Sample tank · live estimateOn schedule
Home heating oil · Barrie · Orillia · Midland
You already give propane customers a tank they never have to think about. This is the same thing for heating oil, driven by the degree-days over their roof and the burn rate we already know. No more will-call phone tag. No more emergency runs at −22. The truck shows up before the gauge does. This is a concept view of what that could look like.
The prediction, one account at a time
Sample customer — 909 Concession 4, Oro-Medonte. 910 L tank, no telemetry. The model reads degree-days off the local weather feed, multiplies by this home's measured K-factor, and draws the tank down day by day. It schedules the truck before the tank hits reserve — not after the customer calls cold.
Colder days burn faster, so the line steepens. The model watches the forecast and moves the delivery up when a snap is coming — it doesn't wait for the calendar.
Built for the person in the dispatch chair
This isn't a portal you log into all day. It's a queue that fills itself: the accounts that need oil this week, already sequenced by route, before anyone panics.
Every oil account the model flags for the next 7 days, sorted so nobody's driving to Midland twice in a week.
A cold snap moves deliveries up automatically. The dispatcher sees the change, doesn't have to do the math at 6am.
The expensive drops are the surprise ones after hours. Predicted delivery turns a panic into a planned stop on tomorrow's route.
Every fill recalibrates that tank's burn. The longer a customer's on it, the tighter the estimate — same as your propane telemetry, minus the hardware.
Clustering predicted drops by geography means more litres per shift and fewer half-empty runs across three branches.
It rides on the burn history you already have. It's the dispatcher's memory, written down and always awake.
You already proved this works
The propane app, the tank telemetry, the auto-delivery — that was the hard part, and you already shipped it. Oil is the same customer promise reached a different way: forecasting instead of a sensor.
Why this pitch, from someone who's run ops
I spent years in IT operations — dispatch, SLAs, field service — so I know where the day goes sideways: it's the surprise call, the emergency run, the thing nobody saw until it was already on fire. I don't sell software and I'm not adding to your payroll. I take route knowledge that already exists and give it a forecasting engine, so the trucks do more drops per shift and dispatch stops chasing runouts.
"Free the dispatcher" isn't a slogan here — it's the whole job. A predicted list beats a panicked phone every single winter.
Pick one route — Barrie, Orillia, or Midland — and we run the model on that book for a season before it touches anything else. Prove-on-one, then scale. Coffee's on me.