AI in the Water Industry: Lessons from Early Adopters

On Tuesday, November 17th the Asian American Architects / Engineers Association Orange County hosted a panel on “AI in the Water Industry: Lessons from Early Adopters.” The session featured speakers Charlie Eckstrom, CIO/IT Group Manager with the Metropolitan Water District of Southern California (MWD), Francisco “Frank” Hernandez, Director of Information Services with the Orange County Water District (OCWD), and @Christopher Tull, Principal Data Engineer with the Moulton Niguel Water District (MNWD) and Chief Data Officer for the California Data Collaborative. Moderator Peter Martin, Digital Solution Consultant with Trinnex, led the panelists through a discussion establishing baseline knowledge around types of AI, current applications of AI within the various districts, maintaining a policy or guiding principles that allow for innovation, and how to get started. 

The audience was shown an image mapping out various types of AI including large language models (LLMs), generative AI, AI agents, and agentic AI. This distinction is essential for safe AI experimentation, and for understanding and distinguishing the many offerings currently marketed as “AI”. Panelists shared that many are already using AI to brainstorm event titles, help with emails, search for public feedback online, write grants, or assist with basic coding but they emphasized the importance of establishing certain guardrails within an organization before letting a more independent type, like agentic AI, run wild. 

Panelists spoke about how to draft an AI policy or set of guidelines. OCWD released an AI policy this summer that emphasizes the need to protect sensitive information for both customers and employees and prohibit data usage for model training. Since AI is constantly evolving, flexibility is key. MWD is working to stay on the cutting edge of AI integration at their agency. They update their policy every six months as technology and needs evolve, and they encourage internal pilot projects. One major example of AI use at MWD is the use of AI-assisted software development to create proofs of concept in just a few hours to address specific agency needs. Charlie Eckstrom shared that their development team can now bring an internal application to production in just two weeks. 

The discussion mentioned that a good way to initiate AI use at an agency is to integrate it into specific projects, rather than applying it on top of all operations. OCWD is using AI to reverse engineer some helpful but outdated legacy software, and MWD is using a drone team equipped with AI to predict dam anomalies before issues arise. Meanwhile MNWD has applied AI to demand forecasting for recycled water in order to right-size water storage.

Christopher Tull highlighted how data governance is core to successful AI deployment. By pulling from quality source data, having clear access points, and working to establish a single source of truth so AI outputs can be validated, you end up with more useful results. MNWD has been working on building this foundation through an initiative called Building a Leading Utility Ecosystem (BLUE) which involves both establishing governance over data assets and also centralizing disparate data into a single platform for unified analysis. 

Adopting AI will continue to present challenges as advancements are made. However, by working to establish a strong data foundation, allowing innovation within boundaries, and placing cybersecurity at the forefront, more players in water can reap the benefits.