How industry can accelerate its development processes

The challengers from China and other regions do not stop at the automotive industry. The consequences are measurable: “Where Germany still held 18 percent of the global mechanical engineering export market before 2020, this share fell to 13 percent by 2025. In the same period, China expanded its export share from 15  to 20 percent and is increasingly displacing German providers from the Chinese domestic market as well,” reports Ingo Neumann, Managing Director at Unity China. Chinese champions are being systematically forged through scaling and a government-supported “manufacturing operating system” optimized for speed that accepts waste as the price for market dominance. Anyone who cannot keep up with the Chinese pace slips to the margins.

However, the need to reduce development times and costs has already reached the executive floors. “For us, all customers are the same - whether Europe or China. But we have to keep up with the speed, especially in China,” says Rainer Eidloth, Senior Vice President Engineering Digitalization & IT at Schaeffler, for example. Efficiency can only be achieved through digitalization. “An important point is the data flow: Data must be available and quickly accessible. Only in this way can we achieve the necessary speed,” says Eidloth. But the transformation of product development is complex. Although a great deal has happened technologically in recent years with digitalization, IoT, and the digital twin. But now a change in the entire way products are developed - with AI - is on the agenda.

Among the biggest changes is the shift toward the software-defined product: More and more functionality is moving to the software level. The automotive industry and the aerospace sector can be regarded here as pioneers in engineering, from whom other industries can learn a great deal. In order to be able to manage the complexity at all in the future, the concept of Model Based Systems Engineering (MBSE) has become established. In this approach, requirements, circuit diagrams, and documents are managed in a shared system model in order to be able to simulate functions at a very early stage. Systems Modeling Language SysML plays an important role here in describing complex technical systems: software, mechanics, electrical engineering, electronics, software, but also the relationships between all these components. Around 60 percent of companies in mechanical and plant engineering currently develop a large portion of their software themselves, the VDMA study  “Software Development in Industry - Status Quo and Outlook” finds.

AI is currently revolutionizing product development

The gamechanger, however, is AI, which on the one hand makes design suggestions, but above all greatly shortens requirements management. One example: By integrating the AI software of the Finnish startup Raiqon, BMW achieved a considerable reduction in search times for requirements. According to this, 60 engineers previously needed one day to find an entry in hundreds of thousands of document pages; now semantic search does it. AI also helps with change management by tracking down the effects of a change in the overall context, which until now has been extremely labor-intensive manually. This is particularly relevant for companies that develop change-intensive individual products together with their customers.

The exchange in the product development community at the annual meeting of the Prostep Ivip association with 650 participants showed: Many companies are already on their way and have developed solution approaches in the software-defined product, credible simulation, and AI-supported requirements management. In the trusted community, innovations and strategies are discussed with great openness.

"As Prostep Ivip e.V., we are increasingly noticing the desire of additional industries such as mechanical engineering to exchange views on the pressing topics of product creation. We consciously want to open ourselves to this in order to share knowledge and standards," says board member Philipp Wibbing (Unity). Because the shortening of development times, the individualization toward customers, and dealing with costs in change management as well as in complying with governance are becoming more critical for mechanical engineering, medical technology, aerospace, and defense as well in economically difficult times and under growing innovation pressure.

Leadec: why industrial service is becoming the backbone of the factory

Industrial service is increasingly determining efficiency, flexibility, and availability in production. Leadec CEO Markus Glaser-Gallion explains why data and AI are strategically upgrading service.

Software-defined products are coming

"The complexity is growing enormously: Many product variants, long life cycles, stricter regulations. Old products have to meet new requirements - that is hardly manageable with paper anymore," notes Dr. Thilo Jania, Global Head of R&D Center of Development Excellence at the Japanese medical technology manufacturer Olympus, for example. Therefore, the company is moving in the direction of digitalization and AI. At the same time, the signed drawing is often still the only "single source of truth."

For companies that now want to enter the defense sector, the focus is on software-defined defense. If mechanical engineering companies want to work as suppliers for European armaments or aerospace programs, they are not only component suppliers, but part of a highly complex overall system. For that, classic CAD drawings and bills of materials are often no longer sufficient.

Large system integrators such as Airbus, Dassault Aviation or Rheinmetall are now increasingly developing on a model-based basis. Their suppliers must therefore be able to provide requirements, interfaces and system descriptions in compatible models. “We work with real models and not just with diagrams or PDF documents. Therefore, it is crucial to enable all parties involved to access these models. The key to mastering this complexity lies in model-based work and in the early verification of hypotheses through simulation of the battlefield and the overall system,” reports Dr. Jörg Wirtz, Head of FCAS Common Working Environment & PMT program at Airbus Defence and Space.

Model-based development and sharing of data

In Boeing's complex supply chain, the design is worked on with many suppliers in different ways. In some cases, parts are built according to specifications; in some cases, there are co-design processes. “The problem: We are still working heavily on a document-based basis. The challenge with lead times lies in arriving at a completed design and then continuing to work efficiently,” notes Kenneth Swope, Senior Manager Enterprise Interoperability Standards & Supply Chain Collaboration at The Boeing Company.

From his point of view, the answer lies in moving from a document-centered to a model-based approach. To that end, the aim is to increase digital competence internally and in the supply chain.

Engineers had previously been trained to create models with a view to isolating the data and to carry out engineering in order to solve their respective problem: “All this data was structured for years - and still is today - precisely in this context. But now we want to turn the equation around: We want to disclose the data and make it relevant across domains,” Swope explains of the paradigm shift.

At present, then, traditional approaches and data models are meeting technology such as AI agents, which thrive on context. For this, data must be structured and provided in such a way that it is semantically understood and new standards such as  Model Context Protocol (MCP) can take effect. “That means a transformation and reskilling of the workforce and a transformation of the company’s own approach,” Swope concludes. MCP is an open protocol that is currently gaining strong importance because it enables AI assistants to access external data sources and applications in a standardized way without having to build an interface each time: essentially a kind of USB-C for AI.

Semantically understanding data and unlocking contextual knowledge

Simulation continues to gain importance, closely linked to the development toward the Industrial Metaverse. "Simulation is always also about simulation credibility, in order to be able to make decisions without hardware through virtual testing," explains Hans-Martin Heinkel, spokesperson at Robert Bosch GmbH and head of the Smart Systems Engineering (SmartSE) project at Prostep Ivip. For that, information is needed, meaning the harmonization of metadata, semantics, interfaces, and data formats. "There is no way around aligning these four points; that is a painful lesson that all companies have to learn," says Heinkel. MBSE and simulation are mutually dependent on each other.

Data is very clearly becoming the key to innovation in product creation. The experts also agree that LLMs cannot realize their potential without contextual knowledge and semantics. Data is indeed available in PLM/PDM systems, in ERP, and in MES. Here, Eidloth sees the challenge as companies having to clarify how their software suppliers will enable access to the data through AI.

Because by no means have all providers already opened up to this. The association is therefore calling for consistent standards with the “Code for PLM-Openness.” But there is another hurdle: The engineering knowledge behind the decisions as to why a product was designed exactly this way and not another does not lie in the structured data of these systems, but in countless PowerPoint presentations, Excel and Word documents.

What advantages agentless access offers in OT

Why is agentless access becoming relevant for OT systems? In the interview, Felix Wirthmann, Director Product Management at TeamViewer, explains how secure remote access to OT systems works without software installation at the endpoint.

Mapping long product life cycles in data terms

Aviation, for example, has very long product cycles and works with data over decades. It is similar in mechanical and plant engineering. “Our data goes back decades, but language, rules and documentation have changed. The problem: The data does not speak for itself,” says Swope. It is therefore important to examine the “vocabulary” closely and understand how terminology is used in all areas of the company to ensure that the product, its integrity and its safety always remain at the center.

The EN/NAS 9300 series of standards emerged from the Prostep-Ivip project Lotar (Long Term Archiving and Retrieval), which was originally driven by aerospace members. Lotar ensures that 3D CAD, PDM, simulation and now also MBSE data (artifacts) are still readable and reusable decades later - regardless of the original software manufacturer and changes in data formats. For mechanical engineering companies that use SysML or model-based development, this now ensures that not only CAD but also system models are preserved in the long term. With Lotar’s standardized information models and processes, companies in other industries can also build a continuous digital thread (Digital Thread) that extends from engineering through manufacturing all the way to service.

Driving innovation with AI

Medical technology is heavily regulated worldwide with regard to patient protection and data protection; many requirements must be complied with for market approval. Olympus primarily manufactures medical endoscopes, a highly innovation-driven field. “It is extremely important for us to bring new products and new innovations to market quickly. Because of the regulatory requirements, this is an additional challenge,” says Jania.

The goal is to expand the functionality of the products and, in the long term, to broaden the range of applications through robotics. AI is thus being incorporated into the products, for example for computer-assisted diagnostics. Software is increasingly providing expanded functionality and individualization according to patient needs - medical technology is thus becoming a software-defined product.

“Our vision is to have a fully closed patient pathway - from detection of the disease through treatment all the way to reprocessing the products,” explains  Thilo Jania. This includes sterilization, for example, which is subject to particularly strict regulations and is therefore very labor-intensive.

Dealing with increasing regulation

“The diversity of the different regulations and local particularities makes it nearly impossible for anyone to have the complete overview. That is why AI will help us monitor the development of regulations and extract the correct interpretation with regard to risk minimization,” the expert reports. LLMs are particularly suitable here, because the regulatory affairs process in medical technology is heavily text-based. Here, work is being carried out together with the consulting company Unity, which has developed a tool to describe the Intended Use of a medical product. At the same time, it identifies which norms and standards must be taken into account in the development process. They can then be incorporated into the company’s own requirements management system in order to continue working with them in the development process.

“Since regulation requires us to validate every kind of algorithm, there will be no automatic approval of these data and no automatic decision-making,” says Jania. Nevertheless, AI is “enormously helpful,” as long as there is a guided process and, at the end, a human grants the approval. In his view, AI is also indispensable for preserving the knowledge of highly specialized skilled professionals who will retire in the next few years.