WHAT YOU’LL TAKE AWAY

  • Look across departments for repeated decisions and information handoffs.
  • Treat a use-case catalog as a map of possibilities, not proof of value.
  • Build shared data and review capabilities that support several workflows.

AI opportunities appear wherever people turn information into a decision or a completed task. In a utility, that spans almost every department: finding an inspection, reviewing a capital proposal, explaining a bill, preparing evidence, or planning tomorrow’s work.

The Senpilot source workbook contains 139 named use cases across engineering, regulatory, customer service, operations, and platform capabilities. The count describes the breadth of its catalog. It does not mean every case is implemented, independent of the others, or suitable for every utility.

Use that breadth to find a practical entry point and to see what different workflows could share.

Engineering and asset management

The workbook’s engineering examples include transformer health, loading analysis, asset comparison, demand forecasting, vegetation management, and capital prioritization. These tasks combine records, time series, spatial context, and domain analysis.

A useful starting point might prepare a reviewable asset evidence package or surface loading exceptions. More consequential planning applications need validated inputs, appropriate analytical tools, uncertainty treatment, and qualified review.

AI can help assemble and explain the information. The method used to establish an engineering conclusion must remain appropriate to the decision.

Regulatory and finance

Regulatory examples include precedent research, filing checklists, consistency checks, interrogatory organization, variance explanations, and commitment tracking. They offer repeated opportunities to connect a question with the evidence needed to answer it.

An effective workflow preserves source versions and links material claims to supporting passages or calculations. It should also make contradictions and unanswered questions visible. A well-written paragraph is only one component of an accepted filing package.

Finance and regulatory teams can share reconciled figures, assumptions, and evidence with planning teams, reducing the need to rebuild the same context at each handoff.

Customer service

Representative assistance, billing anomaly detection, information capture, and outage communications appear in the source catalog. Start by identifying what the customer needs resolved and which systems hold the authorized answer.

For an illustrative representative assistant, useful behavior includes finding the current policy, showing the relevant account context, and preparing the next step. Sensitive or complex cases need clear escalation. Communications about operational conditions need current approved information and an honest treatment of uncertainty.

Measure resolution quality and repeat contact alongside handling effort.

Operations and field work

Work order tracking, workforce replanning, route planning, inventory forecasting, and inspection support connect analysis to physical work. These opportunities depend on current job status, locations, skills, constraints, and asset identities.

An initial application can organize incomplete work packages or prepare suggested schedules for a dispatcher. Any extension into consequential operating decisions needs an explicit authority boundary and evaluation suited to that role.

Field information also improves the foundation: a confirmed correction or a well-linked inspection can support future planning and reporting.

Adapting the patterns to gas and water

The supplied catalog’s engineering content is predominantly about electricity. Gas and water examples here are editorial adaptations of the same workflow patterns, not demonstrations or claims from the workbook.

A water utility could explore linking inspection notes to main records or organizing evidence for recurring service complaints. A gas utility could explore matching asset documentation or checking work package completeness. Each requires domain-specific data, validation, and operating controls; an electric network model cannot simply be relabeled.

Find the shared foundation

Across these domains, the recurring needs are recognizable: dependable identities, current sources, access controls, evidence, review, and a place to record the completed action.

Explore the use-case library, then choose one workflow with a named owner and a measurable outcome. Build its supporting capabilities so that the next team can use them too. The opportunity becomes more practical when each project strengthens the foundation for the next.

Sources & further reading

Practical guidance combines the source material below with editorial analysis. Examples and suggested approaches are illustrative.

  1. Senpilot Global List of AI Use Cases in Utilities, September 2026

    Contains 139 named cases across Engineering, Regulatory, Customer Service, Operations, and Platform; engineering examples are predominantly electric.

  2. Senpilot website: engineering, regulatory, and customer service

    The connected platform and shared model inform the cross-department perspective.

Back to articles