At seven on a Monday morning the tool-room attendant at an engineering site has three jobs on the issue list and three torque wrenches on the shelf. The site is invented, and every figure in this article is chosen to show a rule rather than measured anywhere. The asset tracking record, which in this tool room is a sign-out book, says the wrenches are in. The stickers on two of them say the calibration is current. The third sticker says the calibration ran out on Saturday, and nobody has moved that wrench to the quarantine shelf because nobody has looked at it since Friday. One of the two good wrenches expires tomorrow evening, and one of the three jobs runs until Wednesday.

Counted, there are three wrenches for three jobs. Assigned properly, there is one wrench for the three-day job, one for a one-day job, and a wait for the third. The count was accurate and the answer it gave was wrong, because a count says where things are and the attendant needs to know something else.

Three questions decide whether a job can start with the tool it needs. Where is the tool? Who holds it, and for which job? Is it permitted to do this job for the whole time the job will take? Most asset tracking answers the first question well and the other two badly or not at all. This article is about why the second and third matter more, and how to tell which of the three your own system can answer.

Asset tracking and inventory in a tool room

The searches that lead people to this subject say “tool room inventory”, and the phrase hides a distinction worth making before anything else. Inventory is stock that is counted and used up: abrasives, fasteners, gloves, cutting fluid. A tool room holds some of that, and it follows the counted-stock logic set out in inventory management. Most of what a tool room manages is assets: items with their own identity, issued to someone, returned, checked and maintained over years. Wrench W3 is one particular wrench with its own history and its own certificate, and the question at the counter is always about a particular item.

Three properties belong to each asset. Location is where it physically is: the crib shelf, bay four, the calibration lab. Custody is who has it and for what work. The two differ more often than people expect. A wrench can be on the right bench in the right bay and still belong to the night shift’s job until that job closes at half past seven. Presence says nothing about whether you may take it.

Fitness is whether the asset is allowed to do work at all, and it rests on checks. A calibrated instrument is measured against a standard, a certificate records the result, and the check is repeated on an interval. Lifting gear and powered trucks are inspected periodically. Some items carry condition or use limits: a drop, a number of cycles, a number of running hours. Each check is a rule with a date or a count attached and a piece of evidence behind it. The intervals differ by equipment, by client and by jurisdiction, and this article quotes none of them. What matters for the argument is the shape of the rule: every check has a due date, and past that date the tool is unfit whether or not anyone has moved it.

The tool room, or crib, is the point where issue and return are recorded. Everything else here depends on what gets written down at that counter and which relationship the entry captures.

Why knowing where a tool is runs out

A tool cycles through states. It sits in the crib, fit. It is issued. It comes back and waits for a return check. It goes to quarantine or out to the lab for service, and eventually back to the crib. Each transition is recorded by a different person at a different moment: the attendant at issue, the technician on return, the lab on the certificate. With a sign-out book and a sticker, each of those people sees only their own entry. The attendant does not see the lab’s result until the wrench comes back. The technician borrowing a wrench from the next bay does not see that it was due back on the previous shift.

Fitness runs on its own clock. A calibration expires on a date whether the tool has moved or not, so a location record can be accurate and still wrong about availability. A wrench sitting on the shelf can be unfit. A fit wrench can be on the shelf and committed to someone else’s job. “Is it in the crib?” and “can it do this job?” are two questions, and a location count, a barcode scan or an RFID reader answers the first one only.

A job adds a third condition. It needs a class of tool with the right range, and it needs the calibration to be valid on every day it runs. Assignment therefore has to match a specific asset against the job’s window, not against the moment of issue. A wrench in date today and expiring tomorrow can do a one-day job. It cannot do a three-day one, even though the sticker looks fine at the counter on Monday.

The same three questions read across to a warehouse with the equipment changed. Swap the torque wrench for a powered truck that is due its periodic inspection, or a hand scanner shared between the day and night shifts. The truck can be parked in its bay, charged and ready to drive, and unfit because its inspection date passed at the weekend. The scanner can be sitting in its cradle and still signed out to the night shift’s pick wave until that wave closes. A three-day stock count starting Monday needs a truck whose inspection is valid through Wednesday, and a truck due on Tuesday can do Monday’s replenishment and cannot do the count. The warehouse operations manager asks about a truck exactly what the attendant asks about a wrench: where it is, whose job holds it, and whether it is fit for the whole of the work in front of it.

When the record cannot answer those questions, people fill the gap. The attendant reads stickers and does the date arithmetic in their head, the planner phones round for a spare wrench, and the technician who cannot wait borrows one from a colleague. Two outcomes follow. A job starts late while somebody hunts for an in-date tool, or a job starts on a tool that should never have left the crib. Neither shows up in a location report, because in both cases every tool is exactly where the report says it is.

The rule at the counter

Back to the six wrenches. The site holds six torque wrenches of the same range, W1 to W6. At 07:00 on Monday the attendant prepares issues for three jobs starting that morning. Job A runs one day. Job B runs one day. Job C runs three days, Monday to Wednesday. The state of the pool at 07:00:

The rule is one sentence: a wrench may be issued only if its calibration is valid on every day of the job it is issued to.

A count of what is in the crib says three wrenches for three jobs. The rule applied to each job’s window says something different. W5 is the only wrench on the shelf valid through Wednesday, so W5 goes to Job C. W6 is valid through Tuesday and Job A finishes today, so W6 covers Job A. W4 is expired and goes to quarantine. Job B has no wrench on the shelf that qualifies. It waits for W1 or W2 to come back from the night shift at 07:30 and pass a return check, and starts around half an hour late.

Half an hour late is the good outcome. Had the attendant issued on location alone, three wrenches would have gone out to three jobs and W4 would have been one of them. The mistake would have surfaced at the next audit or the next failure, if at all. The other error was available too: an attendant working from stickers could have sent W6 to Job C because its sticker read as current on Monday, and Job C would have run its Wednesday on an expired tool.

Two things happened at the counter that a count cannot do. Fitness was evaluated over each job’s window, and custody was checked: W1 and W2 were on the floor and fit, and still unavailable, because they belonged to another job until 07:30. Availability is fitness over the window plus custody. The question the attendant needs the record to answer is “which of these can do this job for as long as it runs”, and a record that cannot answer it leaves the attendant working from stickers and memory.

What the same record does when a tool fails

On Thursday W3 comes back from the lab. The certificate says out of tolerance at its lowest setting. Its last pass certificate is eleven months old, so every job it has touched since then is suspect, at least where the low setting was used.

The issue log shows fourteen issues of W3 in those eleven months. What happens next depends entirely on what each of those fourteen entries recorded. If each issue was recorded against a job, the quality lead has a list of fourteen jobs within minutes, can narrow it to the ones that specified a low torque, and can begin the review the same morning. If each issue was recorded against a technician’s name, which is how most sign-out books work, the same fourteen entries have to be matched back to jobs from job cards, shift rotas and memory before the review can start. The wrench was never lost. Its location was known every day. The record still failed, because it captured the wrong relationship.

This is why custody to the job matters on a site that has never mislaid a tool. The entry that made Monday’s assignment safe, this wrench issued to this job for these dates, is the same entry that turns Thursday’s failure into a finite list. Recording issue against the job costs the attendant one more field at the counter and saves the quality lead the reconstruction.

The tool room running on Keel

Keel is an operations platform shaped around how your workshop or warehouse runs its jobs, its tools and its checks. For a tool room, that means the issue-and-return counter, the return checks, the calibration and inspection returns and the jobs themselves run as one set of guided tasks, with your rules built in: which tool may go out to which job, what must be checked when it comes back, and when a tool leaves the pool. Suppose the site with six wrenches ran on it. What follows describes what an implementation for that site would run; there is no ready-made tool-room module.

The attendant’s issue task shows the job, its dates and the tool class it needs, and lists only the assets whose fitness covers the whole job window. On Monday morning the list for Job C contains W5 and nothing else. Scanning W6 against Job C is refused, with the reason shown: valid until Tuesday, job runs to Wednesday. W4 does not appear on any list, because it left the available pool on Saturday when its calibration lapsed, and a service task for booking the lab appeared in the tool-room queue at the same moment.

When a wrench comes back, the return step asks the condition questions that matter for that class: dropped, damaged, reading suspect. The answers decide whether it goes back into the pool or into quarantine, and the attendant does not have to remember which questions apply to which kind of item. W1 arriving at 07:30 passes its return check and appears on Job B’s list a moment later.

For the quality lead on Thursday, the asset history for W3 lists every job it was issued to, every return check and every certificate. The fourteen-job list is that view filtered from the date of the last pass certificate. For the planner on Friday, a report on pool availability by class over the planning horizon shows that Monday has two fit wrenches for three jobs before Monday arrives, which is when the conversation about moving Job B or hiring a wrench should happen.

Underneath, each asset is a record with its own identity, a current state (available, issued, awaiting check, quarantined, in service), a custody link to a job, and its fitness attributes with the certificate as evidence. The rule “valid for the whole job window” is business logic written in code within the implementation and checked when the issue task is completed. State transitions are workflow states. Permissions decide who may override a refusal, so a quality lead can grant a recorded concession where the operation allows one. Keel supplies the platform: workflow state, tailored operator screens, permissions, rules in code, task queues and reporting. The implementation builds the asset model and its classes, the fitness rules for each class, the issue and return tasks, any import of certificates from the lab, and the link to the job or work-order records, which may already live in Keel or in an existing system. Keel leads the implementation and hands over to your team; what Keel supplies sets out where that boundary sits.

Code needs building, testing and maintaining, and that is the cost of a rule that says exactly what your site means by fit. Tool fitness is one of several conditions that decide whether a job’s next step is permitted; production control covers readiness as a whole, and this article has taken one of its conditions to full depth.

One place AI could sit inside this workflow is reading a returned calibration certificate and proposing the new due date for the attendant to confirm, a step an implementation would define along with the confirmation and whatever follows it.

When you don’t need any of this

If your problem is lost tools, a barcode or RFID location system, or a controlled cabinet with a check-out log, solves it. Those answer the first of the three questions, and if that is the question you have, they are the right answer.

If the pool is small, runs on one shift, and one attendant knows every tool, the tool or asset module of the maintenance system you already own probably records due dates and can record issue against a work order. Check both. If it does both, you have the essentials, and the gap between you and the Monday described above is process discipline at the counter rather than software.

If your existing system already refuses to issue an out-of-date tool against a job, do not replace it for this reason.

Keel fits when the fitness rules are your own: a client specification that sets its own interval, a job-window rule, a pool shared across sites with different intervals. It fits when issuing a tool has to become a step on the job the operation already runs rather than an entry in a separate system, and when you expect the rules to change every time a contract adds a specification. Whether yours is that kind of operation is something the three questions, put to the system you have, will tell you.

Three questions for your own tool room

Take the record you issue from today, whether it is a book, a spreadsheet, a cabinet log or a module, and ask it three things about one specific tool. Where is it? Who holds it, and for which job? Can it do a job that starts tomorrow and runs for three days, and how does the record know?

Then ask the backward question: if this tool came back from the lab out of tolerance this afternoon, could the record list every job it has been used on since its last pass, without anybody opening a filing cabinet?

A system that answers the first question only is a location system, and it may be all you need. A system that answers all three at the moment of issue, and the backward question afterwards, is joining the tool record to the job record, which is a different thing to buy and a different thing to run. If you recognise that your fitness rules are your own and the record at the counter is not applying them, talk through how issue and return would run as steps on your jobs. The conversation starts from your tool classes and the rules you already apply at the counter.