Selected work

Five cases. Skim first, read the ones that fit.

Each one is the same shape: what everyone assumed, what was actually wrong, and the calls I made. The numbers below are the scan. The link is the full story.

Hatch · Industrial operations

The AI worked. Nobody trusted it.

What everyone assumed
Operators need better predictions.
What was actually wrong
Operators were not doubting the numbers.
30%Less energy and chemical use
Read the case study
Hatch · Mining operations

Turning a node amber before anything is wrong with it

What everyone assumed
Teams cannot see enough.
What was actually wrong
They had plenty of screens.
30 minWarning before the loss reaches the crusher
Read the case study
Veeva · Life sciences

They were taping printouts to the wall.

What everyone assumed
Users need better search and filtering.
What was actually wrong
The data was there but the relationships were not.
45→1Printed pages replaced by one screen
Read the case study
Veeva · Life sciences

Eight hundred organisations. The right one was never at the top.

What everyone assumed
The problem was visibility.
What was actually wrong
Visibility alone does not help at this scale.
818 → 91Organisations, after one filter
Read the case study
Concierge Plus · Property software

The software still worked like the binder.

What everyone assumed
The flow is too long.
What was actually wrong
Length was a symptom.
80%Faster booking, about 2:00 down to 0:24
Read the case study

Have a problem like one of these?

Send a link and a sentence. If I can help, I will record a teardown. If I cannot, I will say so.

nate@nate.design One engagement slot open this quarter