably just a bit more philosophical and, um, you know, a bit more sort of in your life, um, as you, as you progress in your life journey. Um, well, you know, what's the best advice you've ever received from, from somebody? Uh, you can't expose who it is, but you can if you want to.

I don't actually remember who said this to me, but somebody said, uh, you spend more time on the journey than at the destination, meaning you set yourself goals in life, but really all you do in your life is being on the journey towards that goal. But the instant where you have arrived at the goal is just a very short instance in time, right? So what that means is you should set your goals not based on where you want to be, but based on what journey you want to take.

Yeah, that's, that's very good advice.

Yeah.

Yeah. So there's— I think there's, there's also a very, um, wise person, isn't it? Sort of the path is the The journey, isn't it, as well? Similar.

Yeah, exactly.

Yeah. Okay. Yeah. If you could be Olympic athlete, what sport would you choose?

Uh, I used to do a lot of fencing when I was younger. I think fencing is— yeah, fencing is an Olympic sport. I remember watching the Olympic fencers on TV, and it's just incredible how skilled they are. I mean, I was fencing for years and years and years, but I never got to that level of skill. So if I could be an Olympic dancer, I would love to do that.

Yeah, cool. Very good. So yeah, thanks a lot, Henrik, to be on the podcast. It's been a pleasure. Yeah, as I mentioned earlier, if you like to come back, you're welcome anytime. And we have another session and talking about AI, process and automation, or any other topic which is really related here to this show. Before I will, before we close here, maybe some listeners would love to get in touch with you. What is the best way to do that?

You can find me on LinkedIn, Henrik Lakseberg, not very hard.

I put that link into the show notes as well when we share the podcast, and everyone will find it there as well, plus the books and the link to The link to any website is not possible yet, isn't it? So if you—

no, okay, unfortunately not yet.

So, okay, so I keep that out. So we will share that with everyone very, very soon. And yeah, thanks again for being on the show. And yeah, we will be back here with another recording very soon. And until then, let's automate it.