About — Manitiq

Manitiq was started by people who have stood in a plant at 3 a.m. with an alert, a spectrum, and no part on the shelf. We are building the half of predictive maintenance that everyone else stopped short of.

Why now

Sensors got cheap. Low-power radios got good. And models that once needed a vibration analyst to interpret a spectrum can now learn a machine’s own behaviour and name the mechanism when it changes. Every vendor in this category has used that to sell alerts.

But the plants we work with do not have an alert problem. They have a planning problem: a fault has been identified, and someone still has to work out the part, check the store, find the procedure, estimate labour and cut the order. That gap is where predictive programmes go to die, and it is the part we decided to build.

Principles

The work order is the unit of value.
A prediction nobody schedules is a cost. We measure ourselves on orders raised, parts on the shelf, and outages that didn't happen — not on alerts sent.
Every machine is its own machine.
Fleet thresholds fail on the plant floor because no two installs load the same way. We model each asset against itself, and we show the evidence.
Cheap enough for the long tail.
Most outages come from assets that were never 'critical' enough to instrument. Coverage matters more than precision on ten machines.
Meet planners where they work.
Nobody wants another dashboard. The output lives in the CMMS, the parts system and the technician's phone.
Say what we know, and how sure we are.
Confidence bands, evidence links, and a feedback loop that admits when the model was wrong. Trust on a plant floor is earned per work order.

Team

We are a small team of reliability engineers, embedded and RF engineers, and machine-learning researchers who have shipped condition-monitoring hardware and maintenance software before. We are hiring across firmware, signal processing and CMMS integrations.

careers@manitiq.com