Abstract
Of the four readiness gates, materials is where the largest number of jobs actually fail, and it is the gate least represented in industrial AI research. This paper argues that the materials question is not whether the part is in stock but a chain of six conditions, any one of which turns a ready job into a cancelled one. The sixth condition is not in any inventory system I have seen, and it is the one that destroyed a hundred-thousand-dollar motor without anybody touching it.
Six conditions, not one
Identified. Does the work package name a specific part, or a description? Many name a description, and the substitution then happens at the counter, from memory.
Correct. Is the identified part the one the asset takes now? Assets get modified and the bill of materials does not, so the catalogue describes a machine that no longer exists.
Present. Does the physical count match the system count?
Reachable. Can it be found and issued inside the job window, by someone with access, on the shift the work is scheduled for?
Complete. Is the whole kit there? A job needing eleven items and holding ten is not ninety-one percent ready.
Viable. Has the part survived storage in a condition that still works?
The first five are in every textbook. The sixth is the one nobody counts, and it is the reason for this paper.
The spare that died standing still
The stranders in my plant ran motors so big the mill had to phone BC Hydro every time they started them. Easily a hundred thousand dollars apiece.
A spare came in, and somebody decided a good idea would be to park it right underneath the stranders. The highest vibration point in the entire mill.
I agreed to it at the time, and I want to be exact about why, because it matters. There was a mechanism built to hold the shaft up so the rotor did not ride on the bearings. It worked. It actually held the shaft up. With the rotor's weight off the bearings, the vibration had nothing to chew on. That was the correct mitigation and it was in place when I signed.
Then somebody removed the mechanism. For whatever reason. No record, no rationale anybody could point to.
The motor sat there, full rotor weight on stationary bearings, vibrating, for at least three years.
Then another mill needed it. They shipped it, installed it, and it lasted about three months before it failed.
That is false brinelling. Three years of the stranders hammering the rollers against one spot on the races, lubricant squeezed out of contact points that never moved, wear divots pounded in at exact roller spacing. The bearing was destroyed years before it ever turned, and the failure surfaced at a different mill where nobody could see the cause.
I never learned whether the receiving mill worked out what actually happened. My guess is they blamed manufacturing, because that is what usually gets blamed.
I had warned our own outfit many times that a machine sitting there vibrating was not a good thing. The answer I got was "you signed off."
Yes. With the mechanism holding the shaft up. That sign-off died the day the mechanism was removed. I pushed to have it reinstated afterwards, and they would not go for that either.
What that story does to inventory
Run the motor through the five standard conditions. It was identified, it was correct, it was present, it was reachable, and it was a complete assembly.
Five green gates. And it was dead before anybody turned it.
"In stock" is a statement about location, not about condition. Every inventory system I have worked with treats a part as a quantity in a place, and quantity in a place says nothing about what that place has been doing to it for three years.
There are two separate failures stacked here and they are worth pulling apart.
The first is that storage conditions are a maintenance variable and nobody maintains them. Vibration, humidity, temperature, and shaft loading on stored rotating equipment all degrade a spare in ways that leave the count unchanged.
The second is worse and it is a records failure. A mitigation existed, was correct, and was removed with no record of who removed it or why. The sign-off survived. The condition it depended on did not. And nothing in the system connected the two, so three years later the paper trail said the spare was approved and stored, which was true and completely misleading.
An approval is only valid while its assumptions hold. If the assumption is not recorded alongside the approval, then the approval outlives the reason for it, and that is exactly how you end up defending a decision you did not make.
The catalogue describes a mill that no longer exists
An OSB mill got restarted after roughly twenty years shut down. Through the whole shutdown the company had kept one millwright on site. Not to do millwright work, but to keep the place from being overrun by wildlife and to do property and building maintenance. Twenty years, alone, with a mill nobody was running.
I came on for the restart. Every locker in the shop was full. Not full of junk exactly. Full of sorted, curated, decades-old stock, some of it left by millwrights who had worked there before the shutdown and filed away by the caretaker in the years since. An entire locker of halogen work lights. Another of strapping tools. Another for cutting torches. Another for fall arrest. One holding nothing but brass, kept to resell.
I needed the lockers for the new hires coming in, so I pulled a dumpster up to the shop door and started pitching. Outdated fall arrest first, then the halogen lamps that did not work, then the strapping tools.
The caretaker came running and started yelling about what it was all worth. He pulled a strapping tool back out of the bin and told me it was worth seven thousand dollars.
I told him I knew exactly what it was worth, and that was why it was going in the bin. It used to be seven thousand dollars, twenty years ago. We do not even run metal strap any more. Halogen lamps are not made any more. Fall arrest expires, and nobody is re-certifying twenty and thirty year old gear.
After I went home he pulled every bit of it back out of the dumpster and put it in the lockers. The next day I put it all back in the dumpster and took it outside to the recycling bins.
That is the Correct condition failing at scale. Not one wrong part number. An entire inventory that was accurate about a plant that had stopped existing. The count was right. The location was right. The valuation in anybody's head was right for 2005.
And here is the other half, which every parts paper leaves out. On my first day I was hunting for something, and the caretaker walked me to a sea can, unlocked it with the only key anybody had, and handed me the part.
Every other manager on site wanted him gone. He was the only person in the building who knew where anything was. The Reachable condition was, for twenty years, entirely dependent on one man and one key, and no system anywhere recorded that dependency.
Verification, and who should do it
Here is how a part gets ordered in most plants. The tradesman says he needs it, the planner looks it up in the books or the computer, orders it, and builds the schedule around the supposed delivery date. Nobody in that chain verifies anything. Not the part, not a measurement. So it shows up wrong, with the line down, and a crew standing there holding a part that does not fit.
We had a conveyor replacement arrive the wrong length that way. A sixteen-hour scheduled window became an unplanned week, because I had to retrofit both the infeed and the discharge to make it fit.
After that happened a few times we made a rule: you cannot trust the books, anything like this gets its size verified before the order goes in. It did not matter who did it. Usually a millwright put a tape on it. Once in a blue moon a planner came down himself.
The fairness of that matters, and it is the same argument I make about tradesmen and workarounds. It is technically the planner's job to verify. But most of the time the planner does not know where the machine is or how it works, and asking him to verify a measurement he cannot physically reach is the system failing him rather than him failing the system.
So design the verification step in, and route it to whoever can actually perform it. It does not matter who does it, as long as there is verification.
The unofficial inventory as a diagnostic
Every plant has a second inventory, held in toolboxes and lockers by tradesmen who learned the official one cannot be relied on.
Read it as a diagnostic rather than a discipline problem. The contents are a precise, empirical list of the items the formal chain fails to deliver in time. That is free, high-quality data about which of the six conditions is failing and on what parts, and most plants respond to it with a memo about accountability.
Where automation helps, and where it must not
Checking the first five conditions against a work package is retrieval and comparison. Current systems do that well and safely. Oracle and IBM both expose agents that read and construct maintenance records, and a kit-completeness check across a backlog is a legitimate early application with a clear stop condition.
Two things should not be automated.
The substitution decision. Choosing an equivalent part requires knowing what the machine is doing now, and that judgement belongs to a named person whose name stays attached.
The viability judgement. No system can infer from a stock record whether a spare has been sitting under a vibration source for three years. What a system can do is flag that a stored rotating asset has no recorded storage condition and no inspection since receipt, and turn that into a question. That is a readiness finding and it costs nothing to produce.
Conclusion
Materials is the gate where readiness is most often lost and most cheaply recovered, because unlike evidence or authority it fails on conditions that are each individually checkable.
A plant that converts its parts metric from an average to a kit-complete count will find its true readiness lower than reported and its path to improving it much shorter than expected.
But the harder lesson is the sixth condition. A part can be identified, correct, present, reachable and complete, and still be destroyed. The motor was on the books the entire time. The count never changed. It cost a hundred thousand dollars and three months at somebody else's mill, and the record of the one thing that would have explained it, the mechanism that held the shaft up, was never written down at all.
References
- Oracle, “AI Agent: Maintenance Work Order Builder.” https://docs.oracle.com/en/cloud/saas/readiness/scm/25d/maint25d/25D-maintenance-wn-f40777.htm
- IBM Docs, “Creating custom AI tools in Maximo by using object structures.” https://www.ibm.com/docs/en/masv-and-l/cd?topic=tools-by-using-object-structures
- Misra, Vyas, Gutta & Mercangöz, “ADMITBench: A Safety-Governed Reference Framework for Evaluating the Admissibility of Industrial LLM Advisories,” 2026. https://arxiv.org/abs/2608.03866
- Liu et al., “ConMem: Contribution-Aware Memory for Long-Horizon Manufacturing Inspection Logs,” 2026. https://arxiv.org/abs/2607.28126