MoldVision — an instrument
How worn is it, really?
Photograph a mold surface with a phone. MoldVision gives you a recorded, comparable observation of its condition — so the answer next quarter can be measured against the answer today, instead of against what somebody remembers.
The problem it goes after
Tool condition is an opinion until someone writes it down
Ask three people whether a cavity needs attention and you can get three answers, all defensible, none written down anywhere you can compare next time. The decision that follows — pull it now, watch it, or run it — is usually made on experience and a torch.
That works, right up until the person with the experience is on holiday, or the tool goes out to a shop that has never seen it before.
What you get
A record, not a verdict
The image, kept
Every observation stores the photograph it came from. Anyone reviewing the call later can see exactly what was looked at.
Comparable over time
The same surface, photographed again after another few thousand cycles, lines up against the last one. Change is the thing worth seeing, and change needs two points.
Hardware you have
It is built around a phone camera. No fixture-mounted rig or specialist inspection hardware to buy before you can find out whether it helps.
The part most vendors skip
What has actually been tested
MoldVision has been validated against synthetic imagery only. It has never been pointed at a real tool, and there is no pilot site yet.
The measurement core works, and we can show you how we know. But everything we know, we know from images we generated ourselves with the right answer known in advance.
Why there are no accuracy figures on this page
We have numbers from those synthetic tests. They are good. Publishing them here would tell you the method is sound — and imply something we have not earned, which is that it will behave the same on your steel, under your lighting, on your parts.
Real numbers come out of the first pilot. If you run it, you will see them before we do, and you will see the ones that did not go our way as well.
What a first pilot looks like
- Your tool, your choice. Pick one you already have an opinion about — that is the useful test.
- Nothing gated on it. No production decision waits on the output while we are finding out whether it works.
- Your toolmaker's call still stands. If the software and the person disagree, the person is right until proven otherwise, and the disagreement is the most useful data we can get.
- You see everything. Including where it was wrong. Especially where it was wrong.
Being first is a real thing to ask of you. You would be spending your time to find out whether this works. We price and scope a first pilot accordingly — talk to us about what that looks like rather than assuming.
Willing to be the first tool it sees?
Tell us what you run and which tool gives you the most trouble.