01Reactive maintenance
When the first clear signal is a breakdown, work becomes unplanned: parts are rushed, schedules shift, and technicians spend their time responding instead of preventing.
In development · pilot conversations open
Turn machine signals into earlier maintenance decisions.
MechSignal is being built to help maintenance teams bring machine readings and maintenance records together, notice unusual equipment behaviour sooner, and decide which machines to inspect first.
Illustrative product preview
One view from signal to inspection task
A working concept of the MechSignal interface. Every reading, asset and name below is sample data created for illustration — not output from a real facility or a validated model.
Equipment
5 assets| Asset | Status |
|---|---|
| P-104Centrifugal pump · Line 2 | Review suggested |
| C-302Air compressor · Utilities | Watch |
| M-211Drive motor · Line 1 | Within baseline |
| F-118Exhaust fan · Line 3 | Within baseline |
| G-407Gearbox · Line 2 | Baseline building |
Vibration velocity (RMS)
mm/s · 30 daysSample chart requires JavaScript. Values rise from about 2.3 to 4.1 mm/s over the last five days.
Motor housing temperature
°C · 30 daysSample chart requires JavaScript. Values rise from about 64 to 74 °C, above range on the last four days.
Anomaly flags
2 openVibration above baseline range
P-104 drive-end bearing — 5 consecutive days above learned range, increasing.
Temperature trending upward
P-104 motor housing — above range on last 4 days. May be related to the vibration flag.
Technician review tasks
Priority order1 · Inspect P-104 drive-end bearing
Check bearing condition, lubrication and coupling alignment. Record what was found.
Issue confirmedNo issue foundNeeds more data
2 · Check C-302 intake filter
Minor pressure deviation flagged last week. Technician noted filter was due.
Issue confirmedNo issue foundNeeds more data
Status labels use text and shape as well as colour. Flags indicate a deviation from that machine’s own baseline — they are prompts for inspection, not diagnoses or failure predictions.
The problem
Maintenance teams have data. It just isn’t in one place.
Many plants already collect machine readings and maintenance records — but in separate historians, spreadsheets, PLC exports and work-order systems. That makes it hard to see which machines are changing before someone has to react.
02Scattered equipment data
Sensor trends live in one system, inspection notes in another, and asset history in a third. Connecting them for a single machine can take hours of manual work.
03Unclear priorities
With dozens or hundreds of assets and limited inspection time, it’s difficult to know which machines genuinely need attention this week — and which can wait.
Planned capabilities
What MechSignal is designed to do
These capabilities are in active development. Scope and priority will be shaped with pilot partners.
Equipment health overview
A single list of monitored assets showing which are within their usual range, which deserve a look, and which don’t yet have enough data.
Sensor trend analysis
Readable vibration, temperature, pressure and current trends, plotted against each machine’s own operating baseline.
Anomaly detection
Statistical and machine-learning models that flag readings deviating from learned behaviour, with the evidence shown alongside each flag.
Maintenance prioritisation
A ranked inspection list that combines flag severity, asset importance set by your team, and recent maintenance history.
Technician feedback
Technicians record what they found. That feedback is kept with the asset and used to review and tune how flags are raised.
Asset history & reporting
A timeline of readings, flags, inspections and outcomes per machine, with exportable summaries for maintenance reviews.
How it works
Four steps from raw readings to a reviewed inspection
Connect compatible equipment data
Bring in readings from existing sensors, historians or gateways, plus maintenance records where available.
Establish operating baselines
Learn what normal looks like for each machine across its typical loads and operating modes.
Flag deviations
Highlight readings that move outside the learned baseline and rank them for review.
Support technician inspection
Turn flags into inspection tasks; technicians confirm what they found and the outcome is recorded.
Planned architecture
Planned on AWS industrial and machine-learning services
The current architecture plan. Components may change as development and pilot work progress.
AWS IoT SiteWise
Collect
Ingest and model industrial equipment data.
Amazon S3
Store
Keep historical readings for baselines and review.
Amazon SageMaker AI
Detect
Train and run anomaly detection models.
Amazon SNS
Notify
Send alerts to the right maintenance contacts.
MechSignal app
Review
Technicians inspect, respond and record outcomes.
Before you start
What a deployment needs
MechSignal works with data your equipment can provide. Before any pilot we’ll review your machines and data sources together.
See deployment requirements- Suitable sensors or existing data connections — for example vibration, temperature, pressure or motor current, via a historian, PLC, gateway or new sensors.
- Enough operating history to establish a baseline for each machine and operating mode.
- Validation on your machines. Model performance has to be measured on the relevant equipment before results are relied on.
- Existing safety systems stay in place. MechSignal is a decision-support tool, not a protective or safety system.
Exploring a pilot with your maintenance team?
We’re looking for a small number of manufacturers and service providers to help shape and validate MechSignal on real equipment.