'/%3E%3C/svg%3E){kind=small}
How Companies Can Put AI to Practical Use: Case studies demonstrate how to make information available more quickly, intelligently link applications, and streamline service processes.
Artificial intelligence is becoming increasingly important in the mechanical and plant engineering sectors. At the same time, many companies are asking how they can identify specific use cases and successfully integrate them into existing processes.
The online event “AI Solutions for Mechanical Engineering – NRW,” organized by ProduktionNRW in collaboration with VDMA Software and Digitalization on October 29, 2026, provided practical answers to these questions. Using concrete examples, the event presented various potential applications of AI as well as key prerequisites for successful implementation.
AI Activities at the VDMA
Carsten Rückriegel, a specialist at VDMA Software and Digitalization, provided an overview of the current state of AI use in mechanical and plant engineering. The results of the 2025 VDMA survey presented show that AI is of medium to high strategic importance for more than 80 percent of the companies surveyed. Forty-three percent are already using relevant solutions, while another 48 percent plan to implement them by 2028. A preliminary look at the 2026 survey suggests that the actual use of AI has since increased significantly, and that AI agents, in particular, are gaining in relevance.
Both last year’s survey and the current results show that the leading sectors in AI adoption are software development, marketing, and communications. Companies associate AI with, among other things, reduced staffing costs, increased employee motivation, and shorter process times. At the same time, hurdles remain: a lack of human resources, unclear economic benefits, and shortcomings in strategy and data quality complicate implementation. Successful projects therefore require the support of senior management, interdisciplinary teams, clearly defined use cases, reliable data, and the early involvement of employees.
AI Integrates with Business Applications
Stefan Kirch, Managing Director of Bauer Kirch GmbH, used the example of resource scheduling to demonstrate how language models can successfully exchange data with other business applications via the open Model Context Protocol. The standard acts as an interface between AI and existing software systems. This allows the AI to retrieve current data directly from line-of-business applications and, provided the appropriate permissions have been granted, to create tasks in other systems as well.
In the demonstration, the AI reviewed planned international assignments for service technicians and identified missing A1 certificates or those with insufficient validity periods. Subsequently, the necessary tasks were automatically prepared in a connected HR system. This example illustrated how AI can consolidate information from multiple applications, take business rules into account, and relieve employees of time-consuming routine tasks. The business decision and approval were deliberately left to humans.
Multi-Agent Systems in Technical Support
Florian Erbs of lise GmbH used a real-world support process to explain how multiple specialized AI agents can collaboratively handle technical customer inquiries. An analysis of approximately 1,400 support tickets showed that a large portion of the relevant information is not contained in the message text, but rather in customer photos, manuals, previous tickets, or machine data. Photos alone were included in 56 percent of the tickets examined. In nearly half of the inquiries regarding a motor, the model number was visible only in an image.
Instead of a single comprehensive AI system, a division-of-labor approach was presented: An image agent reads type plates and visible features, another agent checks the completeness of the information, a follow-up agent requests missing information, and a research agent searches existing knowledge sources. Finally, a response agent creates a well-documented draft. Clear responsibilities make errors easier to trace. Low-risk tasks can be automated early on, while technical responses should initially be reviewed and approved by a human due to potential follow-up costs.
AI Strategy for Technical Documentation
Christopher Rechtien, a management consultant at kothes GmbH, presented a structured approach to the use of AI in technical documentation. Technical information is far more than just an operating manual: it supports commissioning, operation, maintenance, troubleshooting, the spare parts business, and training. In the future, AI-based assistance systems will be able to provide users with context-sensitive answers without requiring them to search through extensive documents.
However, this requires a robust knowledge base. Incomplete, contradictory, or inadequately tagged content leads to unreliable results, even with powerful language models. The strategic approach presented therefore comprises five stages: structuring content and terminology, reviewing existing systems and expertise, testing initial use cases in a controlled environment, defining measurable benefits, and establishing clear rules, roles, and approval processes. Approved tools, the protection of sensitive data, and the expert review of AI results are particularly important.
Identifying Replacement Parts via Photo
The presentation concluded with a demonstration of a visual search solution by Justus Lnyris[CR1.1]. It addresses a typical challenge in the spare parts business: If the part number is unknown or no longer legible, service staff often have to search through manuals, drawings, and catalogs or consult additional experts. The presented solution makes it possible to identify a spare part based on a photo and link it directly to the corresponding catalog entry.
In addition to existing product images, three-dimensional design data can also be used for this purpose. The system generates synthetic images from this data, which are used to train and improve the visual search. Companies can either deploy the solution via a preconfigured search interface or integrate it into existing service, sales, and retail applications via APIs. This reduces search times, eases the workload on service departments, and simplifies the ordering process for replacement parts.
Conclusion: AI Creates Added Value Through Clearly Defined Applications
The event demonstrated that the benefits of AI do not depend solely on the performance of a speech or image recognition model. What is crucial are clearly defined use cases, reliable and structured data, suitable interfaces, and clearly defined responsibilities. Particularly great potential arises where AI supports employees in time-consuming search, verification, and routine tasks while being able to directly access the necessary information in existing systems.
At the same time, it became clear that sustainable implementation cannot be achieved through technology alone. Companies should start with manageable pilot projects, make their benefits measurable, and gradually expand the scope of AI’s operations. Expertise, quality control, and the final human decision remain essential components of successful AI applications.
Organizer
The event was organized by ProduktionNRW in cooperation with VDMA Software and Digitalization. ProduktionNRW is the cluster for mechanical engineering and production technology in North Rhine-Westphalia and is managed by VDMA NRW. ProduktionNRW serves as a platform to connect, inform, and promote companies, institutions, and networks with one another and along the value chain. Significant portions of the services provided by ProduktionNRW are funded by the Ministry of Economic Affairs, Industry, Climate Protection, and Energy of the State of North Rhine-Westphalia.