Preventive maintenance aims to avoid equipment failures and unplanned production line downtime. Implementation begins by identifying the tasks required for each machine. The frequency of maintenance must then be adjusted and the people responsible assigned. Maintaining a comprehensive service history will subsequently help refine the maintenance plan with the support of artificial intelligence.
1 - Identify the maintenance tasks that need to be anticipated
These planned tasks may include routine servicing, checks, inspections or component replacement. The plan must specify the nature of each task and the component or equipment concerned. At this stage, the maintenance requirements of each machine are translated into actions that can then be scheduled at an appropriate frequency.
2 - Set the frequency and create the maintenance schedule
The frequency may be defined as a time interval, a number of units of use or according to changes in a parameter monitored on the equipment. This is known as condition-based maintenance, which is triggered according to the observed condition of the equipment, whereas predetermined maintenance is carried out at predefined intervals.
The maintenance schedule distributes tasks over time. It must also allow for periods of downtime and take into account maintenance preparation, including technician availability, access to the equipment and the required tools or parts. Following the work, an inspection or test may be scheduled before the equipment is returned to production.
3 - Identify and assign responsibility for maintenance tasks
The plan must designate the person responsible for each maintenance task. These tasks are not necessarily carried out by the same people. Maintenance teams may, for example, handle routine work. Certain inspections, tests or routine servicing tasks may, however, be the responsibility of the equipment supplier. Clearly assigning responsibilities makes it possible to know in advance who will carry out each type of task when required.
4 - Document the history of failures and maintenance work
Every failure and maintenance task provides useful information for the ongoing development of the maintenance plan:
- nature of the failure;
- possible causes considered;
- maintenance procedure followed;
- solution selected.
In an article published by L’Usine Nouvelle on 8 February 2026, Daniel Freyd, Chief Information Officer at Schmidt, explains: “For twenty years, we have been recording failures and maintenance operations, following a highly structured troubleshooting process: possible scenarios, how the failure unfolded and the conclusion.” This historical record directly supports a diagnostic assistance tool designed for technicians.
5 - Develop the maintenance plan using data and AI
A comprehensive historical record can support new diagnostic assistance tools. At Schmidt, data from past failures and machine technical documentation are also used to train a large language model (LLM). When a failure occurs, technicians can query the model in writing or by voice. They remain free to follow the proposed repair procedure or use their own judgement.
Artificial Intelligence (AI) can also help document maintenance work. Technicians dictate their reports, which are then transcribed and classified. This information can be reused when future failures occur. The accumulated data gradually expands the available knowledge about the equipment.
Lessons learned may justify changes to the process, such as strengthening an inspection, bringing a maintenance task forward or revising the frequency of a component replacement. The preventive maintenance plan therefore evolves as knowledge is acquired.