{"id":589936,"date":"2026-10-19T21:28:22","date_gmt":"2026-10-19T21:28:22","guid":{"rendered":"https:\/\/www.esri.com\/en-us\/industries\/blog\/?post_type=blog&p=589936"},"modified":"2026-10-22T16:09:37","modified_gmt":"2026-10-22T16:09:37","slug":"covington-water-district-unlocks-utility-knowledge-with-an-in-house-ai-assistant","status":"publish","type":"blog","link":"https:\/\/www.esri.com\/en-us\/industries\/blog\/articles\/covington-water-district-unlocks-utility-knowledge-with-an-in-house-ai-assistant","title":{"rendered":"Covington Water District Unlocks Utility Knowledge with an In-House AI Assistant"},"content":{"rendered":"

GIS-Powered Innovation Is Breaking Down Information Silos and Putting Answers at Staff and Customer\u2019s Fingertip<\/strong>s<\/h2>\n\n

A developer evaluating a property contacts Covington Water District<\/a> with seemingly straightforward questions: Is the parcel within the district\u2019s service area? Where is the nearest water main? Is there an existing meter? Will a developer extension agreement be required? What might be the connection cost?<\/p>\n\n

Answering those questions often requires staff to review geographic information system (GIS) data, development standards, permitting requirements, customer records, asset information, and engineering documentation stored across multiple systems. What appears to be a simple request can quickly become hours of research.<\/p>\n\n

Covington Water District saw an opportunity to change that. Rather than building another reporting tool, the district developed an AI-powered assistant that allows employees to interact with utility information through natural conversation, receiving maps, charts, reports, and answers in seconds. The result is a fundamentally different way to access and use knowledge across the organization.<\/p>\n\n

\n
\"\"
The Covington Water District Service Area<\/em><\/figcaption><\/figure>\n<\/div>\n\n

Covington Water District serves approximately 50,000 customers across a 55-square-mile service area in King County, Washington. The district manages more than 90,000 assets and 325 miles of transmission and distribution infrastructure with a staff of just 45 employees.<\/p>\n\n

GIS is a core operational system, with roughly 80 percent of employees using GIS-related applications as part of their daily work. Over the years, the district implemented specialized systems for billing, asset management, permitting, GIS data, and operational monitoring. While each system supports a critical business function, information often resides in separate repositories.<\/p>\n\n

As the volume of data grew, staff increasingly needed to search multiple systems, consult subject matter experts, and assemble information from different sources before providing complete answers.<\/p>\n\n

The Challenge<\/strong><\/h2>\n\n

Covington Water District\u2019s challenge was never a lack of data, but was the effort required to access it.<\/p>\n\n

Developers, customers, engineers, operators, and managers routinely need answers from information that spans multiple systems. Questions about water availability, infrastructure proximity, consumption patterns, service connections, and development requirements often require significant research.<\/p>\n\n

The district processes hundreds of water availability requests annually along with countless phone calls and customer inquiries. Valuable information existed throughout the organization, but accessing it often depended on knowing whom to ask and where to look.<\/p>\n\n

\n
\"\"
The assistant allows users to interact with the results in a variety of ways, ranging from raw numbers to charts to map outputs.<\/em><\/figcaption><\/figure>\n<\/div>\n\n
\n
\"\"<\/figure>\n\n
\"\"<\/figure>\n\n
\"\"<\/figure>\n
Users interact with the data using plain-language prompts.<\/em><\/figcaption><\/figure>\n\n

\u201cWe have many systems and many experts in each system,\u201d said Covington Water District lead developer Dan Dulan. \u201cBut we didn\u2019t have a layer that sat on top of everything and allowed people to interact with all of that information at once.\u201d<\/p>\n\n

GIS at the Core<\/h2>\n\n

The GIS team knew that GIS had to be at the center of any solution.<\/p>\n\n

\u201cGIS touches everything,\u201d said Dan Dulan. \u201cEverything in the district has a spatial component.\u201d<\/p>\n\n

Pipes, meters, parcels, customers, facilities, and infrastructure all share this characteristic: location. That made GIS the logical framework for bringing information together across departments.<\/p>\n\n

\u201cLocation stays the same even when ownership changes,\u201d said engineering department manager Steve Lee. \u201cGIS is the central nervous system.\u201d<\/p>\n\n

The district\u2019s ArcGIS Utility Network<\/a> implementation created a modern geospatial foundation capable of connecting information across business systems and operational workflows. Today, the AI assistant can access utility network information, GIS feature attributes, business records, and supporting documents through a single conversational interface.<\/p>\n\n

\n
\"\"
Users can generate maps based on attribute information stored in the GIS data.<\/em><\/figcaption><\/figure>\n<\/div>\n\n

Staff can ask questions in plain language and receive maps, reports, analytics, and visualizations without needing specialized GIS or database expertise.<\/p>\n\n

From Reports to Conversations<\/strong><\/h2>\n\n

Unlike traditional reporting tools that answer one question at a time, the district\u2019s assistant supports an interactive conversation.<\/p>\n\n

A user might begin by asking which service area consumed the most water and then request a map, compare historical trends, filter results, or generate a chart. Each question builds on previous responses, allowing staff to investigate issues as they emerge.<\/p>\n\n

\u201cBeing able to talk with your data in natural language is very powerful,\u201d said Dan Dulan. \u201cThe data is out there, but it\u2019s locked.\u201d<\/p>\n\n

Instead of requiring users to know where information resides, the system helps them focus on the questions they need answered. Questions that once required multiple systems, several experts, and days of effort can now be explored in minutes.<\/p>\n\n

\n
\"\"
 Each question becomes obvious only after the answer preceding it.<\/em><\/figcaption><\/figure>\n<\/div>\n\n

ASKED AND ANSWERED\u2014REAL QUESTIONS, AS TYPED<\/strong><\/p>\n\n

Location\/Spatial Identification<\/strong><\/p>\n\n