Like many other industries, the machine engineering industry is currently undergoing digitalization. Many questions arise in this context: What digital innovations are we encountering in this sector? Do digital service models like "Predictive Maintenance" or "Condition Monitoring" directly generate new revenue, or are they merely tools that create new customer value only when applied correctly?
Sebastian Fuhrich from Putzmeister, Lena Weirauch from ai-omatic, and IIoT expert Christoph Moisel provide exciting insights into the future of AI-driven machines and the digitalization of plant and machinery construction in this interview.
Is it more difficult for the machine engineering sector than for other industries to engage with digital innovation, and if so, why?
Christoph Moisel: Yes, it is. I believe the major challenge with digital innovation is that the technology is largely new to many companies. Besides the technology, there is the significant issue of "culture." This means mobilizing internally to alleviate fears surrounding "data security" and at the same time translating this into new business models. These are three major fields that signify innovation. This makes it difficult and is not just a technical innovation but major transformation on many levels.
Lena, in 2020, you founded ai-omatic with two friends and a software company that focuses on making machines smarter, specifically addressing the topic of "Predictive Maintenance." Can I, as a machine builder, simply connect to your system on my machine, and "voilà," "Predictive Maintenance" is there?
Lena Weirauch: I wish I could say “Yes.” The reality is not so straightforward. One may have the best algorithm for this evaluation, but the first step, which is crucial for most companies, is to make available the data needed for such processes. Once that's done, a lot has been accomplished. The goal of ai-omatic is essentially to offer a method for the area of "Predictive Maintenance." However, it still requires various adjustments. I always say one cannot expect a method that works for a wind turbine to also function in the automotive sector or shipping. These are different applications, and I believe, even in ten years, there will still be no solution that works in a "plug-and-play" manner across these various industries.
Christoph, you earlier suggested that "Predictive Maintenance" might simply be a bubble and mentioned three essential topics that a machine builder must tackle to enable new business models. Plug and play is not a viable option, as we just heard. So how else can this be achieved?
Christoph Moisel: “Predictive Maintenance” is on everyone's lips, and there are countless possibilities and enormous potentials, which I certainly do not deny. However, what I observe is that most of the customers we had in our company eagerly want "Predictive Maintenance" and consider it somewhat a given, but their willingness to pay is significantly challenging. This is partly because it is relatively difficult to calculate for many companies that are not yet very digital. It is crucial for machine builders or providers to engage in dialogue with the producing companies and try to establish priorities under the aspects of “Where can we get the most benefits?” and “What might not be so abstract at first?” “Predictive Maintenance” involves discussing a fault condition that might occur. The question arises, once it occurs, how I can identify it beforehand. In many discussions on similar topics, I learned from other machine builders that, for example, topics like “Predictive Quality” or “Optimization of Throughput” were simpler to communicate. A classic salesperson finds it easier to say, “I will make the machine a bit faster” or “the quality will improve” than to discuss unfamiliar topics like “Predictive Maintenance,” which makes it even harder to sell. It is a journey on many levels.
Is there really demand for Predictive Maintenance, and are the data the drivers, or is it the customer? To rephrase: Does the customer actually want it, and are they willing to pay for it?
Sebastian Fuhrich: Our customer is in a vicious cycle today. We have customers who sometimes have parts worth ten or even twenty thousand euros in stock, which affects cash flow. We have the customer who has already independently reduced maintenance rhythms we suggested and consequently changes parts more frequently. Every little step we provide is positively received and definitely creates added value, even if it is merely a preventive measure and not yet predictive. Additionally, when we talk about “Predictive Maintenance” with our customers, the understanding of what we are discussing is often absent on their side. Hence, yes, even small steps definitely bring customer value that customers are willing to pay for. Now you are, of course, selling a suggestion to replace a part. From my perspective, this isn’t sold as a service, but the customer is indeed very willing to buy the part. You often have the discussion about whether to buy an original part or a counterfeit part. This provides us with another significant leverage point, where we can potentially prove that the original part makes significantly more sense since we can only sensibly track the original part and do not want to or cannot track the counterfeit part. Hence, we see many opportunities to generate additional revenue.
Let’s say we have a machine and need to extract data from it. Lena, how do you approach this with “Ai-omatic,” and what tips do you have for others dealing with this topic?
Lena Weirauch: If you are investing energy into “Predictive Maintenance,” you should first think about where it is financially worthwhile. Once you have answered this question, you can discuss the topic, whether with an external company or in-house. There are certainly several factors to consider, like whether data exist that provide information about wear. If data are queried only once an hour, no alterations will be visible. The data must have appropriate quality. I would recommend considering the use case where it truly costs money; if you want to approach this topic, bring along a particular mindset and be ready to face several hurdles that may arise. I have yet to see a single “Predictive Maintenance” project fail when the individuals truly expressed their enthusiasm and desire to tackle it. The projects that have failed are usually those where the motivation was absent from the outset. Therefore, if you decide to approach this topic, bring the right energy and it will become successful.
Christoph, if you could ideally form a team that deals with digital innovation in machine engineering, what competencies would you assemble?
Christoph Moisel: When I speak on this topic, I often show a picture of a bridge under construction. On one side, there is industry and medium-sized machine builders, the producers, etc., who stand with all their engineering know-how and especially domain knowledge – the experts who work directly on the systems or are heavily involved in service. I believe we need to have a few engineers who genuinely understand the machine in-depth, are well versed in application technology, and know the typical problem cases. On the other side, we need data experts. There is a substantial question of whether to bring them in from outside or build them in-house. What I find very central is that the bridge is under construction and in the gap, which I often see, sit those who mediate between the data scientists and classical industry or the start-up world and industry. They who understand both worlds to some extent, see the challenges, and can ensure both worlds speak the same language, integrate and harmonize. I believe that is a very essential intermediary role that must be present in the entire interaction of various experts.
How do I coordinate the different experts? Sebastian, you mentioned human “Condition Monitoring,” meaning experimenting and stating, “Let’s conduct two or three manual processes simply to understand if it works.” Is that well received? How did you achieve acceptance within Putzmeister AG?
Sebastian Fuhrich: I believe the major challenge was that you definitely need users who possess this knowledge within themselves. You then need a good feel for it and a very straightforward way to extract this knowledge from employees. We learned to approach people simply and, ideally, sit next to them for a day to see how they tackle such problems, what they do, and what they inquire about to eventually speak on equal footing, despite having minimal knowledge of the topic. The goal is to turn the camera on them through this human trust. At least for us, once this opened up, they implemented something, appreciated it, and then it gradually flourished. Then the machine started to run a bit, and I believe this is a starting point that must be made as simple as possible so that people genuinely join in. That was the pivotal point for us.
Lena, what is your experience when interacting with traditional machine builders? Can you learn from each other in a culture that initially does not match with yours, and what is the challenge?
Lena Weirauch: In our case, we often tread on already burnt ground. I must say, “Predictive Maintenance” is a topic where many have tried something, and then the 18th company comes along and claims, “But now it really works.” They have already heard that 17 times before, and persuading them is not always so simple. We solve this by always initiating a “Proof-of-Concept” with each company. We say, “Give us the chance; if you have data, let us prove that we can help you.” Once that’s happened and we can show in black and white, preferably using historical data, and the company knows there was a failure, which with our system would have received an alarm eight hours in advance, it certainly becomes evident to most and then acceptance is there.
Is the fire department support from employees also provided remotely, and are live machine data utilized?
Sebastian Fuhrich: I will try to answer two questions simultaneously. Our customer calls the operator on-site, who typically first calls someone within their office, where one or another technician is sitting. Previously, everything was done via phone. Now, we have this approach where we can look at the data, meaning we have at least one unidirectional connection where our service technician can see what has occurred with the machine. This is quite beneficial since they receive the entire history of fault messages. On that feedback loop, we are currently working on providing emergency responders with better tools available to genuinely connect to the machine, make changes, alter parameters, and deactivate sensors. Regarding video support, we have not implemented that yet. The significant problem at that time was the glasses we used. Those glasses tend to disappear or become damaged, leading to two issues. I expect you to send me new hardware without fault. This is naturally your product, and it’s now broken, and I obviously want to hold you accountable for it. So at that time, according to the technological conditions, that didn't provide the added value we hoped for.
How can a service like "Condition Monitoring" or “Predictive Maintenance” actually be sold? Christoph, what is the chance of successfully bringing a digitally innovative service to the customer?
Christoph Moisel: One must genuinely think anew, and corresponding business models must be incorporated. I believe this originally comes from energy management, where you say: “I will extract more from your processes, and we will share a portion of the performance gains you achieve. We track this, you take zero risks; we are convinced we can extract something more from your process, and we will take a little cut of that, making the risk lower.” I believe these are concepts you have to reconsider to approach the outer world, ease the client's fears, and potentially reduce barriers.
Sebastian, how do you view this? You mentioned that you already have 100 paying customers for your cockpit. Was that easy?
Sebastian Fuhrich: It is a massive challenge. In our case, the sales force is not 23; even grasping the product itself can be quite daunting. Sales is ultimately often driven by volume. If I sell a machine for half a million or a digital product that I must first understand myself, it’s clear which one sales will seize first. In the end, we managed quite well because we genuinely included target agreements with sales management. That helped us significantly.
Do you have a tip in that direction, Lena?
Lena Weirauch: I see that it is indeed possible to generate revenue from it, especially since the machine builders we speak to are actively being requested by production companies. Sometimes, machine builders reach out to us because they were approached by clients. One should consider whether to take on this topic or not, as it is certainly quite complex. If you possess the competencies internally, it makes sense to do it in-house. I can share that we see machine builders who profit financially from improving their value proposition. They promote lower downtime, which naturally gives them a competitive advantage.
Now, one can sell the machine with the service model or the service model as after-sales service. Do you see the future of machine engineering in platform business models?
Christoph Moisel: Yes, I genuinely believe that platforms will play a major role. Currently, there are numerous island platforms, fragmented across the value chain. The significant point for me here is “cooperation” and “cross-platforms.” With a vision of comprehensive transparency over products, etc., it will enable platforms in this area to grow, as it will allow data to be aggregated across the board, thus deriving numerous additional values. However, I believe the journey is still a bit long.
Lena, are you considering integrating services into platforms?
Lena Weirauch: Yes, definitely. I think there is always a present issue of data security. That is, when people hear the word “platforms,” they generally are not very enthusiastic. I believe that an attitude shift will need to occur. I understand that if it involves highly sensitive data, one may not be eager to share it outside the company. However, there may be less sensitive data where a platform or cloud environment does not seem so dreadful. I can envision that this will be the future.
Lastly, Sebastian, do you have a platform concept ready?
Sebastian Fuhrich: Yes, we do have platform ideas, but it is definitely the pinnacle of challenges for various reasons, including the competence required to build such a system. One also has to be willing to concede some control. This means that with such topics, we must genuinely transition to a true partnership, which is inherently difficult in the real world. Finding a platform that generates enough value for everyone to participate is indeed tough. I believe this would probably be the golden path, but it would involve significant work to arrive there.
Great! Thank you all. It was a very engaging discussion. At this point, I’d like to remind you that we have established our Innovation Club. We warmly invite all those involved in companies focusing on digital innovation. We organize exchanges on specific expert topics. You are welcome to contact me directly to potentially become a member, and we can provide you with additional information.
The interview was conducted by Ulf Valentin as part of the series “Digital Entrepreneurial Spirit | The Conversation” on November 17, 2021, on the topic of "Digital Innovations in Machine Engineering." The questions and answers presented here are summarized excerpts from this discussion.
About the Experts
Sebastian Fuhrich: Head of the Putzmeister Innovation Factory. Sebastian develops digital services here.
Lena Weirauch: Psychologist, CEO & Co-Founder of ai-omatic solutions GmbH, a German AI startup offering an innovative Predictive Maintenance solution.
Christoph Moisel: Freelancer for all matters related to the Internet of Things & Digital Transformation.
