“Agentic AI will matter most in utilities when it becomes the disciplined bridge between data and action. Water, gas, and electric utilities already collect enormous amounts of information from meters, sensors, maps, customer systems, and field records. The real breakthrough is not simply having more data. The breakthrough is using AI to interpret that data, recommend the next step, and help utility professionals act faster, safer, and with better evidence.” – MJ Martin
Introduction
Agentic AI is artificial intelligence that can understand a goal, plan the required steps, use software tools, examine data, make decisions, and take action within defined limits. It is an agent that works on your behalf, just as an employee might do.
For water, gas, and electric utilities, this is a major development because these utilities already operate large networks of meters, sensors, customer records, field crews, billing systems, asset systems, and regulatory processes. The challenge is not only collecting data. The real challenge is converting data into useful operational action.
A traditional AI tool may summarize a report, explain a billing pattern, or identify an abnormal meter read. Agentic AI can go further. It can review the abnormal read, compare it to historical consumption, check the customer class, examine recent field work, confirm meter configuration, prepare a work order, draft a customer notice, and flag the issue for approval. In utility operations, this ability to connect data, judgment, and workflow is where agentic AI becomes valuable.
Background
Water, gas, and electric utilities have spent decades building digital systems. AMR and AMI collect meter readings. MDM systems validate meter data. CIS platforms manage customer accounts and billing. GIS maps assets and service locations. Work management systems schedule field activity. SCADA monitors operational networks. ERP systems manage finance, inventory, and procurement. These systems are powerful, but they are often separated by function, vendor, data structure, and operating department.
This separation creates manual work. Staff may need to open several systems to understand one problem. A missing read may require checking the meter, endpoint, account, route, installation history, network status, and billing record. A high bill complaint may require reviewing consumption history, temperature, occupancy, meter exchange records, leak patterns, estimated reads, and customer communications. Agentic AI is useful because it can help assemble this context and guide the next step.
Applications
Agentic AI does not replace utility expertise. It supports it. It acts as a workflow layer across systems, helping trained staff interpret evidence, manage exceptions, and complete routine processes more consistently.
In water utilities, agentic AI can support leak detection, continuous consumption review, district metered area analysis, pressure zone monitoring, non revenue water programs, and customer notification. A water agent could review AMI data for continuous low flow, compare usage against historical patterns, check whether the property recently had a meter exchange, confirm whether a leak notice was already sent, and prepare a recommended customer communication. For larger systems, it could help correlate pressure, flow, acoustic leak alerts, and reservoir data to prioritize field investigations.
In gas utilities, agentic AI can support consumption anomaly review, meter renewal planning, pressure related analysis, safety documentation, and field service coordination. A gas agent could compare usage against weather normalized history, identify unusual zero consumption, review meter age, check Measurement Canada seal timelines, and recommend whether a site requires inspection. It could also help prepare job packages for technicians, including meter type, regulator information, access notes, photos, and safety requirements.
In electric utilities, agentic AI can support demand meter validation, bidirectional meter reconciliation, distributed energy resource monitoring, outage analysis, transformer loading review, and billing exception management. An electric agent could review interval data, compare delivered and received energy, confirm rate class, check meter form, validate CT or PT ratios, and flag discrepancies before billing. For customers with solar generation or other distributed energy resources, the agent could help reconcile import, export, net energy, demand, and time of use values.
Across all three utility types, agentic AI can improve field operations. It can help group work orders by geography, check parts availability, prepare technician instructions, review completed photos, compare scanned serial numbers against expected assets, and flag incomplete work. It can also support customer service by summarizing account history, explaining consumption patterns, preparing plain language responses, and reducing the time spent moving between CIS, MDM, GIS, and work order systems.
Issues and Problems
The greatest concern is that utilities are critical infrastructure organizations. Agentic AI must not be allowed to take high consequence actions without proper controls. Changing a billing record, dispatching a crew, modifying an outage record, issuing a disconnect notice, or making an operational control decision must require defined authorization. The more serious the consequence, the stronger the human approval requirement.
Data quality is another major issue. Utility data is often imperfect. Meter numbers may not match across CIS, MDM, GIS, and field systems. Address formats may vary. Service locations may be wrong. Photos may be missing. Old meter exchange records may be incomplete. If agentic AI relies on poor data, it may produce poor recommendations. This is especially important for utilities because a small data error can affect billing accuracy, customer trust, regulatory compliance, and field safety.
Cybersecurity must also be treated as a core design issue. Agentic AI may need access to customer records, meter data, network information, work orders, maps, emails, documents, and asset records. Access should be controlled through role based permissions, multi factor authentication, audit logs, and clear separation between business systems and operational technology. Utilities should begin with low risk administrative and analytical use cases before considering any connection to operational control systems.
Accountability is essential. Every agentic workflow should record what data was reviewed, what conclusion was reached, what action was recommended or taken, and who approved it. This is necessary for customer disputes, regulatory reviews, safety investigations, and internal quality assurance.
What’s Next
The best near term use cases for utilities are practical and controlled. Water utilities can begin with leak review, high consumption notices, meter exception management, and non revenue water support. Gas utilities can begin with consumption anomalies, meter renewal planning, safety documentation, and field work preparation. Electric utilities can begin with demand meter review, bidirectional meter validation, interval data exceptions, transformer loading analysis, and distributed generation reconciliation.
The future model should be supervised autonomy. The AI agent gathers evidence, compares records, prepares recommendations, drafts communications, and completes low risk administrative work. Human staff approve actions that affect customers, billing, safety, compliance, or network operation. This model protects the utility while still delivering efficiency.
Over time, agentic AI may become a standard layer in utility operations. It will sit between AMI, MDM, CIS, GIS, ERP, OMS, SCADA, and work management systems. Its role will be to reduce the friction between data and action. Utilities that have clean data, strong governance, secure integrations, and clear operating procedures will gain the most benefit.
Summary
Agentic AI is artificial intelligence that can understand a goal, plan a workflow, use tools, analyze data, and act within controlled boundaries. For water, gas, and electric utilities, it offers a practical way to improve exception management, customer service, field operations, billing support, asset planning, and regulatory documentation.
Its value is strongest where utilities already have data but still depend on manual effort to interpret and act on that data. Water utilities can use it to improve leak response and non revenue water programs. Gas utilities can use it to support meter renewal, anomaly review, and safety processes. Electric utilities can use it to validate demand, interval, bidirectional, and distributed generation data.
Agentic AI should not be deployed as an uncontrolled autopilot. It should be governed as a critical enterprise system with permissions, audit trails, cybersecurity, human approval points, and clear accountability. Used properly, it becomes a new operational intelligence layer for utilities, helping skilled professionals make faster, better supported decisions across complex networks.
About the Author:
Michael Martin is the Vice President of Technology with Metercor Inc., a Smart Meter, IoT, and Smart City systems integrator based in Canada. He has more than 40 years of experience in systems design for applications that use broadband networks, optical fibre, wireless, and digital communications technologies. He is a business and technology consultant. He was a senior executive consultant for 15 years with IBM, where he worked in the GBS Global Center of Competency for Energy and Utilities and the GTS Global Center of Excellence for Energy and Utilities. He is a founding partner and President of MICAN Communications and before that was President of Comlink Systems Limited and Ensat Broadcast Services, Inc., both divisions of Cygnal Technologies Corporation (CYN: TSX).
Martin served on the Board of Directors for TeraGo Inc (TGO: TSX) and on the Board of Directors for Avante Logixx Inc. (XX: TSX.V). He has served as a Member, SCC ISO-IEC JTC 1/SC-41 – Internet of Things and related technologies, ISO – International Organization for Standardization, and as a member of the NIST SP 500-325 Fog Computing Conceptual Model, National Institute of Standards and Technology. He served on the Board of Governors of the University of Ontario Institute of Technology (UOIT) [now Ontario Tech University] and on the Board of Advisers of five different Colleges in Ontario – Centennial College, Humber College, George Brown College, Durham College, Ryerson Polytechnic University [now Toronto Metropolitan University]. For 16 years he served on the Board of the Society of Motion Picture and Television Engineers (SMPTE), Toronto Section.
He holds three master’s degrees – in business (MBA), communication (MA), and education (MEd). As well, he has three undergraduate diplomas and seven major certifications in business, computer programming, internetworking, project management, media, photography, and communication technology. He has completed over 80 next generation MOOC (Massive Open Online Courses) [aka Micro Learning] continuous education programs in a wide variety of topics, including: Economics, Python Programming, Internet of Things, Cloud, Artificial Intelligence and Cognitive systems, Blockchain, Agile, Power BI, Big Data, Design Thinking, Security, Indigenous Canada awareness, and more.
Martin in a volunteer, a photographer, a learner, a technologist, a philosophizer, and a romantic optimist.