I spent more than ten years inside real manufacturing environments — automotive and semiconductor — leading industrialization and process improvement work, sitting with quality, maintenance and operations teams on the days things went wrong.
Every failure followed the same script. A quality escape, a line stop, a customer complaint. Somebody opened Excel. Somebody copied an old Word template. Somebody spent two days chasing people for information that already existed somewhere in the building. And then we rebuilt, from scratch, an analysis that another engineer — in another plant, in another year — had already completed and closed.
The question I kept asking was not “why did this part fail”. It was: why are we still doing this manually?
What I actually saw
It took me a long time to name the thing properly, because it hides behind a word that sounds like a compliment.
We say a plant has “experienced engineers”. What we usually mean is that the plant’s knowledge lives in a handful of people’s heads, and everyone else is working with a fraction of what the organization actually knows. The distance between a first-year engineer and a twenty-year engineer is very rarely a distance in intelligence, rigour or care. It is a distance in access — to the failure that looked exactly like this one four years ago, to the fix that was tried and did not hold, to the reason a parameter has that tolerance in the first place.
That is not a talent problem. It is an information problem wearing a talent problem’s clothes, and it costs at both ends: the engineer who joined last month is capable and slow, because they are starting from a blank page the company could have filled in for them — and the engineer with twenty years is the single point of failure for everything the plant has learned, until the day they retire and it leaves with them.
Neither of those people is doing anything wrong. The system around them is.
Our mission
Every engineer performs at their maximum from the first day, regardless of how many years they have.
I want to be exact about what that does and does not mean, because the overclaim is easy and it is the fastest way to lose an engineer.
It does not mean a tool turns a junior engineer into a senior one. Judgment is earned on the floor and there is no shortcut. It means the part of the gap that is pure access — what has already been tried here, what held, what came back, what the statistics actually say — should not exist. That part is not experience. That part is just information nobody organized.
So the newest engineer in the building opens an incident and immediately sees what this plant already resolved that looks like it. The most experienced engineer stops re-deriving conclusions they reached years ago and spends their time on the problem that is genuinely new. Both are working at the top of their capability instead of at the top of their recall.
What it refuses to do
DatumKeep guides the investigation — 5-Why, Ishikawa, 8D — with your own resolved cases surfaced in context, statistics computed by real engines beside the reasoning, and corrective actions tracked to closure. What I care about more is the list of things it will not do, because that list is the product.
Those constraints cost us features. They are the reason the thing is trustworthy on a shop floor, which is the only place its opinion matters.
Where we honestly are
We are early. The first plants are onboarding now. We are not SOC 2 or ISO 27001 certified — that is ahead of us, not behind us. Some capabilities ship switched off until an admin turns them on, and we say so on every page rather than letting “shipped” and “on by default” blur together. I would rather tell you now than have you find out later; it is the same principle the product runs on.
Mid-sized manufacturers — roughly 50 to 500 people — are underserved and overpriced by the existing tools. They have the same failures as the large groups, the same regulatory pressure, the same customer complaints, and none of the corporate quality apparatus. They are the plants where one engineer’s week decides the month.
If you are trying to make your engineers better at their jobs — not by asking them to work harder, but by giving them everything your organization already knows — I would like to talk to you. Bring a failure you already understand. That is the fastest way to judge whether any of this is real.
José María González Gómez
Founder, DatumKeep