‘Rocket fuel’ for employees: Inside a utility’s AI journey

When Startex-Jackson-Wellford-Duncan Water District (SJWD) in South Carolina started experimenting with artificial intelligence, it started where a lot of utilities start — with chatbots.

Following the COVID-19 pandemic, the utility experienced population growth leading to an increase in customer inquiries. To help meet demand, the utility introduced AI-powered customer service agents, including Rosie, a chat assistant; Eve, a voice assistant that handles after-hours calls; and Vespera, a Spanish-language voice assistant.

They saw immediate success: Rosie resolves about 99% of customer interactions without requiring a transfer to a human while maintaining high customer satisfaction scores, said Jeff Diaz, SJWD’s chief information services officer. Eve and Vespera handle hundreds of after-hours calls each month and successfully resolve about 84% of those interactions.

That success led the utility into more specialized AI applications, including department-specific assistants tailored to individual business functions. The utility is also exploring digital twins and systemwide modeling capabilities that could simulate operations from source water through billing.

“Our CEO [Billy Cothran] calls AI ‘rocket fuel’ for our employees,” Diaz said, because it enhances productivity. “Each group has their own [AI assistant] and we structured it that way so we had a team lead — a knowledge lead, someone who had a lot of experience in the district and in the industry — that could help us train and hone these things.”

Today, AI supports a variety of activities across the district in Spartanburg. Some tools help analyze non-revenue water data, identify property and easement opportunities, or perform geographic information system (GIS) queries. Other AI assistants have been trained in utility policies, procedures, and operational knowledge to provide department-specific support. The engineering team’s assistant, for example, can access mapping data and help evaluate new development needs and flow calculations.

The utility views AI as an enhancement tool, not a replacement strategy, Diaz said. Rather than eliminating positions, the utility focuses on using AI to help employees access information faster, reduce repetitive tasks, and make more informed decisions. Leadership involved employees in developing use cases and training the systems, which ultimately garnered trust and encouraged adoption across the organization, he said.

“Everybody has embraced this because we designed it this way,” he said. “We designed it so it was everybody’s system.”

By working closely with experienced employees, the utility is capturing decades of operational and institutional expertise and incorporating that knowledge into its AI systems. Diaz said he believes future utility teams will combine highly trained staff, AI agents, and advanced data models to solve complex water challenges.

Beyond its own operations, utility leaders share lessons with other utilities through industry collaborations and research initiatives. Diaz said that knowledge sharing is part of the utility sector’s broader responsibility to help advance the industry.

“Water and power are used to generate a lot of these ideas,” he said. “We have a stewardship duty to share some of the things we create so we don’t have to reuse water and power to do that.”