Solutions/Physics-informed reasoning
Trace

Go beyond predicting failure. Know what to do about it.

Trace is a specialised small language model for equipment. It interprets behaviour, reasons about causes within physics and engineering constraints, and recommends the intervention.

ANOMALY MISALIGNMENT BEARING WEAR IMBALANCE INTERVENTIONFOR REVIEW ILLUSTRATIVE EXAMPLE
From anomaly to possible causes to a recommended intervention, reviewed by your engineers.
Problem

Alerts without answers

Monitoring tells you something is wrong. It rarely tells you why, or what to do about it on this machine.

Solution

A model built for your machines

Trace uses custom-built small language models, grounded in physics and engineering knowledge, to predict the failure and prescribe the fix.

Trace SLM · Compressor C-204Custom model
Sensor dataSignalsVibration and temperature
Physics modelsValidatedBearing and rotor dynamics
Engineering knowledgeFailure modesHow this machine fails
Equipment dataHistoryCompressor C-204
ConstraintsOperating limitsSpeed, load, temperature
Maintenance recordsPast workEarlier interventions
Predicted failureDrive-end bearing wear, progressing.
Prescribed actionReplace the bearing at the next planned stop and check lubrication.
Illustrative example
Trace SLM engine

Predict the failure.Prescribe the fix.

Built for your equipment.

Each Trace model is a small language model shaped around specific equipment and its data. It is not a general-purpose chatbot.

Read more →

Grounded in physics.

Engineering knowledge, validated physics models and operating constraints keep its reasoning inside what is physically plausible.

Read more →

Prescriptive, with people in control.

Trace assesses possible causes and recommends the intervention. Your engineers review every recommendation.

Read more →

Bring us a hard equipment problem.

Discuss a Trace use case