Implementing New Technologies - Part Two - With Florian Beil

Welcome to the Trend Detection podcast, powered by Senseye, an industry leader in using AI to drive scalable and sustainable asset performance and reliability. This is a new publication designed to help you go away with ideas on how to achieve maintenance efficiencies.


For this 4-part series, I’m joined by Florian Beil, CEO at Axulus Reply, an accelerator to enable manufacturers to embrace the Industrial Internet of Things.

In the second episode of this series, we look at the challenges manufacturers face capturing and utilizing their data and how they can scale their technology deployments across a global industrial network. 

Please subscribe via your favorite podcast provider (Apple, Google and Spotify) if you'd like to be notified about future episodes and let us know your feedback by leaving us a review. 


Key topics covered (click to jump to the section)

  1. Data Issues
  2. Manufacturing Challenges
  3. Overcoming the issues faced
  4. Subscribe to our podcast

Niall Sullivan, Senseye:
Welcome to the Trend Detection podcast, powered by Senseye, an industry leader using AI to drive scalable and sustainable asset performance and reliability. For this four part series, I'm joined by Florian Beil, CEO of Axulus, Reply, an accelerator to enable manufacturers to embrace the industrial internet of things. In the second episode of this series, we look at the challenges manufacturers face capturing and utilizing their data and how they can scale their technology deployments across a global industrial network. I hope you enjoy it.

And you mentioned, it's something that's we've not really talked about yet, but data, and I don't want to frame this question too much, but the question was a wider question about the biggest challenges in manufacturing, and we could go down to roots or the general issues or related to technology. So, I'll let you maybe lead on that a little bit, but we can focus a little bit on the data issues. Or first of all, do you see data issues in the quality in terms of the amount of data or lack of data, let's say personal. So, that's a few questions in one.

Florian Beil, Axulus: Yeah, you just guide me through them. So, the challenges around data manufacturing today are... And of course, there are different type of customers, yeah, don't get me wrong. But in general, if I go through the shop floors, I think the first thing, and this is always inherent, is that you have a highly diverse machine park. You have different type of machines, you have different type of PLCs, you have different type of sensors, you might have a connection network or not. And then, it starts really with the basics. So, first question is, do you have an infrastructure at all to collect data from the machines? And what I see more than once is that, it looks perfect for the outside, like fully automated PCB assembly lines, which have their individual NC programs, which produce a scale. But then, you go with the customer rely and ask, so here you have an AI, so an automated inspection station, where the chips are checked, where does the data from the AI go to?

And I said, "That's staying local." So, okay. Okay. Go to the next one, same answer, right? Same thing, same. Which then leads to the fact that although they have the data, in case there is a problem with one of the PCVs at the end customer side, they need to physically go through the line back, look at every station. So, how was that produced that part, and when, which of course from a digitalization point of view is crazy in the sense that you have some data but you cannot use it because it's just connected and stored it into one, as I call it, unified name space. It means that one space where all this information is collected and can be used across different stations. That means then also the second point, which is of course due to the complexity of software, is of course, some sort of data normalization as I would call it.

It's still a difference if I connect very old machine, maybe where I need to do a retrofit to add some sensors, whereas a new machine like an S7-1500, which has OPC-UA on board or full driver set. And to normalize information, meaning this data point out of the S7-1500 means temperature, which also means the other data point out of an old style machine. That's also quite, can be quite some effort. And I think a lot of customers underestimate that is not only about storing bites, it's about storing contextualized bites. And that data illumination, of course, also is something which takes time, and most of the times underestimated, respectively, not set up in the right way that you can later on it because doesn't help you if it contextualize one machine and then the next machine, which is a different one as a different context, so I cannot compare and cross- correlate. So, I think that's for me on the data side, really the issue.

But again, I also see technologies there emerging which help to really scale connectivity in that sense as I described it. So collecting data from different machines and normalize them. So, for example, we integrated a partner solution called EasyEdge from the [inaudible 00:04:45], which basically the core concept is they build drivers for different type of assets and you can really, out of our solution design and connect to the machine, can configure the data model online and normalize it in a scalable manner. So, although there are challenges still today, I think there's also technology coming up helping us out there at some point of view. And then, of course when you have the data then and you start to collect it and see also the first value, add maybe already with condition monitoring, then the sky is the limit because then you have it in an infrastructure where it can build or add more solutions as needed.

Niall Sullivan, Senseye: And if we just-

Florian Beil, Axulus: Long answer for first part.

Niall Sullivan, Senseye: No, no, no. And we've got one part left by multifaceted question. They broadened out those, the manufacturing challenges they face today, aside from the technology and maybe the data side and IoT side.

Florian Beil, Axulus: Yeah, look, I think the general manufacturing challenges, I would say they're intact, in some sense. So, it's all about more efficient, more variance, single products or lot size one topics, much more complex reduction processes. If I only look at electronic manufacturing, it's a highly automated, highly intrinsic industry whereas so much domain knowhow in the production process alone that one person cannot grab it. And of course, they are then no full up challenges there. So, how do I optimize on very automated, very good process? How do I ensure, I would say, knowledge consistency if people leave or experts leave? How do I preserve the knowledge? We also have a challenge, which is a little bit of, I would say homemade from a lot of software providers, that's due to this high complexity of the production processes, a lot of standard software which is typically used like MES systems, also SCADA systems, whatever, are delivered as a standard but then they need to be highly customized to the individual installations.

Which then of course gets the customer off the main release path with all the additional costs and complexities involved. So, I think these are the key challenges, just perform better in operations. Also, of course, as sustainability energy is also I would cluster under that, which is of course a huge topic. And to be honest also, for some customers at the moment, a life-threatening topic. So, if gas prices again quadruple, what does it do to a bakery? So, it can kill your business. So perform better, more efficient, constantly push the limit of that. Also, involve, preserve the knowledge you have in your workforce, and make it scalable, in a sense that not only one person knows how the MES works, but maybe a system or a group of people. And then also, the transition of, I would say classic software, proprietary monolithic software setups into a new modular market service based IoT world.

I would say these are some of the key challenges I see at the moment, or at least where some of the customers are working on or struggling with even. And yeah, let's see how this goes on. But I think the inner drive or motivation to solve these problems is there, more maybe than ever with all the crisis we had. Now, the gas crisis in or gas price crisis in East Germany and maybe Europe, there's a lot of motivation also to drive the edge there. That would push a limit there to get more efficient and avoid costs, which can be taken out with technology out of a system like energy prices, like down times where of course since I play is a huge role, and all these things come together. And again, I'm optimistic in a sense that technology, there's still much more to come, which can help us there. And innovation power is quite big still. So, I see a lot of things coming up there which are promising.

Niall Sullivan, Senseye: Yeah, and bear in mind the current situation. Well, yeah, obviously across Europe and the UK as well with energy prices, sustainability, and obviously with things like climate change, it's always been a focus, but I guess when it's really impacting and hitting home, I guess that's really starting to make an impact on the manufacturing industrial world.

Florian Beil, Axulus: Right, absolutely. 

Niall Sullivan, Senseye: Yeah. So what I wanted to do, flip that round and really dive into it, those reply really and how you solve some of these challenges, and I know you've got quite an interesting approach. When I first learned about you guys, it was very eye opening. I haven't seen anything like it personally myself. So, it'd be great for you just to explain briefly, you've talked about a lot of challenges, so it might maybe the top three. Maybe, I don't know.

Florian Beil, Axulus: Yeah, maybe. We [inaudible 00:10:26] think about some ideas for solving them, yeah. No, happy to do that. So, I think the core idea behind Axulus is that really in order to really adopt digital solutions or technology in a production environment, you need to tackle the issue not only from a technology point of view, so what type of cloud, what connector, whatever, but really think that through end to end and provide the customer a system that supports them along the way. So, the first question, of course, starts if I talk to a line manager, it's not so much which cloud is Azure, or AWS, or whatever, the first question is, so what type of problems you can solve with that? So, can you increase the OEE of my press line? Can you reduce the outgoing inspection costs of my warehouse management system?

Because this is the reality that people live in, right? So, it's their language, it's their lingo. So, the first question would be, "Okay, I don't care too much about technology, but what type of problems you address?" Then the next question is, "Of course, so if you address them, what are the target KPIs that you improve with that? So is it OEE, is it outgoing warehouse costs, is it scrap rates?" So, what are the KPIs that are impacted by a digital solution that improve my KPI set and how does that relate to my current production processes, operating procedures? What do I need to change? And I think this is something which sometimes was forgotten, and I'm also reflecting on myself. So, don't get me wrong, I'm not talking about others here, but it's sometimes was forgotten when we started with all cloud and IoT platforms that we went to customers and said, "Look, we have an IoT platform, we can do everything." But actually that means nothing to the customer, right?

So, the first step is what is the value added, and what are target KPIs, which processes are there to be changed? Then the next thing that needs to happen is somebody needs to translate that use case, as they call it, into a technology stack. And that technology stack, typically what we have is a presales type of approach. You design it, maybe you have an IT platform, so some of the parts are standardized, but still there's a lot of complexity translating that technology stack into something meaningful operational for a customer. So, which type of sensor... Let's maybe stick with predictive maintenance, so let's assume I identify it, and my first question, "Okay, this is my process, this is the critical asset I need to monitor that I can't afford a downtime." Then the next technical question is, "Okay, maybe I can collect data, but which type of sensor do I need to pick up the right KPIs or parameters in order to predict the failure?"

"Where do I need to mount the sensor? Because it doesn't help if amount of abrasion sensor and the wrong access, so it cannot pick up a failure mode." Which then relates again, "So, how is your equipment mounted?" So, what I'm trying to say is, there's a lot of complexity in that, and the way you can manage that is not on PowerPoint or, I don't know any, Excel, or whatever, or email. But for these type of complex design and simulation topics, you have today an engineering system which does it for you. So, a system that breaks down your target value design into individual building blocks. So, what is the KPI, what are the driving factors, which processes are attached to it? That breaks down your technology stack into bricks, into microservices, into connectors, and also allows you to easily adopt it for a specific installation.

Meaning, it tells you which type of modules I have, which type of building blocks I have, which type of integrations are already there, which type of modules are missing, so I forward it into an R&D processes and get it developed once and put into my repository. But it's a system that supports this whole process of translating the use case into program installation. And then last thing for the end customer, which is important, "So how do I get this thing installed? What are the 5,050 steps I need to do in which sequence that my service provider or my service team can now execute in order to get this solution for the use case I defined, get it installed." And then we are talking about what we call digital installation workflows, so we want to provide the end user the experience.

These are the different type of use cases you can address. You can go in, you can configure your own requirements there. Then you load a so-called template, solution template, or the solution, whoever does the technical solution load, the solution template against it, can engineer it with the engineering system. So, Axulus will tell you this works, you have that in pattern, you can configure it differently, it fits, or there's something missing so build it and then you have it in the system. And then, you publish it again back to the end user who ordered the problem statement. And what he gets is actually, of course, information, he gets the tech stack, and then a step by step instruction which you can start to execute to really get the solution installed and onboarded. Plus, of course, a lot of industrial hardened issue raising, and descending, and so on. But I think, this is actually the core idea of Axulus.

We don't look at the technology alone, we look at the end to end process from, what is my problem down to I have it installed, it's up and running, and I have to value get generated. And we do it based in an engineering system which allows to configure these... First of all, we can templify use cases, we can templify solutions, we can configure with the mono-solution designer the delivery foresight, so it means it's also feasible to adapt a solution easily to a specific technical setup, and we can execute with installation workflows, the whole enablement and rollout of these things. And with that engineering system, or the first time, you have a lot of information at hand, which allows you to go further than that. So for example, as we have a templified use case, a templified solution, you can compare, so which type of customers are using which type of templates?

And you can do recommendation engines like, your customer, other customers in your field, which have the same scope tags, they 80% have these three solutions installed, maybe you want to have a look at it. We can automate the connection, so we can integrate with EasyEdge and out of the solution design, directly connect to the machines, of course, giving all the access rights and firewalls are open for access. And out of the solution design configured the data mode. We can automate software configuration deployment. That means in the tech stack that you design or in the solution template, if you need to configure that for the customer, we basically can trigger a CI/CD pipeline out of the configuration and pull out the modules, software modules, as configured in Axulus, and then triggered the CI/CD part, and deployed, and run it. So, it means actually, for the customer, for end use, it means, in the installation workflow there's only one button, so deploy software, and the configuration that was delivered to him is automatically installed on the cloud and it runs.

So, there are many, many things we can do because we have this engineering system. And for me, it was a little bit eyeopening when I first gave it to my actually a first customer, which is Autotech, which works with Axulus engineering system. Is it's a machine builder. So, they build their own machines and they want to build their own digital solution templates and deliver it. And as they start to work with it, they give it into sales, which then promotes use case templates to the end user and so on there. So, many ideas coming up of what it can do to ease that whole process of getting a solution installed at a customer. That is really amazing. So, we are now going into regime for example, where we also manage the so-called exposed data.

So, we manage the installed base of which type of assets are there in the field, what type of data are they exposing, and then the system automatically matches this exposed data with potential use cases and what type of information they need. So that you have, actually, you can go to the customer, you can say, "This is your machine based on our library, these are the template of use cases which would work on your machine. Which one you want? Which one should we configure and deliver?" And again, it made much more easier for sales to sell a solution. So, instead of starting a project going, somebody going there, looking at the machine, doing a PowerPoint, doing an architecture, it's all automatic out of the system, and you're very quickly to the installation workflow, so that also the end customer, the end user, has the feeling, "Oh, that's a product, right? It's nothing like the project."

"Ah, they know that? Okay, I want this use case, I want to save energy. I want to save chemicals." Next thing you'll see is the insulation workflow and the service team installing in and up and running. That's a complete transformation of how you do digitalization. And that's what excites me, right? Because I was suffering for 10 years maybe to scale that whole thing. And now seeing the first customers really doing and it's really great. But it's Axulus. It's actually an accelerator for the adoption of digital solutions, templified, an engineering system which does that. And also, not last to mention is, it's also a huge partner model, because a lot of partners like Senseye, you have a great technology which addresses very specific use cases, but your end customers, of course, partners, are struggling to sometimes scale that into all these very different or sometimes different plan setups.

And that's actually where we want as Axulus to help also or work with our partners because we are focusing really, on this end to end part, so how can we make it seamless and scalable to get solutions installed? Whereas you and others have great technologies which can help really customers, it would be easy for them to use it. And that's actually what Axulus can contribute. So yeah, end to end, it's a great story so far and we are very happy with our customers working on that, because I still think that the basic idea is viable, and that especially it opens up a lot of new functionalities and features which we can build now based on information and the engineering system we have to support our customers and partners.

Niall Sullivan, Senseye: No, that was a really, really great.

Florian Beil, Axulus: I don't know if thats clear for you. 

Niall Sullivan, Senseye: No, no, it is really, really clear explanation. Really interesting seeing some of the work you're doing. And what I like about it is like, say you've got an approach, a proven approach, which you can then go to other similar company and say, "Oh look, we've got this approach, similar challenges and this is what we did here, here, here, and here." And does that helps to solve that plant by plant, as we're saying, within an organization, different plants, so much work independently, that must really help to overcome those issues that we've sort of covered already? 

Florian Beil, Axulus: Yeah, definitely. The point is that, with Axulus engineering system, you have these use case and solution templates, but of course, you also have a, we call it portfolio manager roles. So, the central role which overlooks which type of use case, which solutions, but more importantly, which type of building blocks, so microservices, sensors, connectors, UI elements, whatever, are in the system which can be used to build a solution. And of course, this allows very, very efficiently to centrally manage which technology is really used. I don't know. If there's no AWS in the system, then of course you cannot configure a solution based on AWS if you centrally decided it, but use another cloud platform. And what I also like about this approach is, look, you could always do a directive top down and say, "This is our technology." but most of the time people... I'm always struggling with top down. You do that and nothing else, if you're not convinced.

But what Axulus provides is actually, if the people understand that with Axulus, they can very easily and quickly get a solution without spending a lot of workshops, a lot of effort, a lot of traveling, and so on. And it's actually also community of templates coming up. So, you can also, as a plan, very easily add a use case request, which then others can see and they can learn from each other. So, if you get that spirit together that people collaborate and get something delivered as a customized product, to be honest, then nobody really questions anymore what type of filling blocks there in. Of course, there are some discussions, but at the end the experience of, "Okay, I get a problem solved. I get across it quickly." And with a scalable approach is more important than really the last detailed technology question. That means you centrally decide, you just put in the legal bricks, which are allowed, and if the users see the value add of the overall exercise, meaning get problem solved, get them solved fast, then this discussion more or less ties out.

And you don't need a directive, you just say, "Hey, I give you the playground, you can play around with it. By the way, the toys are in." which I define, and then you set a standard without really doing or force you to do something, but you provide a clear value at also in the way you digitalize, and then, the customers or the users start to use it without having a lot of technology discussions. And I'm not against one or the other technology, there are all great technologies out there, but the only thing I would like to say is that sometimes, customers spend more on discussing basic technology features or solutions versus making sure that this technology is translated into a concrete value added shop floor. And I think that is going away with that approach, because it makes it easier to users, they use it and then the rest is just set in the system.

Niall Sullivan, Senseye: So, that was the second part of our series. Looking at the challenges of implementing industry 4.0 solutions. I hope you enjoyed it. It's clear that before manufacturers even think about the technology they'd like to implement, they must be clear on the KPIs they'd like to improve, and whether they have a culture in place to deliver scalable projects. Please subscribe via your favorite podcast provider if you'd like to be notified about future episodes. And it would mean a lot if you could let us know your feedback by leaving us a review. You can find out more about how Senseye can reduce unplanned downtime. And contribute towards improved sustainability within your manufacturing plants by visiting Thanks a lot for listening.

Subscribe to our Trend Detection podcast

We post weekly episodes on a wide range of different subjects including predictive maintenance, industry 4.0, digital transformation and much more.

Here are all the places you can access/discover the latest episodes of the Trend Detection podcast: