Most Platforms Make You Choose: Edge or Cloud. Why Not Both?
Edge-only platforms give you real-time but no cross-site visibility. Cloud-only platforms give you dashboards but die when the internet drops. TwinEdge runs both — each layer does what it does best.
Four Layers, One Platform
Data flows up from equipment through the edge to the cloud. Commands flow down from the cloud through the edge to equipment. Work orders flow sideways from insights to Operations Intelligence.
Watch the green data packets rise, the purple OTA updates descend, and the amber work orders dispatch sideways.
The Edge-Only Trap
Edge-only platforms are great for real-time. But without a cloud layer, you end up with islands of data that never talk to each other.
No Cross-Site Visibility
Each site is an island. You cannot compare pump #3 in Plant A against pump #3 in Plant B. No cross-site learning, no enterprise-wide failure patterns.
No Centralized Management
Updating a physics model means SSH-ing into 200 edge devices one by one. No OTA, no model registry, no version control.
No Scale Economics
Every new site is a greenfield deployment. No shared configurations, no templates, no knowledge transfer between sites.
The Cloud-Only Trap
Cloud-only platforms look great in demos. But when the internet drops, when latency matters, or when bandwidth bills arrive, the cracks show.
Latency Kills Real-Time
A 200ms round-trip to the cloud is 200ms too late for a safety interlock. Real-time decisions cannot wait for internet.
Cloud Outage = Total Blindness
Internet goes down at your rural water plant? Cloud-only means zero monitoring, zero alerting, zero data logging until it comes back.
Bandwidth Costs at Scale
Streaming raw 1-second sensor data from 500 devices to the cloud costs thousands per month. Most of that data is normal and does not need to leave the site.
Not edge OR cloud.
Edge AND cloud.
Each layer does what it does best. Real-time physics at the edge. Cross-site analytics in the cloud. Maintenance action in Operations Intelligence. When one layer is unavailable, the others keep working.
What Each Layer Does For You
Four layers, each solving a different problem. Together they cover every operational need from sensor to wrench-turn.
Equipment Layer
Speaks every protocol your equipment speaks
Modbus TCP/RTU, OPC UA, BACnet, EtherNet/IP, S7, GPIO. No protocol translators, no middleware. Direct connection to PLCs, sensors, VFDs, and controllers.
Edge Layer
Real-time intelligence that never goes offline
Physics models at 1-second intervals. ML inference on ONNX Runtime. Local alerting in under 1 second. Local storage for days of data. Runs autonomously when the cloud is unreachable.
Cloud Layer
Cross-site visibility across every facility
Cross-site benchmarking. Failure pattern recognition across 10,000+ assets. Compliance reports aggregated from all facilities. Model registry pushes updated ML models to any edge device via OTA.
Ops Intel Layer
Closes the loop from insight to action
Condition-based work orders triggered by actual degradation. Technician dispatched with the right parts to the right asset. GIS-mapped spatial context. Maintenance history feeds back into ML training.
What This Means For You
Three real scenarios showing how the architecture works in practice. Not theory -- actual operational workflows.
Bearing Failure at a Remote Pump Station
Physics model detects vibration signature change at 2:14 AM. Fires local alert. Logs data locally.
Correlates with 47 similar failures across fleet. Predicts 3-4 weeks to failure. Estimates $12K repair vs $85K catastrophic.
Auto-generates work order. Reserves bearing from nearest warehouse. Schedules technician for next planned window.
Internet Outage at a Treatment Plant
Continues monitoring all 24 sensors. Runs physics and ML models locally. Fires alerts to on-site operators. Stores data for 7 days.
Shows "Last Sync: 4h ago" on operations dashboard. No data loss. When connection restores, edge backfills all stored readings.
Critical work orders cached locally on edge. Technician mobile app works offline with synced asset data and inspection checklists.
New ML Model Deployed Across 200 Sites
Data science team trains improved anomaly detection model on fleet-wide data. Publishes to model registry. Targets pump assets only.
OTA update delivers new ONNX model to 200 edge devices in staged rollout (canary, 10%, 50%, 100%). Auto-rollback on error rate spike.
New model catches bearing defects 2 weeks earlier. Work orders now created with higher lead time. Parts availability improves 40%.
Switching from another platform?
We'll migrate your data from ANY CMMS or analytics platform — free. Assets, work orders, maintenance history, sensor configurations — everything.
See the Architecture in Action
Book a demo and see how edge intelligence, cloud analytics, and operations intelligence work together for your specific use case.