Technology

The intelligence layer for energy control

Hardware-agnostic. Protocol-rich. Integrated with KNX, Control4, BACnet, Modbus, and major BMS platforms. The platform does not only measure power — it orchestrates it within engineered bounds.

Platform in Operation

Live monitoring, load prediction, and flexible control

Deployments today in real buildings — integrating with existing automation to turn sites into responsive contributors to grid stability.

Real-time monitoring and control

Monitor electrical signatures, track device behaviour, and coordinate loads — processed on premises.

Electrical Signatures (NILM)

Every load has a measurable electrical fingerprint

Non-intrusive load monitoring identifies and tracks individual devices from high-frequency data captured at a single measurement point, without adding per-device sensors.

Non-intrusive monitoring

Single-point measurement identifies all devices. No per-machine sensors.

Pattern recognition

Models learn the signatures of loads in each environment.

Anomaly detection

Surfaces unusual patterns that indicate wear, fault, or misconfiguration.

Behavioural Modelling

Load patterns, occupancy, and cycles — learned continuously

Occupancy patterns

Understand when spaces and plant are in use; adapt systems accordingly.

Routine recognition

Daily, weekly, and seasonal cycles identified automatically.

Adaptive modelling

Continuously refine as operational patterns evolve.

External Context

Integrates weather, generation, pricing, and calendar data

Weather integration

Anticipates heating/cooling demand from forecast data.

Daylight tracking

Adjusts artificial lighting against natural light levels and sun position.

Calendar awareness

Adapts to events, holidays, and scheduled activities.

Seasonal adaptation

Adjusts behaviour models across seasons and climate conditions.

Edge Architecture

All inference, locally

All models run locally on the on-premises device. Operational data stays on site. Real-time responsiveness is guaranteed — no round-trip latency to a cloud.

Optional secure cloud connectivity is available for remote access, fleet operations, and multi-site reporting — but is never required for normal operation.

Predictive Optimisation

From signatures to interventions

Fault detection

Identifies equipment issues weeks ahead of failure.

Pre-conditioning

Prepares spaces for optimal comfort in advance of occupancy.

Load balancing

Shifts non-critical loads to reduce peak demand charges.

Continuous improvement

Performance improves as model learns site-specific patterns.

Technical Summary

Device characteristics

Local processing

Dedicated NPU handles inference locally at high frequency.

Open integrations

KNX, Control4, Vantage, BACnet, Modbus, MQTT, and more.

Resilient connectivity

Ethernet, Wi-Fi, optional cellular. Operates without connectivity.

Deploy the intelligence layer on your site

Active grid participation with deterministic, on-premises control.