Before you fund the next AI pilot, answer five questions.
They come from Chapter 10 of Why Industrial AI Fails. Answer for your own team, honestly. If the room goes quiet, the book is for you.
Answer each one yes, partly, or no. Then read what your answers mean, below.
01
Can you define what “good” or “bad” means for your most important processes and metrics?
An actual limit, an actual KPI, not three people giving three different answers. Ch. 3
02
Do you have a defensible model of your operation?
A hierarchy, an asset model, a common language people would agree on. Ch. 5
03
Do you have access to real-time data?
Or is everything in a static report by the time you see it? Ch. 4
04
Can your experts’ knowledge be tied to specific assets?
When an expert explains something to others, does that land where the next person will find it? Ch. 6
05
Do users act when a system says something surprising?
Or is it explained away? Nothing you build matters if nobody believes it. Ch. 8
How to read your answers
if you answered
Any “partly”
The middle answer is the dangerous one. It means somebody already believes you have it, so nobody is working on it.
if you answered
Some “no”, no “partly”
At least you know where the gaps are. That is where the book starts.
if you answered
Five yeses
Good. Now ask the people who run the night shift the same five questions.
