Whitepaper: AI is widely used in maintenance, but knowledge retention is lagging behind

Of the 39 respondents, 36 report using generative AI regularly or daily. For 16 participants, this even involves intensive daily use at work.

However, this familiarity has yet to translate into widespread adoption within technical maintenance processes.

AI is primarily used for administrative tasks, summarising documents and answering questions. Its use is significantly lower for more specialised applications, such as Root Cause Analysis (RCA), technical documentation and maintenance planning.

Traditional information sources also remain dominant when troubleshooting equipment failures. Technicians mainly consult digital manuals, internal documentation systems and maintenance history records in their CMMS/EAM systems. AI-powered documentation systems with chatbots rank lowest as an information source, scoring just 1.2 out of 4.

The technology is therefore already in use, but has yet to reach the core of maintenance operations.

You can find the full white paper at the bottom of this blog post.

The departure of experienced maintenance technicians presents a significant challenge for industrial organisations. When employees leave, they take with them not only technical knowledge but also valuable practical experience that is rarely fully documented.

Examples include recognising unusual machine noises, diagnosing recurring equipment failures and correctly adjusting older installations.

Respondents rate the risk of knowledge loss at 4.1 out of 5. At the same time, they believe AI can help capture this expertise, make it accessible and transfer it to other employees.

The survey explored four possible approaches to improving the retention and accessibility of technical knowledge.

• Expert knowledge: Generative AI can help capture the tacit knowledge of experienced employees and make it available to other technicians.
• Decision trees: AI can support the creation and improvement of decision trees that guide technicians step by step through fault diagnosis.
• Digital twins of expertise: In addition to digital models of physical assets, technical documentation and human expertise can also be made digitally accessible.
• AI agents as knowledge resources: Specialised AI systems can consult relevant sources and support technicians in carrying out knowledge-intensive tasks.

AI agents in particular are generating considerable interest. Respondents rate their potential value in maintenance departments at 4.5 out of 5, the highest score in the survey.

Interestingly, however, Agentic AI is much less firmly established on corporate agendas, scoring just 3.1 out of 5.

The findings highlight a significant challenge. Maintenance organisations recognise the risk of knowledge loss and the potential of AI, but have yet to translate this awareness into concrete applications.

The white paper therefore outlines a four-step roadmap:

1. Build on existing AI applications and strengthen employees' skills.
2. Make technical documentation, manuals and maintenance history searchable using AI.
3. Launch a clearly defined pilot project, such as an AI-assisted decision tree for a common equipment failure.
4. Embed knowledge retention into the digitalisation strategy, with clear ownership and a dedicated budget.

Throughout this process, data quality, reliable information sources and human validation remain essential. AI should support the expertise of maintenance technicians, not replace their critical thinking skills.

The white paper makes it clear that the next step does not necessarily involve implementing increasingly advanced AI technology.

The greatest opportunity lies in making better use of information that is already available: technical documentation, maintenance history and employees' experience.

By systematically capturing this knowledge and making it accessible, organisations can reduce their dependence on individual experts and lay the foundations for more sustainable knowledge transfer.