Abstract

As grid and load complexity increases and distributed generation proliferates, utilities face a torrent of sensor data that outpaces human capacity to interpret it. This paper introduces Merlin™, an analyst-grade AI system from Power Monitors, Inc. (PMI) that applies large-language-model reasoning to power quality (PQ) analysis. Rather than replacing engineers, Merlin™ amplifies their interpretive capacity by classifying, ranking, and summarizing PQ findings in plain language — backed by deterministic, standards-based computations. The paper explores how this human-centered AI architecture improves detection, diagnosis, and institutional learning, while preserving full auditability, explainability, and cybersecurity integrity.