Story of a solution · Industrial kitchens

Oil filtration management, machine by machine.

How a simple question from a client — "how do I know my people are filtering the oil?" — became a system that today watches more than 40 Broaster Company fryers across stores in Santander, and writes a report to management every night.

+40fryers monitored in Santander
9stores
24/7real cycle detection
1daily AI report

How it started

A regional restaurant chain with Broaster Company pressure fryers had a problem familiar to anyone running several locations: oil filtration depended on each operator doing it, on every shift, on every machine. The paperwork said one thing and the state of the oil said another. The oil degraded early, product quality varied between stores, and a supervisor cannot be in nine places at once.

The client's question was blunt: "how do I know they are filtering?" And the honest answer was that, with paper forms, there is no way to know. We had to measure the machine, not ask the operator.

How we solved it

We started with sensors at a pilot store and grew to the full fleet. Along the way, real operations taught us everything a system like this has to withstand.

01

Measure the machine, not the self-report

We installed a sensor on the filtration pump circuit of every fryer. Each real cycle is detected by its electrical signature: when it started, how long it ran, and whether it was a routine filtration or a deep oil treatment. Impossible to dress up on a form.

02

Learning from the false positives

The first versions over-alerted: machines that were switched off and "were not filtering", end-of-shift routines that looked like faults. We refined the rules against thousands of real cycles — grace windows, powered-off detection — until we removed 86% of the alerts that arrived stale. A Goldcas alert today means something.

03

Surviving the real world

A power cut mid-filtration? The system recovers the cycle without a false alarm. The pump jams? It is caught by its current signature the same day. Need to improve the sensor software? It updates over the air across 30+ machines, without visiting a single store and with a safe rollback.

04

Closing the loop with intelligence

Data nobody reads is useless. Every night, an AI engine reviews more than 40 fryers and writes the report a human analyst would take hours to assemble: who filtered and how many times, who did not, which machine behaved oddly, ranking by store.

What it does today

Watches the whole fleet

More than 40 Broaster Company fryers across 9 stores, every filtration cycle detected and stored, every day for over a year.

Alerts with judgment

The store that has gone too long without filtering gets its alert within minutes — and management finds out without calling anyone.

Reports every night

The AI summary arrives by email before the morning meeting, with what matters first.

Zero data entry

Nobody fills anything in. The machine measures itself and the report writes itself.

Do you run fryers across several locations?

Tell us how many machines and where. We design the solution around your real operation — just as we did with this one.

Write to Goldcas