The platform
One platform for the whole loop.
From the first sensor sample to the signed-off report, In-Sight runs detection, investigation, action, and learning in one system. Here is what that includes.
Detect what matters, not everything.
Detection in In-Sight is layered: statistical baselines catch drift, trained models catch what statistics cannot, engineering limits catch what physics forbids, and alarm analytics catch what your control system is already telling you.
Machine learning in production
Train models on your own historian data, deploy them to live inference, and watch their predictions become findings when risk crosses the line.
- SHAP explainability on every prediction
- Drift detection with scheduled checks
- Live precision published in the app
SPC and anomaly detection
Western Electric rules, Hotelling T² multivariate detection, and self-recalibrating baselines that learn each signal's normal by operating mode.
- Mode-aware baselines
- Multivariate correlation breaches
- Business-rule breach detection
High-rate capture
Stream vibration and other fast signals at up to 2 kHz per tag, loss-lessly, with momentary overshoot detection at ingest.
- Zoom from months to milliseconds
- Overshoots become findings instantly
- Configure rates from the app
Alarm management
Ingest OPC UA alarms and events and measure your alarm system against ISA-18.2 and EEMUA 191 benchmarks.
- Flood and chatter detection
- Bad-actor rankings
- Alarms correlated with findings, never merged
Limits with provenance
Engineering limits extracted from datasheets and SOPs, proposed to an engineer, and armed only on human confirmation.
- Every limit traceable to its source page
- Tiered: engineering, alarm, SPC
- Nothing is auto-armed
Operator rounds
Manual log sheets bring operator readings into the historian, so paper rounds join the same detection pipeline as sensors.
- Structured capture forms
- Readings flow into baselines and limits
- One pipeline for sensors and people
Investigate with evidence, not guesswork.
A finding is a starting point. In-Sight turns it into a case: an investigation with a timeline, an evidence chain, ranked hypotheses, and an AI copilot that has read your plant's documentation.
Worn sealing element
evidence 87%Film supplier batch change
evidence 41%Ambient temperature shift
evidence 12%Replacement interval exceededSealing Unit Manual · p.12
Investigation cases
Launch a case from any finding, roll related findings into it, and close it with a disposition that writes back to every rolled-up finding.
- Case timeline and ownership
- Disposition write-back
- Recurrence tracking after closure
AI copilot and Ask In-Sight
Ask questions in plain language, across live data, findings, and documents. The copilot drafts RCA narratives and ranks likely causes by evidence.
- Answers cite data and pages
- Abstains when it cannot source a fact
- Charts generated from your question
Document intelligence
Datasheets, SOPs, and manuals are parsed, chunked, and retrieved with hybrid search, so every answer can ground itself in the exact page.
- PDF parsing with vision for drawings
- Page-level citations
- Facts mined into proposed limits
Act, and leave a paper trail.
Analysis only earns its keep when it changes what happens on the plant. In-Sight carries a confirmed cause through to a dispatched work order, a notified assignee, a scheduled report on the right desk, and an audit trail of who decided what.
Replace sealing element, Station 2
Work orders and assignments
Turn a confirmed finding into a work order with the evidence chain attached, emailed to the planner or team, with optional SAP write-back.
- Evidence-rich dispatch emails
- Assignment notifications and reminders
- Stale-assignment nudges built in
Alert rules
Route what matters to the people who need it: metric thresholds, anomaly triggers, and compound conditions, delivered in-app and by email.
- Per-rule recipients and channels
- Confirm or dispute from the email
- Test delivery before you rely on it
Reports that write themselves
Compose reports from live sections: KPIs, charts, query tables, findings, and AI-drafted narrative. Generate Word documents on demand or on a schedule, emailed to your list.
- Real charts embedded in Word
- Scheduled generation and delivery
- Natural-language query sections
Dashboards
A builder with live tag binding on every tile type, from KPI tiles and SPC charts to heatmaps and flow diagrams.
- Per-tile signal binding
- Templates to start fast
- Cross-filtering drill-downs
Learn, so next shift starts smarter.
This is the stage most platforms leave out. Every confirmation, dispute, and disposition your team records becomes training signal. The system's knowledge base is not static configuration; it is an asset that appreciates.
Live precision
92%The learning loop
Operator verdicts become model labels. Baselines recalibrate on confirmed reality. Detection precision is tracked live, in the open.
- Feedback-driven retraining
- Baseline recalibration
- Live precision vs backtest, visible to all
Institutional memory
Every investigation, disposition, and root cause stays queryable. The plant's hard-won lessons stop walking out the door.
- Historical precedents on new findings
- Recurrence detection after fixes
- Knowledge survives crew changes
In-app guide
A full guide surface inside the product, so every role learns the platform in the place they use it.
- Role-aware articles
- Written in plain operational language
- Always current with the product
Connectivity
Meets your plant where it is.
No rip-and-replace. In-Sight connects to the systems you already run, from the control network to the ERP, and backfills history so day one starts with context.
Plus historical backfill, connection health monitoring, and gap tracking on every source.
See it run on your own data.
Bring a persistent problem from your own operation and we will walk the full loop on it in one session.