01 — The challenge
A hundred controllers, each telling its own story.
The controllers already measured conductivity, pH and dosing at customer plants — but the vendor portal showed one unit at a time, event logs left only as manual CSV exports, and nothing pushed an alarm to a person. Reading the fleet meant hundreds of clicks. Spotting a plant drifting out of its treatment band meant somebody happening to look.
For a data-science team that wanted to predict problems rather than log them, the missing piece was not a model. It was a system that held every reading, every band and every plant in one structure.
02 — What we built
One structure for the fleet, and an engine that explains itself.
Ranger360 went live on 2 August 2026, two days after the project started, and has been shaped in review rounds with the client's team since — a 22-point review on 13 August, a 13-point catch-up on 18 August, a demo on 25 August. Each round became a shipped release, not a wish list.
The core is a detect → diagnose → explain → recover → prevent → predict loop. A reading that crosses a band raises an event with the likely cause and the next action; a reading trending towards a band is flagged up to fourteen days early; a fresh in-band reading closes the event on its own. The troubleshooting knowledge came from the client's own contingency list and cause-effect diagrams — twenty workflows an engineer can walk through on screen.
03 — The results
From a vendor portal to a diagnostic system the team owns.
A live controller at a customer plant now reports every minute into a historian of more than 91,000 rows, and the platform models the fifteen lab parameters the client's list called for. Fifty-six controllers across five customers were loaded in the first weeks, with a pilot set of fourteen chosen for the direct feed.
The platform runs on the real fleet only — no demo data — and every administrative change passes a second pair of eyes. That is what a corporate team needs before it trusts a number on a screen.
Before · After
Dorf-Ketal's controller monitoring: before vs after Ranger360
| Metric |
Before |
After |
| Seeing the fleet |
One controller at a time in a vendor portal |
Every controller in one hierarchy |
| Alarms |
Nobody notified; logs exported by hand |
Four-band alarms with a notification matrix |
| Troubleshooting |
Contingency list in a spreadsheet |
20 guided workflows on screen |
| Prediction |
— |
Trend flagged up to 14 days before the band |