Two weeks, on your own data, before you commit to anything.
Run in your own environment, reviewed with your process owner as the results appear. Free and non-binding.
One line, one target variable, one export. We take it from there — ingestion, cleaning and shaping are ours, not yours.
Reviewed with your process owner as results come out, not presented at the end when it is too late to redirect.
A validated case, or a documented no that you keep. Both are real outcomes and both cost you the same: one export.
One line. One target variable. One export.
That is the whole ask. If you can name the line and say what you want improved, we can start.
- The line
- One production line, one plant. Plant-level scope, not an enterprise programme.
- The target
- The one number you want to move — downtime, yield, scrap, consumption.
- The export
- Whatever you already produce. Any source or format — a CSV is a starting point, not a blocker.
- One decision
- Which of your variables are settings and which are results. That is a process question, and only your team can answer it.
A paid pilot, on the same scope with more depth.
Between €5k and €20k, over four to eight weeks. Where we can, we condition the pilot on conversion once the targets are reached — so the pilot is a step toward a decision rather than a destination.
Why we do it this way
Pilots kill promising industrial software. They also cannot be avoided in this market — nobody lets an unproven model near a running line. So we designed around them instead of pretending otherwise: two weeks free to establish whether there is anything there, then a paid pilot only if there is.
Who you are talking to
The proof of value puts us in direct contact with the production team rather than with a procurement process. They are the people who will use the product, and they are the ones who can tell in two weeks whether the answer is any good.
Sometimes the log carries nothing.
On one machine in the aerospace case, major breakdowns were preceded by no build-up of micro-stops at all. There was nothing to see, it was not a tuning problem, and we said so in the deliverable rather than tuning until something appeared. You keep that finding — it tells you where not to spend next.
Name one line and one target variable.
We review your data extract within a week — before you commit to anything.
