Ask a production manager what they did the day before yesterday. The answer is rarely what was in the calendar.
The day was supposed to start with the planning review and then move on to capacity planning for the autumn. Instead it went to questions. Where is the material for order 45120? Which drawing revision applies? How did we set the machine last time we ran this item? Can I use the other sheet if the first one has run out? Has the customer approved the change?
Each one takes two minutes. Together they take half the day. And the questions are not the problem. The problem is that they arrive in a sequence that makes qualified work impossible to settle into.
The questions are predictable
What looks like noise is in practice four kinds of question, and they recur every day in every manufacturing company.
| Type of question | Examples | Where the answer sits |
|---|---|---|
| Where is it? | Material, tooling, fixture, batch, pallet | The ERP, the stock location |
| How is it done? | Setup, tool change, packing instruction, measuring method | Work instruction, manual, a colleague’s head |
| What applies? | Drawing revision, tolerances, quality requirements, deviation handling | Drawing archive, quality routines |
| What is the status? | Order, inbound delivery, next operation, customer delivery date | The ERP |
Almost none of it calls for a production manager’s judgement. It only calls for knowing where to look.
The answers already exist, just not where they are needed
In a company running Monitor, Garp, Pyramid or any other ERP, the answer to the status questions is already in the system. The instructions sit in the quality management system or in a binder by the station. The machine manual is on a server. The experience from the last run sits with the person who ran it.
The operator still goes to the manager. That is rational: leaving the machine, digging through the binder and then searching the ERP takes ten minutes, while the manager answers in twenty seconds. As long as the manager is cheaper to ask than the system is to search, the manager will be asked.
That is the equation you can turn around.
A colleague on the screen
A digital coworker can sit on a touch screen by the line, in production engineering or at the packing table. The operator types or speaks, and gets an answer in the language they used. Behind the screen there are two things.
The first is the company’s own material: work instructions, setup routines, machine manuals, quality requirements, safety rules, past deviations and the routines that otherwise exist only in the habits of a few people.
The second is a live connection to the ERP. Orders, items, structures, operations, stock locations, balances and delivery dates are fetched when the question is asked, not from a copy that is a week old.
That means a question like “what am I running after this one?” can be answered from the actual schedule, and “how do I set it up for item 4471?” can be answered with the setup instruction plus a note on what went wrong last time.
Just as important is what happens when the answer does not exist. It does not guess. It says the information is missing, logs the question and passes it to the right person.
What it does not do
A digital production manager does not take over the management. Prioritising between customers, deciding to deviate from a quality requirement, staffing, working environment and everything involving people stays with a human. It should.
What it takes over is the interruptions. The difference between a production manager answering forty questions a day and one answering five is not that the second cares less. It is that thirty-five of those questions had an answer that was already written down somewhere.
What changes
The production manager gets uninterrupted time. Capacity planning, supplier dialogue, improvement work and developing the team all take more than seven minutes at a stretch. They are also the first things to be postponed.
Evenings and nights get the same answers as days. On the shift where no manager is present, today’s options are to guess, wait until morning or call someone at home. None of the three is good.
New hires get up to speed faster. Questions you would rather not ask a third time are easy to ask a screen. That includes anyone who does not have the local language as their first language, which in many production environments is a substantial share of the workforce.
You get a map of where the instructions fall short. Every question is logged. After a month it is plain which operations generate the most questions, and which instructions are unclear or out of date. Nobody has had that picture before.
Start with one station
The lesson from robotisation applies here too. No workshop became robotised by deciding to become robotised. They put one robot on one station, measured the outcome and took the next one once the first paid for itself. More on that pattern in Swedish industry has automated for 150 years.
The same applies here. Pick one work area, put up a screen, and start from the ten questions the production manager gets most often. Measure two things: how many questions the screen answered and how many went on to a human. The ratio between them tells you whether the next station is worth taking.
What a digital coworker is and how it connects to your ERP is described on the pages about the platform and integrations. How other manufacturers have gone about it is under customer stories.
Wondering where it would pay off most for you? Start by counting the questions on an ordinary Tuesday. Get in touch once you know the number.