2027 Event Agenda

Electric utilities are under pressure to move beyond AI pilots and deliver measurable operational value, yet many struggle to integrate advanced analytics into day-to-day grid decision-making. This session presents a utility-led implementation of AI-enabled decision support for engineering and operational planning, developed by Essex Powerlines in collaboration with Utilismart Corporation.  

At its core, the project integrates large volumes of metering and operational data to pre-work investigations, generate actionable operational insight, and improve engineering productivity bottlenecks. The project addresses manual, time-intensive investigations of grid issues, such as deviations, transformer stress, and event-related grid issues across disconnected systems, unnecessary truck rollouts due to a reactive response to one-off anomalies or non-actionable noise, and slower outage diagnosis and restoration contributing to higher SAIDI/SAIFI due to manual root-cause identification.  

The AI/ML solution was integrated into Utilismart’s SmartMAP as a decision-support layer to pre-work voltage, transformer, and event investigations the way experienced engineers do, reducing outage time and truck rollouts; consistently interprets grid operational data to predict failures in advance; reduces manual effort while preserving operator and engineering judgement through explainable outputs and human review; and enables proactive grid operations through real-time insights, improving decision making. 

Rather than focusing on model complexity, this work addresses a more persistent industry gap: how to operationalize AI outputs in a way that is trusted, explainable, and usable by system operators and planners. The solution combines predictive analytics with human-in-the-loop validation, enabling utilities to interpret peak forecasts, assess system constraints, and take informed action within existing workflows.  

If selected, the session will walk through technical architecture, data integration challenges, and the governance approach required to move from fragmented data sources to a production-ready decision support capability. It will also highlight where traditional approaches fall short, particularly the overreliance on “black-box” outputs, and how embedding explainability and operator feedback loops improved adoption and outcomes.  

Attendees will gain practical insights into:  

• Designing AI systems that augment (not replace) utility decision making 

• Integrating disparate operational and metering data for real-time planning  

• Establishing governance, validation, and trust in AI-supported processes 

• Lessons learned from deploying AI  

This session challenges the industry to re-think what “AI success” looks like, shifting from model performance metrics to operational usability and decision support.