Data before AI. Fifty steps back, five hundred forward.
By the end of this module you can explain why AI is the payoff and not the starting line, spot the pump with three names in your own systems, and start the Monday list without calling a single vendor.
Put the AI down for a second
I build AI for water utilities for a living, so this is either the worst sales pitch you have heard all year or the most honest one.
Every room in our field now arrives at the same comfortable sentence within ten minutes: AI should augment the people who run our utilities, not replace them. True, and by now obvious. Everyone nods, including me. Then the conversation stops at the exact moment it should start, because the real question is not whether AI should help. It is whether we are ready to be helped.
We are playing catch-up, and the numbers are not a rounding error.
$0T+US DRINKING WATER NEEDS, 25 YRS
≈0%OF REQUIRED CAPITAL ACTUALLY SPENT
≈0%OF OPERATORS RETIRING IN A DECADE
And while we run to catch up, the workforce that knows how the system actually behaves is walking out the door, with its knowledge scattered across systems that were never built to talk to each other. So my message to the sector is not "move faster." It is the opposite. Take fifty steps back so you can take five hundred steps forward.
THE PUMP WITH THREE NAMES · two plants, one machine, no fleet
CMMS · PLANT EAST
Pump 3Awork orders: 14
SCADA HISTORIAN
P3A-EASTtags: 9
WORK ORDER LOG
Pump #3 (east)entries: 22
PUMP 3Aone name · one history · 45 records, finally one machine
✓ Now the fleet-failure model can see it. AI is the payoff, not the starting line.
Same model of pump in two plants. The agent you bought cannot predict failures across the fleet, not because the AI is weak, but because it cannot tell these are the same machine. The step back is boring, and it is everything.
✅ CHECK THE FOUNDATION
Your new fleet-failure agent underperforms. What is the first suspect?
Education before automation
Look at who works in a utility: chemical engineers, electrical engineers, civil engineers, modelers, planners, operators who can hear a failing pump before a sensor catches it. A field that folds half a dozen engineering disciplines into one working system every day has no business telling itself it cannot understand a model. We can. We simply have to learn it before we buy it.
Be careful what you put on autopilot. Pilot, experiment, learn, by all means. But the line is not "no AI." The line is no autonomy on ground you have not checked. Do not hand an agent a live treatment process while your understanding of that process lives in one retiring operator's head.
The Monday list
None of the steps below require my company, or any company. You can begin all of it on Monday, with the staff you already have. Click through it.
The Monday List
the fifty steps back, in five moves, zero vendors
1
Agree on the naming.What do we call the pump? A meeting, not a capital program.
2
Standardize the tags.One convention across plants, enforced at entry.
3
Clean the asset register.Make the record match reality, machine by machine.
4
Write down what the retiring operator knows.Before the retirement party, not after.
5
Teach your people what these systems do.A workforce that understands its tools does not get captured by them.
If you do this list and never call me, the advice still worked. That is the only kind worth writing down.
🧠 NOW RUN IT ON YOURS
Name one asset class where the same thing is logged under different names in your systems.
Your answer goes to the sparring brain. Watch the graph light up the pages it reads before it coaches you.
Your data is your utility
We keep saying our data is fragmented. I think that is the symptom, not the disease. Data is not fragmented. People are fragmented. We keep having the same conversation in a hundred separate rooms, and if we want the five hundred steps forward, we take them together or we do not take them at all.
Your data is your utility. The condition of your pipes, the behavior of your plant, the memory of every operator who ever solved a problem at 2am. AI is not there to understand your system in your place. It is there to do far more with what you already understand, and it can only reach as far as your own grasp of your own data will let it.
AI is not the starting line. It is the payoff. You would not hand a new analyst the keys to the trading account on day one.
✅ CHECK THE FOUNDATION
What is the line on AI autonomy in a utility?
TAKEAWAYS
The real question is not whether AI should help. It is whether you are ready to be helped.
The pump with three names is why your agent fails: the record does not match reality.
The Monday list needs no vendor: naming, tags, register, the retiring operator's knowledge, education.
No autonomy on ground you have not checked. Pilot, learn, but keep the controls human.
Data is not fragmented; people are. The next module is about the weakest of the links.