Leading the Intelligent Factory:  Chapter 10

Learn from yesterday

Every finding becomes a target, so the gain does not drift away.

Spread 1

A fix that is not written down is on loan

The pressure. You have solved most of your problems before. The same break, the same trip and the same drift come back two years later, and the person who solved it has moved on.

What goes wrong today. A fix lives in a report, an email, or one person’s memory. The target sheet does not change. Six months later the process is back where it was, and the next investigation starts from zero.

What AI changes. This is the decision of what to learn. Every finding ends in a changed standard: a new target, a new limit, or a new check. A score watches that standard every hour, so a move away from it shows in days, not months. And every past case stays searchable, so the same symptom starts from the last answer.

What still needs people. Agreeing that the new way is the standard. A target the crews do not accept will not be run to.

FindingFindingFindingWritten intothe standardFixed, notwritten downPerformanceTwo yearsIllustration, not data
Figure 10.1. Three findings in two years. Without a changed standard, each gain drifts back. With one, the gains stack.

A real example

The answer was already in the kiln’s history

Chapter 8 showed a lime kiln that ran its stack O2 near 7.1% through the fall, with 68% of its lime inside the carbonate band. In January the O2 moved up to about 7.6% and stayed there for eight months.

The best operating point was not new. The kiln had already run there. It was most likely not yet a written target with a daily check, so the move went unanswered.

So the study ended with two things, not one. A target: 7.0 to 7.2% stack O2. And a check: report the hours inside that window every day. The target captures what was learned. The check keeps it.

6.87.07.27.47.67.8Lowest cost7.0 to 7.2%Where historysays to runWhere you ranJanuary stepAug 25Nov 25Feb 26May 26Aug 26Monthly average stack O2, %
Figure 10.2. The same chart as Chapter 8, read a second way. The green points are the lesson. The band makes it the standard.

Questions for Monday morning

  1. After our last three investigations, which targets or procedures changed?
  2. If a key setting moved today, how long before someone would know?
  3. Where would a new engineer find what we learned from the last break study?
  4. Which problems have we solved more than once?

In practice. Envoy Intelligence answers a question from your trends, your procedures, your past cases and your centerlines, so the next investigation starts from the last one.