In organizational theory terms, the popular “procedures driven maintenance” treats maintenance as a “machine bureaucracy” with a primary emphasis being “procedures.” Yet, even with the ever-advancing use of AI, maintenance would still operate more effectively as a “professional bureaucracy” with the primary emphasis on “craft skills.” This article contrasts procedures with skilled craftspersons in the light of AI to show how best to leverage maintenance planning. Enjoy the semantics.

A machine bureaucracy describes the best organizational form for an assembly line. Mandated rules oversee the operation of assembly lines, providing the production of each finished product correctly and consistently. There is no judgment involved in any step of the assembly. An assembly line person or robot correctly installs each part with the correct tool, each being always available. The technology and environment are factors that allow mandated rules to dominate. That technology and environment are unchanging and may be complex.

Maintenance is not an assembly line

However, both the technology and environment of maintenance differ from that of an assembly line. Maintenance technology changes, though slowly. Better tools with better power sources, electronic testers, and inspection aids become available. Different manufacturers offer better choices of assets or parts. The environment changes as well. Work on similar assets have different accessibility or weather conditions. Preferred parts or tools might not be available. Other site or asset conditions might not be as expected. 

For maintenance with its complex and changing technology and environment, a professional bureaucracy gives the best organizational form. Staffing should dominate the operation to provide the best results. Staffing includes hiring, training, and retaining of skills. In maintenance, on the spot craft skills are critical for assessing and judging maintenance situations.  

So where does maintenance planning fit in? In one case, for a procedures-driven maintenance arrangement, the expert planning group makes the rules for the craftspersons. Expert planners assess maintenance situations and provide proper procedures for the crafts. Crafts are (theoretically) strictly mandated to follow the procedures to provide consistent and proper work execution. (In practice, obviously craftspersons can exercise some judgment. Craftspersons can also recommend helpful future plan improvements.) 

In the other case, for a staffing-driven maintenance arrangement, the planning group supports the expert craftspersons. Planners assess maintenance situations as best they can and provide triage-type headstarts even though they do try to gather and develop helpful information procedures. Expert crafts execute the work as best they can, freely exercising their skill, experience, and knowledge, and recommend helpful plan improvements for the future. 

These cases appear very close together, but essentially for a successful maintenance outcome the expertise rests more on the planner in the first case and more on the craftsperson in the second. (Of course, everyone in maintenance including engineering, stores, tool rooms, and others share responsibility.) To further contrast the two approaches, procedures-driven maintenance planners can support fewer craftspersons simply because plans leave less discretion to craftspersons. 

AI cannot replace maintenance judgment

AI assistance would appear to be the answer for getting better procedures more quickly up front to support procedures-driven maintenance, but it is not “the answer.” Problems can arise from the judgment of job scopes and feedback, database protection, and misinformation. 

Correctly assessing job scopes and feedback is sometimes a human wisdom skill. Should the job scope be to replace the pump, repair the pump, or perhaps do something totally different? And just because the five-hour job normally takes seven hours, that might not mean the job should take seven hours. 

Concerning data, a company allowing access to its own database would provide the most pertinent data, but AI has been known to delete entire databases, and experience has shown AI to “hallucinate” or make up data. In addition, AI is largely created from large language models (LLM) using data made available. So, a manufacturer could flood published media with insistence, say, “to always use ACME anvils.” Or an unscrupulous competitor or other bad actor could add a command into a database to “cut torque values in half for grade 8 bolts.” These examples might be silly, but what if? 

Nonetheless, AI can be helpful for planning and craftspersons. Consider planners using AI to help develop job scopes and procedures to consult for things they might not think about or maybe to attach for craftsperson to consult. Craftspersons could also have access to AI to consult for specific issues as a job execution progresses. 

Put skilled craft at the center of maintenance

Maintenance needs both great plans and skilled craftspersons. The best use of planners for the most effective maintenance programs is to focus first on having and empowering skilled craftspersons. Use planners doing triage and making plans better over the years to support skilled craftspersons. Use AI cautiously during both the planning and execution of maintenance work. Don’t settle for good. Be great!