Industrial facilities and large-scale campuses face persistent pressure to reduce energy costs, improve reliability and meet sustainability and ISO 50001 continuous-improvement expectations. Yet most conventional Energy Management Systems (EMSs) remain reactive in metering and reporting consumption at aggregated levels while obscuring asset-level inefficiencies such as partial-load losses, standby waste, scheduling misalignment and progressive performance drift. This paper presents an AI-enabled energy management framework implemented on an Intelligent Operations Analytics platform with an Agentic AI system to convert heterogeneous operational data into real-time, equipment-level decision support. The proposed architecture ingests data from smart meters, sensors, PLC/SCADA/BMS systems, enterprise databases and tariff sources; harmonises these streams into a unified analytical layer; and automatically computes standardised Energy Performance Indicators (EnPIs) including energy intensity (kWh/h), cost and baseline deviation. Continuous baselining and anomaly detection quantify avoidable energy losses and translate them into financial impact. When deviations are detected, advanced agents provide Root Cause and Corrective Actions (RCCAs) guidance with incident context, affected systems, criticality, confidence scoring and stepwise troubleshooting aligned with PDCA/DMAIC workflows. Results across representative asset classes (compressors, chillers, boilers, HVAC, pumps/motors and auxiliary utilities) demonstrate that asset-centric analytics reveal inefficiencies masked by facility-level reporting and enable timely, targeted interventions without code-intensive model development. The framework supports rapid time-to-value, scalability across zones and sites, and extensibility to predictive maintenance, cooling optimisation and broader Industry 4.0/5.0 operational intelligence.
Churi et al. (Wed,) studied this question.