An Insight Works product
Pilot-stage · inviting pilot sitesPlantEdge AI — Multimodal industrial intelligence at the edge
Use the data you already have. Add sensing where it matters. Apply intelligence locally.
PlantEdge AI is a local-first industrial intelligence solution designed to combine existing plant automation data with dedicated real-time measurements where necessary, helping engineering and maintenance teams identify sustained equipment changes and review the supporting evidence.
What PlantEdge AI is
An industrial intelligence layer between plant data and engineering decisions
PlantEdge AI is a local-first industrial intelligence platform, initially focused on electric motors, VFD-driven systems and their control and process context. It is designed to acquire existing plant data and dedicated measurements, check their quality and timing, add equipment and operating context, establish normal behaviour, detect sustained change, assemble the evidence and raise diagnostic cases for engineering review, all on site.
PlantEdge observes, analyses and advises. Existing plant-control systems retain authority.
Engineering decisions
Maintenance planning · engineering review · planned intervention
PlantEdge AI — local intelligence layer
Context · baseline · change detection · evidence · cases
Plant data and measurements
PLC · VFD · SCADA · historian · existing and dedicated sensors
Conventional monitoring and PlantEdge
PlantEdge complements PLC, SCADA and historian systems. It does not replace them.
| Conventional monitoring | PlantEdge |
|---|---|
| Individual tags | Multimodal data |
| Instantaneous values | Equipment and operating context |
| Fixed thresholds | Normal-behaviour baselines |
| Alarms after the fact | Sustained-change detection |
| Historical charts | Data-quality awareness |
| Separate systems | Evidence assembled in one case |
| Manual interpretation | Diagnostic-case workflow and engineering review |
Both stay in service. PlantEdge reads from the systems on the left; nothing is removed or replaced.
Hybrid data acquisition
Existing data first. Dedicated sensing only where it improves the decision.
A · The default path
Existing plant data
- PLC measurements and operating states
- VFD measurements
- SCADA and historian data
- Protection and instrumentation signals
- Existing condition-monitoring signals
- Production and process context
B · Added per asset, after assessment
Dedicated measurements where justified
- Vibration
- Temperature
- Motor current and voltage
- Power
- Speed
- Relevant process or equipment measurements
PlantEdge data acquisition
Not every motor needs new sensors. Dedicated sensing is a per-asset decision, made after the existing data has been assessed.
Assess your existing plant dataMultimodal intelligence
Condition data interpreted in context, not in isolation
Electrical
Current, voltage, power, load
Mechanical
Vibration, speed, rotational characteristics
Thermal
Motor, housing and bearing-related temperature, where available
Process and operating context
Running or stopped, production mode, operating state, load, drive status
Historical
Prior behaviour, baseline, event history, previous diagnostic evidence
Why context matters
A vibration rise at constant load means something different from a vibration rise that follows a normal increase in process load. PlantEdge is designed to assess the vibration change alongside operating load, helping distinguish equipment behaviour from normal process variation.
Local-first architecture
Cloud optional. Industrial intelligence local.
On-site / edge boundary · read-only
- 1Plant equipment and existing systems
- 2Data acquisition
- 3Quality and time validation
- 4Asset and operating context
- 5Local analytics and condition intelligence
- 6Evidence-backed diagnostic cases
- 7Engineering review
Optional enterprise and cloud layer: fleet visibility, reporting, remote access
- Cloud required for core operation
- No
- Commands sent to equipment
- None
PlantEdge is read-only. It does not issue PLC commands, operate machinery, or replace protection or safety systems. Existing PLCs, safety systems and drives keep control of the process.
The analytics follow one order: deterministic engineering rules first, then machine learning, then AI-assisted interpretation, with every finding backed by evidence. Today’s MVP uses deterministic rules; the machine-learning and AI-assisted layers are planned.
Detecting sustained equipment change
Sustained, credible change, not a threshold crossing
- 1Normal operation
- 2Baseline established
- 3A change begins and persists
- 4Supporting context from other signals
- 5Attention raised
- 6Engineering review
Attention is raised on persistence and supporting context, not on the first excursion. Before a baseline exists, PlantEdge reports that data is insufficient rather than guessing.
Evidence-backed diagnostics
Every finding answers four questions
Motor M-203
ATTENTIONILLUSTRATIVE- What changed?
- Vibration rose progressively above the motor’s recent normal range.
- Why was it flagged?
- The increase persisted across several assessment periods.
- What evidence supports the finding?
- Vibration trend · motor current · load · temperature · operating state · timestamps · data-quality status.
- What should engineering review next?
- Engineering review recommended.
Attention is not the same as confirmed failure. PlantEdge supports investigation; it does not claim certainty it does not have.
Example case: illustrative asset and traces, not plant data.
Technical and operational value
Where the value comes from
- Earlier visibility of equipment deterioration
- Better maintenance prioritisation
- Less manual diagnostic effort
- More return on existing automation
- Focused instrumentation spending
- Local operation
- Evidence-supported engineering decisions
Mechanisms, not promised percentages. The value of PlantEdge is plant-specific, and each pilot measures it with the plant’s own figures.
Proposed pilot model
Start with a pilot. Prove the value. Scale.
One production line or process area · about 3–10 representative assets, a mix of critical and non-critical.
- 1Assess the existing data
- 2Identify measurement gaps
- 3Add sensing where justified (optional)
- 4Integrate and commission locally
- 5Establish the baseline
- 6Monitor and review with your engineers
- 7Evaluate technical and economic results
- 8Recommend whether and how to scale
Pilot outputs
Technical findings · Asset condition history · Evidence cases · Deployment lessons · A business-case assessment · A scale-up recommendation
Discuss a PlantEdge pilot
Your enquiry goes to our operations team at operations@insightworksng.com.
Product maturity
Where PlantEdge is today
PlantEdge AI is a pilot-stage product. The table separates what is designed, what has been demonstrated and what still needs field validation.
- Approved architecture and design objectives
- Local-first site runtime · two acquisition paths (existing data plus dedicated sensing) · read-only integration · data-quality and source labelling · evidence-backed diagnostic cases · human engineering review · optional enterprise layer
- Implemented and demonstrated in the motor-intelligence MVP
- Read-only acquisition of industrial data over Modbus TCP · data-quality states (current, stale, invalid) · source labelling as LIVE, REPLAY, SIMULATOR or TEST · local history · deterministic demonstration rules that raise Attention · diagnostic cases with an engineering-review workflow · local dashboard · operation without internet
- Simulated, replayed and emulated demonstrations
- End-to-end workflow on a simulated motor and drive · replays of recorded demonstration data · sensor-gateway behaviour against TEST emulators. Every demonstration screen carries its source label
- Planned for future phases
- Production motor-intelligence rules · learned normal-behaviour baselines · machine-learning and AI-assisted interpretation layers · enterprise and fleet reporting · asset classes beyond motors and drives
- Requires field validation through pilots
- Performance on live plant equipment · detection sensitivity and false-attention rates · installation effort · business value for each plant
PlantEdge is not a PLC, SCADA or historian replacement, a cloud-only service, an autonomous control system, guaranteed failure prediction, or definitive root-cause diagnosis.
Learn it on the same rig
SkillKatalyst courses train operators and engineers on the motors, drives and protocols that PlantEdge AI works with.
About SkillKatalyst training