Before Integrated AI existed, before the Industrial AI Index was built, I spent 16 years watching organisations try to adopt new technology.
At SAS across Asia Pacific, working with organisations using data and analytics to drive commercial outcomes. At WesTrac, one of the world's largest Caterpillar dealers, leading marketing, customer experience, digital and intelligence -- and later digital transformation -- through the period when industrial AI went from concept to commercial reality.
The same patterns appeared every time. Different industries, different scales, different technologies. The same patterns. I'm watching them appear again now with AI.
The foundation comes first
At SAS I worked alongside organisations that were serious about using data to drive decisions. The ones that succeeded had one thing in common: they built their data foundation before they bought the tools. The tools worked because the foundation could hold them.
The ones that failed did it the other way around. They bought impressive software, brought in vendors with compelling demonstrations, and then discovered they had nothing to run the tools on. Months of integration work. Budget overruns. Business cases that never delivered.
At WesTrac, I led the development of FitFleet, a customer portal that gave operators a centralised view of their fleet health and proactively identified maintenance needs before they became failures. It only worked because the data foundation existed underneath it. The same principle applies to every AI tool your vendors are currently demonstrating to you.
Technology adoption is a people problem
WesTrac's website optimisation programme won a global Sitecore Experience Award. But what actually made it work wasn't the technology. It was the internal capability built to use the tools, to interpret the data, to make decisions based on what the platform was showing.
I've been in the room when implementations succeeded and in the room when they failed. The technology was rarely the variable. The people and the capability built around the technology -- that was almost always the variable. AI is not different.
The vendor cycle
At every stage of my career, vendors arrived with compelling demonstrations of the next capability before most organisations had built the foundation to use it. The board asks why you're not doing what the competitor announced. The vendor offers a pilot that seems low-risk. And then you've got a scattered set of AI experiments with no connective tissue, no governance, no clear line to business outcomes.
That gap has a name: accidental AI adoption. It's the most expensive problem in industrial AI right now, and it's entirely avoidable.
Where to start
The starting point is an honest view of where you actually stand. Not a vendor's view. Not a competitor benchmark that doesn't account for your operational context. An honest, specific, operationally grounded view of your current AI readiness.
Sixteen years taught me that organisations that know their starting point make better decisions about where to go next. The ones that skip that step spend money finding out the hard way.