On the Shop Floor
Jean, 32, has worked in maintenance for 8 years. He has no IT degree. But he knows every machine in the plant by heart. He can tell by ear whether a compressor will fail within 48 hours. He also knows that the pressure switch on compressor #5 has always thrown false positives in December.
Now you are being offered an AI to "automate" maintenance. Except that the AI does not know Jean. It does not know that true maintenance expertise is exactly what Jean knows: the context, the subtleties, the tacit know-how.
This is where most AI projects fail. Mimorian is an industrial intelligence platform that models equipment, structures failure diagnosis and captures the know-how of maintenance teams through a multi-agent AI architecture. It was designed to make the most of this field know-how rather than bypass it.
Tacit expertise: the knowledge that cannot be written down
Studies on enterprise knowledge management (Panopto, 2018) estimate that the majority of critical knowledge remains tacit: experience, awareness of context, learning through practice. This knowledge is rarely documented. It lives in the minds of the veterans.
Concrete examples:
Climatic and seasonal context. A piece of equipment behaves differently in summer and winter. The manual does not say so. Jean knows.
Local machine specifics. "The variable speed drive on line #3 was moved in 2021 when we redid the cabinet, but the electrical diagram was never updated. If you follow the documentation, you look in the wrong place." No sensor flags this gap between the documentation and the actual installation.
Non-obvious troubleshooting sequences. "To reach the valve, you first have to remove the air filter, not go straight in as the manual says." No sensor can learn this.
Non-linear cause-and-remedy relationships. "If it is the differential pressure that changes, it is 60% the filter, 30% the check valve. But after 5 years in a humid plant, the probabilities change."
Logistical trade-offs. "We could order the premium part at €3,000 (8 weeks), or the standard version at €800 (2 days) and reschedule a preventive task."
Why AI needs this expertise
The more sophisticated your AI is, the more it needs human expertise to interpret it. The sector feedback documented by McKinsey (2023) shows that industrial AI projects which maximise human-machine collaboration achieve far better results than those aiming for full autonomy.
How Mimorian captures field expertise
Step 1: Equipment walkthrough with the senior technician
A natural conversation with Mimorian about everyday matters. "Look, last week we had a faulty contact on the unwinder: the operator tends to lean on the edge of the guard housing, it shifts slightly and it disconnects the sensor cable over time." The patterns emerge.
Step 2: Collaborative structuring
Mimorian proposes a structure: "Here is a guided diagnosis for a pressure drop. Correct? What would you add?" The technicians validate, refine and prioritise.
Step 3: Enrichment through local context
"Does this guided diagnosis work the same way in winter and in summer?" The answers tailor the diagnosis to each context.
Step 4: Capturing field feedback
Every intervention enriches Mimorian. The field feedback adds to the previous entries. The diagnostic process gradually becomes more intelligent.
The multiplier effect
After 6 months, you have 200 structured field feedback entries. You hire Thomas, 22, a beginner. Before Mimorian, he would have learned over 2 years, dependent on the availability of his expert colleagues. After Mimorian, he gains autonomy far more quickly than through oral transmission alone, which lets one of his colleagues mentor him more effectively.
Organisations that structure tacit knowledge into guided diagnostic processes considerably accelerate the learning curves of newcomers.
The virtuous circle: happier technicians
Their expertise is finally valued. Instead of being threatened by the AI, they create it. The result: better engagement, less turnover. The impact on retention and recruitment costs is direct.
Conclusion: Your technicians are your best data scientists
The secret to successful AI in maintenance is to stop seeing technology as the source of innovation. The source is the people. Capture this expertise. Listen to it. Document it. Share it. That is trustworthy AI in industry.
Ready to turn your technicians into AI experts? Contact Mimorian to talk through your context.