Leading the Intelligent Factory:  Chapter 1

The pressure you are under

Fewer people, more data, more AI noise. The answer is better decisions, not more tools.

Spread 1

More data, fewer people, and a new word

You run a plant that was built to last. Some of your equipment is older than your newest engineers. Your control systems come from three vendors and four decades. Your historian holds years of data that almost nobody has time to look at.

At the same time, the people who knew how it all worked are leaving. The engineer who could hear a refiner going wrong retired last spring. The operator who knew why the target changes on the heavy grades moved to days. You are covering more with fewer people every year.

Now there is a new word in every meeting. Your corporate office wants an AI plan. Vendors call every week. Your board has read that AI will change manufacturing. Nobody has shown you, in plain terms, where it helps in a plant like yours.

2015202020252030Process data collected per dayExperienced engineering hours per machineThe gap AI should fillIllustrative, not data
Figure 1.1. The gap AI should fill. Illustrative, not data.

Spread 2

Why most AI pilots stall

Many plants have already tried. A data science team builds a model. It works in the demonstration. Six months later, nobody uses it.

The pilot rarely fails because the model was wrong. It fails for one of three reasons:

  1. It answered a question nobody on shift was asking. A model that predicts something the crew cannot act on becomes a screen nobody opens.
  2. It arrived at the wrong stage. Optimization sold into a process that breaks unpredictably cannot hold. There is nothing stable to optimize.
  3. Nobody owned the result. A finding without a name on it is noise. The model kept flagging; the plant kept firefighting.

None of these is a technology problem. Each is a leadership problem, and that is good news, because leadership problems are the ones you can fix.

This book takes a different starting point. It does not start with algorithms. It starts with the six decisions your plant already makes every shift: what to run, how fast, where everything else is set, when to intervene, what to fix first, and what to learn. AI earns its place by making those decisions better, in an order that fits where your plant is today. The next chapter shows you how to find out where that is.

Month 0Month 3Month 6Month 9Month 12Daily useA project: the demo spikeNear zero by month sixA habit: tied to a daily decision
Figure 1.2. A project and a habit. Illustrative.