Digitization in mechanical engineering and in the manufacturing industry — it’s creative teamwork that goes beyond the boundaries of our company. But how does this “co-creation” actually work? A talk with Steffen Cords, Head of Ideation and Scouting at Körber Digital, reveals that the Industry 4.0 experts at Körber Digital rely on a clearly structured, fast, and stringent process.
Today artificial intelligence (AI), machine learning, and data science are taking factories by storm — at least that’s the conclusion reached by a new survey of 555 German industrial companies by Bitkom Research. The survey found that, on average, a fourth of all industrial machines are already connected with the Internet and that about 12 percent of companies use AI in their factories. However, the companies participating in the survey believed that two of the reasons why a great many Industry 4.0 projects fail are the “high investment costs” and the “complexity of the topic.” But these two challenges in particular can be mastered if companies use the streamlined co-creation process offered by Körber Digital. Before the Industry 4.0 experts launch the actual implementation phase, they gain a comprehensive understanding of the specific problem and the related data. This phase is always followed by a transparent dialogue that results in valuable insights on both sides.
Mr. Cords, many companies are noticeably hanging back when it comes to the establishment of data science and other digitization projects. Why is this happening?
There are various different reasons. In many cases, the people in charge fear that the initial costs will already be too high, that there’s a risk of revolutionary changes, or even that the production process will be disrupted. But they wouldn’t necessarily have to introduce revolutionary elements — even though new business models shouldn’t be entirely ruled out. Nonetheless, small-scale measures yield huge results. The Industrial Internet of Things (IIoT) can help companies selectively shorten delivery times and time to market and enhance a product’s quality and stability. In this case, IIoT safeguards the growth of the company’s existing portfolio.
Couldn’t the companies’ IT departments forge ahead with certain developments on their own and dispense with an extensive co-creation process?
That’s an obvious question, but it leads one in the wrong direction, because in many cases in-house IT experts are mainly focusing on safeguarding everyday processes. Besides, a very specific kind of IIoT know-how is needed in order to develop innovative digital solutions. Without this know-how, there’s a risk of implementation errors — and incidentally, such errors can also pose a security risk for the company. Against this background, Körber Digital offers a wide-ranging co-creation process. In this process, our know-how is combined with the customer’s application expertise so that together we can develop the best ideas.
How do you make sure this process works from the very start?
First of all, you need to know that we divide the shared work into several phases. The first phase is Customer Discovery — in other words, the precise study of the customer’s requirements and current processes. This is followed by a Proof of Concept phase, in which we formulate initial hypotheses and rigorously test them. Finally, when we share the opinion that an implementation will make sense, we create a Minimum Viable Product — an initial prototype that represents the later product in a first step with the initial functions. We offer the Customer Discovery phase at the beginning of the cooperation free of charge, but it’s all the more valuable, because we talk to production managers, machine developers, and machine operators about the question of which concrete problems have to be solved. In addition, we aim to find out what effects the problem is having, why current solutions are not sufficient, and whether the problem is measurable. Moreover, the topic of data also comes into focus through questions such as the following: What relevant data are currently available? How can this data be accessed? And who knows how to deal with this data? To sum up, you could say that during this first phase we acquire a clear and comprehensive insight into the task at hand and the basic prerequisites. Incidentally, we’re also building the customer’s necessary trust in our methods.
How do these insights flow into the subsequent work?
Knowing about the task at hand is one thing — but translating this knowledge into concrete recommendations for action is something completely different. This translation is exactly what happens in the second phase, Proof of Concept. In order to check the feasibility of an idea, we may, if need be, extract further data from the relevant machines and use provisional prototypes to verify our idea. At the end of this process stands the question of whether we should cooperatively launch the third phase, during which the Minimum Viable Product is created.
But hasn’t this step already been agreed on from the very start?
No, not at all. After all, we have to work together with the customer to critically examine questions such as what concrete results the use of machine learning would or wouldn’t have in this particular case. Let’s look at an example. Say you’d like to use machine learning in the future to find out within seconds what has caused a certain machine breakdown. The Proof of Concept phase shows us that the conclusion reached via machine learning can have a success rate of only 75 percent, since the quality of the data makes a 100 percent correct prediction impossible. Consequently, we determine in a discussion with the customer whether this degree of accuracy is sufficient and whether it would already lead to a significant improvement of the key performance indicators. We put all the information into visual form and present it to our customer in a format that can be quickly understood. As a result, the customer can assess the practical value of the solution.
How is the Minimum Viable Product (MVP) created in the following phase?
This concept implies a certain openness in the development phase. The software is developed with least possible expense and in a focused and manageable product size. Even though it’s still in an early stage, it is then directly applied. We basically do this in a way that allows us to refine this provisional prototype and get it ready for production. This method includes a Build Pipeline that enables us to safeguard the product’s quality and stability. We also evaluate whether added value has already been created in the MVP phase. Every solution must demonstrate its practical value in the production environment. And of course this also means that the operator and the production manager must be using, understanding, and applying the software correctly. That’s why we consider their feedback very important. Incidentally, this phase is a very good example of why we talk about co-creation — in each phase, close cooperation is absolutely essential.
Is the MVP phase directly followed by a major rollout in various production locations?
That decision is made by the customer, of course. Finally, if a significant improvement is achieved with the help of the MVP, there are two possible courses of action. On the one hand, the customer can very quickly implement the software at as many locations as possible. After that, we receive new insights regarding its use — insights that will then flow into advanced development and product optimization. On the other hand, it’s also possible that the implementation is limited to a single location. There the use of the software is further validated and workshops are used to find out if there’s a need for new functions. In that case, the aim is to make the software even more powerful by adding new elements before it’s launched at other locations.
How great is the ultimate economic effect of a co-creation?
That depends on the specific application, of course. However, we consider it very important to always provide a clear answer to this question.That’s why we define the existing situation in terms of the key performance indicators (KPIs) at the start. This definition serves as the reference value. After the introduction of the MVP and during its use, we regularly measure the KPIs. In the ideal case, we see a continuous improvement. We also continually monitor whether the operators and production planners are using the new tools in the desired manner and whether the tools’ usability is assured.
I’ll conclude with one more question about co-creation: Do your customers also basically benefit from this sharing of knowledge?
The introduction of Industry 4.0, data science, and related concepts always launches a kind of knowledge circuit at our customers’ companies. The data expertise within the company already starts to grow during the first phase of the cooperation. This in turn leads to improved data use as well as better data, and that too ultimately enhances the company’s data expertise. In other words, the company’s know-how continuously increases. Incidentally, that also happens if we jointly come to a decision that the development process should not be continued after the Proof of Concept phase. It’s always worthwhile to launch a co-creation process. For example, software companies have been forming alliances for decades. Industrial companies have to follow suit if they want to continue growing and holding their own in the marketplace.