Kyber Atlas
The living digital twin for any infrastructure network.
Grid Topology AI
See the real grid
Not the map on the wall. The network as it is, right now.
Atlas core →
- TelemetryIngest
- Digital TwinModel
- Fault LocationLocate
Activate telemetry into a living model of the network.
Atlas is the foundation of the Physical Intelligence Platform. It ingests raw electrical telemetry from the SCADA systems, smart meters, and IoT sensors a utility already owns, then uses AI state estimation and graph neural network models to reconstruct the actual real-time state of the network — solving the Dark Grid problem.
See how a utility implements Atlas →From raw signals to a precise segment of the network.
Operators stop synthesizing five disconnected systems under pressure. Atlas gives them one continuously updating picture of topology, load, and risk.
Telemetry, already in the building.
Connect existing SCADA, AMI, and IoT feeds. Atlas does not require a rip-and-replace of the sensors a utility already paid for.
A digital twin that does not go stale.
Graph models rebuild the live topology as switching, load, and weather change — not a GIS snapshot from last quarter.
Faults located in milliseconds.
Anomalous signatures are detected before they become failures, then pinned to a precise feeder segment instead of a county-wide outage ticket.
How Atlas works
/01
Telemetry ingest
Atlas reads voltage, current, breaker status, and meter events from the systems already in the control room and in the field. Missing or conflicting points are treated as signals, not as reasons to stop.
/02
State estimation
AI-driven state estimation fills the Dark Grid: the parts of the network no one can see because maps are partial and data systems do not talk to each other.
/03
Graph neural models
The network is modeled as a living graph. Every node and edge updates as the physical system changes, so the twin stays aligned with the iron in the ground.
/04
Fault location
When a signature appears, Atlas locates it to a segment in milliseconds — hours faster than complaint-driven dispatch — and hands that location to Command and Sentinel.
How a utility implements Atlas
Atlas is built for municipal utilities and rural electric cooperatives first: organizations that cannot buy enterprise grid software priced for companies ten times their size.
01
Connect what you already have
Point Atlas at SCADA historians, AMI head-ends, and existing IoT. No new sensor fleet is required to stand up the first twin.
02
Reconcile the graph
Atlas compares as-built maps against live telemetry and surfaces the gaps — the laterals, ties, and switching states the GIS never captured.
03
Stand up the twin
Operators get a continuously updating topology: load, voltage, open points, and predicted risk on one surface.
04
Turn on location
Fault and anomaly location goes live. The control room sees the segment before the call center fills with complaints.
05
Hand off to action
Atlas feeds Command for dispatch and switching, and Sentinel for millisecond isolation at the edge. The same twin serves every layer.
Atlas across critical infrastructure
The same topology engine adapts by domain. Atlas maps the network wherever civilization depends on a physical graph.

