Technology

Physics-informed AI built specifically for buildings. 4-layer architecture. Explainable. Robust. Production-ready.

4-Layer AI Architecture

Layer 1

Physics-Informed Digital Twin

Physics-Informed Neural Network (PINN) combines building physics equations with data-driven learning to model your building with high accuracy.

Layer 2

Supervisory Intelligence

Learns patterns from weather, occupancy, and operations to set optimal targets for comfort, energy, and cost across the building.

Layer 3

Plant Optimization

Optimizes chillers, boilers, pumps, and other central plant equipment for maximum efficiency with multi-objective control.

Layer 4

Distribution & Zones

Coordinates hundreds of zones in parallel using distributed optimization to maintain comfort while minimizing energy.

Physics-Informed AI

Most AI platforms treat buildings as black boxes. OCTR combines thermodynamic equations with machine learning, making decisions that obey physical laws while adapting to real data.

Requires only ~7 days of training data vs. 30+ days for pure ML
53% more accurate than baseline ML models in internal benchmarks
3.4× more robust to sensor failures and missing data

Explainable Control

Every control decision is validated against physical constraints and explainability methods (e.g., SHAP/LIME-style analyses), giving operators visibility into why changes happen.

This reduces the trust gap between AI systems and facility teams, and supports regulatory and ESG reporting requirements.

Multi-Objective Optimization

OCTR does not just minimize energy. It optimizes across five competing objectives simultaneously: energy, cost, comfort, equipment health, and regulatory constraints.

Energy Use