Your Journey with Rela AI
How Rela AI transforms your maintenance from reactive to predictive, step by step
Your Journey with Rela AI
Rela AI doesn't ask you to change everything overnight. The system learns progressively from your equipment and gets more accurate every week. Here's how it works.
Level 0: Detection (week 1-2)
What happens: you connect your sensors, PLCs or SCADA systems to Rela via MQTT, OPC UA or HTTP. The system starts collecting data and building baselines of normal behaviour.
What you see:
- Basic alerts when something drifts from normal.
- WhatsApp or email notifications in real time.
- Live event dashboard.
Immediate value: you no longer depend on someone watching a monitor. Rela watches 24/7 and pings instantly.
Level 1: Health tracking (month 1-2)
What happens: with 10+ data snapshots, the Asset Health Index (AHI) stabilises. You start seeing degradation trends in the dashboard.
What you see:
- Health score 0-100 per equipment (A, B, C, D, F).
- Trends — is vibration rising? Did temperature level off?
- First useful-life estimates (low-to-medium confidence).
Value: visibility. For the first time, you see the "health" of your whole fleet in one place.
Level 2: Prediction (month 2-4)
What happens: with 30+ snapshots and at least one logged failure, the predictive model calibrates. RUL (Remaining Useful Life) predictions gain confidence.
What you see:
- Predictions — "this compressor will fail in ~120 hours".
- Failure probability at 24h, 72h, 7 days.
- Automatic recommendations — "intervene within 48 hours".
- Work orders recommended and pre-assigned by the system (subject to approval).
Value: transition from reactive to predictive maintenance. Instead of waiting for the failure, you intervene earlier — planned and controlled.
Level 3: Optimised (month 4+)
What happens: with multiple logged failures and technician feedback, the model reaches high confidence. Recommendations are orchestrated end-to-end, always with human approval before anything runs on equipment.
What you see:
- Predictions with 80%+ confidence.
- Work orders recommended and pre-assigned to the best technician (subject to approval).
- Orchestrated WhatsApp coordination with human approval.
- Impact KPIs — downtime hours avoided, cost saved.
Value: demonstrable ROI. Fewer unplanned stops, fewer emergencies, more production.
The promise
"We start as an anomaly-detection system and build the predictive-maintenance model in real time from your operational history. Every logged failure, every completed intervention, makes Rela more accurate for your specific operation."