Notes · AI and data on the shop floor
Testing a factory idea in days, not months
In business school the Minimum Viable Product is a clean idea: build the smallest thing that can teach you whether the idea is worth pursuing, then learn and improve.
Where I learned it
I met it in my Executive MBA strategic project. We started from a big idea, solar containers where remote communities could charge and swap batteries. Customer research narrowed it to a back-up battery box for small shops in Nairobi, where the grid fails often. A teammate in Kenya built a first unit and installed it with real shopkeepers. It took months of coordination, and it taught us more than the spreadsheet did. We never launched it, but I kept the lesson.
What has changed
In a factory, testing a concept used to cost the same kind of effort: an engineer, a supplier, a budget and weeks. That has changed, and I feel it every month.
Today I can take an idea I have already defined and refined, and use an AI assistant to build a first version quickly: a first design, some code, a calculation, a draft procedure. The assistant does the groundwork. I still decide what is worth building and whether the result is good enough.
An example from the shop floor
We did this with sensors that tell us how intensely our sewing machines are used. The team designed the sensor, bought very cheap electronic components, and built and configured them in-house with AI support. A pilot on a few machines worked well and quickly showed differences worth a conversation with the production team. We are now refining the concept, with something real to learn from rather than a slide deck.
What this changes for a small organisation
- Speed. Ideas are tested in days, so more of them get tested.
- Autonomy. We can develop and implement projects without waiting for an outside specialist for every step.
- Better questions. When a test is cheap, you can afford to be wrong early.
What it does not change
Someone has to define the problem well, judge the result and take responsibility. The assistant makes the work faster. It does not make the thinking optional, and it does not reduce the need for the people who know the factory.
A question to keep
What is a concept in your business that you would test this month if it cost almost nothing?
Questions this note answers
- What is a minimum viable product in manufacturing?
- The smallest working version of an idea that can teach you whether it is worth pursuing, tested early with real users or real machines, then improved. A cheap first version teaches more than a perfect system that arrives late.
- How can AI help a factory test ideas faster?
- Once an idea is defined and refined, an AI assistant can do the groundwork for a first version: a first design, some code, a calculation or a draft procedure. A person still decides what is worth building and whether the result is good enough.
Faced the same question, or see it differently? I like comparing notes. Get in touch →