50th FVI Expert Breakfast

Topic: "50 Topics. 50 Perspectives on Maintenance." A retrospective on two years of expert breakfasts and a look ahead at the next major challenges in the industry.

After 50 episodes, this time there was no single technical topic in focus. Marcel Hahn used the anniversary episode for an open assessment: What has actually improved, which problems have been accompanying maintenance for years, and which developments are still underestimated?

• AI is a tool – not the universal solution: The discussion made it clear that the view on AI has become more professional. Instead of chasing every hype, the focus is on concrete application: "What do I actually need?" and "Which problem do I want to solve?". Especially smaller and medium-sized companies experiment pragmatically with available tools, while larger organizations often get stuck on IT approvals, data access, and complex integrations. At the same time, humans remain indispensable: results must be understood, evaluated, and validated. AI is meant to support, not replace thinking.
• Securing knowledge without creating new data graveyards: The looming loss of experiential knowledge remains one of the central problems of maintenance. The discussion about disturbance knowledge was particularly practical: A technician on the late or night shift may hardly know a rare plant and needs access to previous causes and solutions right then. At the same time, there was a warning against simply collecting huge amounts of data. The crucial question is therefore not only whether information is stored, but whether relevant knowledge can be found and utilized at the right moment. Videos, voice recordings, and automatically generated transcripts were explicitly discussed as pragmatic forms of knowledge documentation.
• The business case decides on digitalization: A participant described a typical situation from the mid-sized sector: The main systems already achieve more than 97 percent availability, while a specialized obsolescence solution causes significant ongoing costs. This raises the uncomfortable but crucial question: What do the last additional percentage points actually bring? Digitalization must be weighed against downtime costs, spare part risks, and total cost of ownership. At the same time, the conditions are changing: Service providers are disappearing, successors are missing, and previously readily available repair services are becoming rarer. The result is higher spare part inventories and additional capital commitment.
• Bureaucracy and lack of decision-making leeway become a maintenance problem themselves: The frustration over internal and external coordination processes was particularly evident. While investments in large systems are possible, even a software license for a few euros per month can trigger extensive approval processes. Even a camera for disturbance analysis can be blocked for weeks by data protection, works council, and internal rules. At the same time, the responsibility for system availability remains with maintenance. The pointed insight of the round: Maintenance managers need a "strong backbone" to distinguish necessary rules from mere formalism and to advocate for pragmatic decisions.

Classification: From data and AI tools to Operational Intelligence

The real problem of maintenance is no longer the lack of software, but the question of how knowledge, data, and digital tools can really be utilized in everyday work.

• Knowledge disappears and documentation takes time: Experiential knowledge today resides in heads, individual disturbance reports, videos, meetings, or unstructured documents and is often missed only when the experienced colleague is no longer available. This is exactly where we start with ADAM. The Digital Lifecycle Record consolidates information asset-centered and turns individual reports, documents, photos, and solutions into a usable treasure trove of experience. Instead of requiring perfect master data, we pursue an ADAM-First and Brownfield-Native approach: With hybrid search and AI, relevant knowledge can be found even where historically grown data structures are anything but perfect. ADAM thus becomes the digital vault for the experiential knowledge of maintenance.
• Employees should document – but don't want to type: The episode describes exactly the known dilemma: Better analyses require qualified disturbance causes and solutions, but at the same time, every additional input form means more administrative burden. We turn this principle around with ADAM. Through reporting and feedback assistants, technicians can speak naturally instead of filling out forms. ADAM structures the information, assigns it to the correct asset, and passes it on to existing systems like SAP if needed. Thus, the documentation obligation becomes an almost incidental transfer of knowledge. The human remains human-in-the-loop, ADAM takes over the repetitive work in the background.
• Digitalization fails due to complexity, costs, and dependencies: The discussion about expensive special solutions, changing providers, follow-up costs, and lengthy approvals shows why large big-bang projects are increasingly viewed critically. Our approach is therefore deliberately pragmatic: ADAM is a SaaS platform for Operational Intelligence that complements existing systems instead of replacing them. Through open integrations and no-code approaches, specific bottlenecks can be resolved step by step without having to rebuild the entire IT landscape beforehand. Instead of promising a supposedly perfect solution for the next 20 years, we start with a clearly defined problem, generate measurable benefits, and develop the process further together. The asset-centered Digital Lifecycle Record remains the long-term basis – also for retrofits, compliance, and safety.

Conclusion: The next 50 episodes will probably be less about whether maintenance needs more technology, but about how it can use technology so simply, economically, and human-centered that it actually helps in everyday life, and this is exactly where ADAM becomes an enabler for Operational Intelligence from an additional software tool.