AI Agent Operational Lift for Des Moines Water Works in Des Moines, Iowa

Why AI matters at this scale

Des Moines Water Works (DMWW) is a mid-sized municipal utility serving over 500,000 residents in central Iowa. With 201-500 employees and an estimated annual revenue around $75 million, it operates at a scale where operational efficiency directly impacts ratepayers. AI adoption in this sector is nascent but accelerating, driven by aging infrastructure, workforce retirements, and the need to do more with less. For a utility this size, AI isn't about moonshot projects—it's about practical tools that extend asset life, reduce energy consumption, and improve service reliability. The convergence of affordable cloud computing, mature machine learning models, and accessible IoT sensors makes this the right moment to move beyond basic SCADA monitoring toward predictive intelligence.

Three concrete AI opportunities with ROI framing

1. Predictive maintenance for distribution mains. DMWW manages over 1,300 miles of water mains, many installed decades ago. A machine learning model trained on historical break data, pipe material, soil corrosivity, and pressure fluctuations can predict failure likelihood. Prioritizing just the top 5% of high-risk pipes for replacement could reduce annual main breaks by 20-30%, saving millions in emergency repair costs, liability, and water loss. The ROI is direct and measurable within the first year of deployment.

2. Real-time leak detection and non-revenue water reduction. By applying anomaly detection algorithms to flow and pressure data from district metered areas, DMWW can identify hidden leaks far faster than manual night-flow analysis. Reducing non-revenue water by even 5 percentage points—from a typical 15% to 10%—translates to hundreds of millions of gallons saved annually, directly lowering treatment and pumping costs. This use case pairs well with existing AMI meter investments.

3. Energy optimization for high-service pumping. Water treatment and pumping are energy-intensive, often representing the largest operational expense after personnel. Reinforcement learning can dynamically schedule pumps to leverage off-peak electricity rates and reservoir storage capacity without compromising pressure or fire flow requirements. A 10-15% reduction in energy costs could yield six-figure annual savings, with the added benefit of reducing the utility's carbon footprint.

Deployment risks specific to this size band

Mid-sized utilities face unique hurdles. Data often lives in siloed, legacy systems not designed for analytics—SCADA historians, work order databases, and GIS platforms may not talk to each other. The IT/OT convergence required for AI introduces cybersecurity vulnerabilities that smaller teams may struggle to manage. Perhaps the biggest risk is talent: attracting and retaining data scientists is difficult for a public-sector utility. Mitigation involves starting with a focused, vendor-supported pilot, investing in data integration middleware, and upskilling existing engineering staff rather than hiring a large new team. Governance around AI-driven operational decisions—especially those affecting water quality—must be established early to maintain public trust.