“Without reliable training data there is no reliable AI”

Editorial team: Dr Grathwohl, at the Packaging Machinery Conference you are speaking about AI as an enabler of new business models. Was there something like an aha moment for you, when it became clear: AI has what it takes to fundamentally change business models in packaging machinery?

Dr Marius Grathwohl: My aha moment came when, internally, we identified the combination of predictive maintenance technology with new business models such as “availability on demand” or “pay per use” as coherent business models. In this context it became clear to me that AI is not an end in itself, but can be a driver for a usage-based model that depends both on the types of machines and on the customer’s needs. That was when it became tangible for me how data intelligence can shift revenue logics and also why many approaches on the market do not work.

But of course, AI also has, more generally, a potential for efficiency gains that until recently was unimaginable, and we are recognising this more and more in different areas of our everyday lives.

Editorial team: Perhaps to be a little provocative: when Industry 4.0 was proclaimed 14 years ago, there was a similarly strong gold-rush mood around new business models. What is different this time?

Grathwohl: The probably most significant difference is that Industry 4.0 came with much greater freedom of interpretation: to this day there are numerous, differing definitions of Industry 4.0 and everyone understands it a little differently. At Multivac we primarily saw Industry 4.0 as an opportunity to move closer together with our customers by offering them value-added services based on the data from their machines.

With AI, in my view, things are already somewhat more concrete, as it is a specific technological field that is now gradually influencing more and more areas of our lives. We are therefore continually exploring further use cases in everyday working life that can be optimised through AI. I would therefore rather compare the current era with the dot-com era of the early 2000s – a technology that fundamentally redefines not only industrial processes, but our everyday lives as well.

“One of the most important prerequisites is openness”

AI is changing engineering less through major upheaval and more through many smart ways of easing the workload. In this interview, Dr Daniel Eckertz from Fraunhofer IEM explains where AI is already helping in mechanical engineering, which applications are mature and why openness is the key.

Philip Bittermann

Editorial team: You yourself work for a packaging machinery manufacturer, so you are very familiar with the industry. Where do you currently see the biggest challenges when it comes to AI? And where might even your colleagues still be underestimating its potential?

Grathwohl: The biggest challenge with AI is certainly to identify meaningful use cases and to really hit the mark with AI. Closely linked to this challenge is the basic prerequisite for AI, which is where many fundamentally sensible and value-adding use cases often fail: data availability and data quality. Without reliable training data, there is no reliable AI.

Then there are the familiar change management topics: employees need to understand what AI can and cannot do, and how it makes their everyday work easier. I see less of a tendency to underestimate the potential. On the contrary, it is often overestimated. Especially in management there is an expectation that all kinds of problems can now be solved with AI. In many cases the challenges are overlooked and, where applicable, the organisational preparations within the company that are necessary for AI projects to succeed are not made.

Editorial team: From predictive maintenance to autonomous process optimisation – in your view, which AI application has the greatest potential to optimise value creation in packaging machinery manufacturing?

Grathwohl: Above all, we see opportunities to massively expand our capabilities in customer service, shorten response times and even provide proactive recommendations to maintain or even increase machine availability. Of course, we are also working on implementing AI directly in our machines, as we have already done, for example, in vision applications.

In optical systems in the food sector, AI can achieve what was previously unthinkable: while classic vision systems make decisions purely on the basis of rules and are therefore overwhelmed by the variety of natural products such as sausages or cheese, AI-based systems can recognise everything that a human eye can see. For example, it is difficult to express the correct shape and appealing appearance of a pretzel in fixed rules – an AI, on the other hand, can be well trained with visually unsuccessful pretzels and correctly shaped examples. In this way, the AI understands when we consider a pretzel to be well formed.

Editorial team: The German packaging machinery sector is the global market leader. How is AI now changing the global competitive situation? Are we seeing a flying start here because our machines already have a high level of digital maturity, or does our high standard of data protection risk becoming a handicap?

Grathwohl: Compared with other regions worldwide that started experimenting with AI at an early stage, the initial focus at Multivac – as probably at many large companies in the EU – was on implementing and complying with the requirements of the EU AI Act. It was also important to us to define ethical standards for the use of AI wherever the current legal situation still leaves room for interpretation. This focus on ethics and compliance seems to me to be less pronounced in regions such as China or the USA – as a result, AI is being experimented with much faster in those regions, as fewer areas are regulated.

In the long term, however, I see this more as an advantage: Just as German engineering expertise used to be defined by precision, responsible AI can become the new hallmark of quality. Many customers, especially in regulated industries, value precisely this approach.

Editorial team: New business models often also mean new partnerships. How does the role of machine builders in their customers’ value chains change when AI comes into play?

Grathwohl: As a provider of AI functions, we want to make our customers’ lives easier and more efficient in the areas where we have points of contact and where we see our core expertise as a machine manufacturer: on our customers’ shop floors, around the topics of packaging and processing food, or in the field of healthcare and consumer products.

“Faster supply chains and hopefully less bureaucracy”

The EU–India free trade agreement is strengthening German packaging machinery manufacturing, as Richard Clemens from the VDMA explains in an interview: lower customs duties, faster supply chains and reduced barriers for SMEs. “Made in Germany” is becoming more competitive, and India is gaining in strategic importance.

Philip Bittermann

Through AI-supported service models, we can secure availability, stabilise processes and increase our customers’ efficiency. This strengthens the partnership and enables our customers to focus more on their core business.

Editorial team: Many companies experiment with AI, but the leap from pilot projects to scalable business models is rarely successful. Based on your experience, what are the critical success factors?

Grathwohl: In addition to the already mentioned topic of data availability and quality, it ultimately comes down to people themselves: If we do not succeed in bringing together the right people for the right use case, we as a company will experience frictional losses that will cost us valuable time in the competitive environment. And of course, we also need support from top management for this topic. After all, AI initially means making investments – in talent, in tools and in infrastructure. Finally, we need standards and processes in order to operate AI solutions reliably and to improve them on a regular basis. This is the only way pilots can become scalable, dependable products.

PMC 2026: Focus on new business models

Whether service, AI or other areas: packaging machinery manufacturers are currently looking for new business models. As a source of revenue and a differentiating feature from global competitors. That is why, in the third edition of the Packaging Machinery Conference, which will take place on 16 and 17 June 2026 in Nuremberg, we have several presentations on this topic on the agenda.

These include the presentation by our interview partner Dr Marius Grathwohl, who will address the question: “How does AI become an enabler of new business models?”

What and who else can you expect this year on the platform of the German packaging machinery industry? We are currently publishing our speakers and their topics one by one at https://www.packaging-machinery-conference.de/