Capabilities · Root cause analysis
Know why, not just what.
The Universal RCA Agent turns an alarm into a ranked list of causes, each with its evidence — in minutes, not shifts.
From alarm to cause in four steps.
Monitoring tells you something changed. Root cause tells you what to fix. The agent does the evidence-gathering your reliability engineers do today.
01
Gather the signals
Vibration, temperature, current and process data around the event, pulled from the device and your historian.
02
Check the failure modes
Each candidate cause is tested against the asset's FMEA, not a generic model.
03
Compare past repairs
Similar events and what fixed them, from your maintenance history.
04
Rank the causes
A short list with confidence and evidence, and the recommended fix for each.
- 1AlarmFrom monitoring or an operator.
- 2RCA agentSignals, failure modes and past repairs become ranked causes.
- 3Engineer confirmsOr corrects the diagnosis.
- 4Work orderDrafted for the confirmed cause.
- 5LearnCorrections become training data.
Skills for every failure mode.
Each failure mode is a reusable RCA skill: imbalance, misalignment, bearing wear, looseness, resonance, process upset. Six skills power the ID fan walkthrough today; new ones are added per asset class and shared through ElixirHub.
Rotating equipment
Fans, pumps, compressors, motors and gearboxes.
Process assets
Kilns, mills, coolers and heat exchangers.
Your own models
Forecasting and anomaly models become tools the agent can call.
Your engineers stay in charge.
- Every cause shows the evidence behind it, never a bare score.
- Engineers confirm or correct the diagnosis in one step.
- Corrections become training data, so the next diagnosis is sharper.
- Every step is recorded and can be replayed in an audit.
Common questions.
How is this different from predictive maintenance?
Predictive maintenance stops at the alert. The RCA agent explains the alert, ranks the causes and hands a diagnosis to the work-order agent.
Do we need years of labelled failures?
No. The agent starts from your FMEA, manuals and repair history, and improves as technicians confirm causes.
Where does it run?
On the device beside the asset, on premises or in a private cloud — your data stays on site.
Start with one critical asset.
A four-week assessment on one asset class: baseline, diagnosis, and a pilot plan your engineers sign off.