AI use cases in energy and utilities: 12 systems to build in 2026

The AI use cases utilities build first are short-term load and renewable forecasting, storm outage prediction for crew staging, vegetation and asset risk scoring from imagery, AMI meter analytics and outage communications. Each one reads data the utility already collects in SCADA, the historian, the OMS, GIS and the meter data system, and each leaves switching, dispatch and safety decisions with operators.

The twelve systems below cover electric, gas and water utilities and independent power producers, grouped by grid operations, field assets, customers and regulatory work. For each we list the inputs, the systems it connects to, the hard part and a planning assumption for team and timeline. Ryz serves enterprises across industries, including energy. Our AI pod teams build in your cloud and repos, working on your team with your grid, GIS and cybersecurity staff.

A note on the figures. Teams and durations are planning assumptions for illustration, never quotes. Every dollar range comes from Ryz's public rates: $7,000 to $10,000 for mid-level engineers, $10,000 to $15,000 for senior engineers and $15,000 and up for tech leads, per engineer per month and quoted per team. We use the $15,000 lead floor in the math.

Compliance boundaries for utility AI

• NERC CIP. Systems that touch bulk electric system cyber assets fall under the NERC Critical Infrastructure Protection standards (CIP-002 through CIP-014), which govern electronic security perimeters, access management and information protection. Most AI work stays on the corporate side and receives operational data through approved one-way paths.
• Gas pipeline records. Operators under PHMSA's 49 CFR Part 192 keep integrity management and maximum allowable operating pressure records that must be traceable, verifiable and complete. AI can find gaps; it cannot fill them with guesses.
• EU operations. The EU AI Act's Annex III treats AI used as a safety component in the supply of water, gas, heating or electricity as high-risk. Under the 2026 Digital Omnibus amendment, those Annex III obligations apply from December 2027.
• State regulators. Public utility commissions review rate cases, wildfire mitigation plans and customer communications, so anything that feeds a filing needs a clear record of its sources.

Grid operations

1. Short-term load and renewable forecasting

Models predict load by feeder, substation and system for the next hours and days, along with expected solar and wind output, so operators and traders plan unit commitment, purchases and switching with fewer surprises.

• Inputs: historical load, weather forecasts and observations, calendar effects, behind-the-meter solar estimates, plant telemetry.
• Connects to: AVEVA PI or another historian, the energy management system, trading and scheduling tools.
• The hard part: rooftop solar hides real demand, and extreme weather days are exactly where history is thinnest. Forecasts need uncertainty bands, not a single number. See machine learning development.
• Planning assumption: tech lead plus two senior engineers and one mid-level engineer, four to six months: $42,000 to $55,000 a month, $168,000 to $330,000 in total.

2. Storm outage prediction and crew staging

Before a storm, a model combines forecast wind, rain and ice with asset age, vegetation and past outage data to predict where outages will cluster, so the utility stages crews, mutual aid and materials in the right yards.

• Inputs: historical outage records, weather model output, GIS asset data, vegetation and soil data.
• Connects to: outage management system, Esri ArcGIS Utility Network, emergency operations dashboards.
• The hard part: big storms are rare and each one is different, so models must be validated storm by storm and explained to the emergency team in their terms.
• Planning assumption: tech lead plus three senior engineers, five to seven months: $45,000 to $60,000 a month, $225,000 to $420,000 in total.

3. Transformer and substation asset health

The system reads dissolved gas analysis, oil tests, load history, thermal readings and maintenance records for transformers and breakers, scores condition and ranks units for inspection, replacement or load relief.

• Inputs: DGA and oil test results, nameplate data, loading history, work orders, infrared inspection results.
• Connects to: asset management or EAM (IBM Maximo, SAP PM), historian, capital planning.
• The hard part: condition data sits in lab PDFs and spreadsheets, and failures of large units are rare. Engineering judgment sets the final ranking.
• Planning assumption: tech lead plus two senior engineers, four to five months: $35,000 to $45,000 a month, $140,000 to $225,000 in total.

Field assets

4. Vegetation risk from imagery and LiDAR

Satellite imagery, aerial LiDAR and inspection photos are processed to measure clearance between conductors and trees, flag hazard trees and estimate growth, so trim cycles target the spans with the highest risk instead of a fixed calendar.

• Inputs: LiDAR point clouds, satellite and aerial imagery, span geometry, trim history, outage cause codes.
• Connects to: GIS, vegetation work management, trim crew scheduling tools.
• The hard part: image dates and seasons, and point-cloud volumes that need careful data engineering. In wildfire regions the output can feed a regulator-reviewed mitigation plan. See computer vision development.
• Planning assumption: tech lead plus four senior engineers, six to eight months: $55,000 to $75,000 a month, $330,000 to $600,000 in total.

5. Pole and line inspection image review

Drone and helicopter inspections produce tens of thousands of photos. A vision model finds damaged insulators, cracked crossarms, woodpecker holes and missing hardware, and queues only the suspect images for an inspector to confirm.

• Inputs: labeled inspection images, defect taxonomy, asset IDs and GPS tags.
• Connects to: inspection platforms, GIS, work management for repair orders.
• The hard part: matching each photo to the right pole, and keeping recall high on severe defects even when that means more images to review.
• Planning assumption: tech lead plus three senior engineers, four to six months: $45,000 to $60,000 a month, $180,000 to $360,000 in total.

6. Field crew standards assistant

A lineworker or gas technician asks how a specific connector is rated or which construction standard applies to a pole configuration. The assistant answers from the utility's own construction standards, switching rules and manufacturer data, cites the page and never improvises on safety procedures.

• Inputs: construction standards, safety manuals, equipment specifications, past work notes.
• Connects to: field service mobile apps (IFS, ServiceNow Field Service), document management, identity provider.
• The hard part: offline use in trucks with poor coverage, and strict revision control. See RAG development.
• Planning assumption: tech lead plus one senior and one mid-level engineer, three to four months: $32,000 to $40,000 a month, $96,000 to $160,000 in total.

7. Pipeline integrity records review

For gas operators, the system reads mill test reports, pressure test records, as-built drawings and work orders, links them to pipe segments and lists segments where records needed for MAOP or integrity management are missing or inconsistent.

• Inputs: scanned historical records, GIS pipe segments, test and inspection data.
• Connects to: GIS, integrity management software, document management.
• The hard part: decades-old handwritten records. The output is a gap list for engineers to resolve, which keeps it consistent with Part 192's records expectations.
• Planning assumption: tech lead plus two senior engineers, four to six months: $35,000 to $45,000 a month, $140,000 to $270,000 in total.

Customers and meters

8. Outage communication voice and text agent

During outages, an agent takes outage reports, gives estimated restoration times from the OMS, sends proactive updates and escalates downed-wire and medical-need calls to a person immediately. A Ryz AI pod built an AI voice platform that has placed 1M+ outbound calls.

• Inputs: OMS events and restoration estimates, customer contact data and preferences, call scripts.
• Connects to: OMS, customer information system (Oracle Utilities, SAP IS-U), telephony, SMS platform.
• The hard part: call surges during storms, and safety calls that must never sit in an automated queue. Outbound messages also need consent under the TCPA. See AI voice agent development.
• Planning assumption: tech lead plus three senior engineers, four to six months: $45,000 to $60,000 a month, $180,000 to $360,000 in total.

9. AMI analytics: high bills, theft and failing meters

Interval data from smart meters is scanned for patterns: a high-bill complaint explained by a new appliance load, a meter that reads zero while the premise is occupied, a bypass signature that suggests theft, or a transformer that is overloaded by the homes behind it.

• Inputs: interval reads, meter events, premise and account data, field investigation outcomes.
• Connects to: AMI head-end (Itron, Landis+Gyr), meter data management system, CIS, field work orders.
• The hard part: data volume and false positives. A theft flag sends an investigator, never an automatic charge or disconnection.
• Planning assumption: tech lead plus two senior engineers and one mid-level engineer, four to six months: $42,000 to $55,000 a month, $168,000 to $330,000 in total.

10. Contact center agent assist for billing questions

While a rep is on the phone, the assistant shows the customer's usage, rate plan, payment arrangements and assistance programs they qualify for, and drafts the notes. Reps resolve more on the first call without searching five screens.

• Inputs: CIS account data, tariff and program rules, call transcripts, knowledge articles.
• Connects to: CIS, contact center platform (Genesys, Amazon Connect), CRM.
• The hard part: tariff rules are complex and change after each rate case; answers must come from the current tariff. See AI agents for customer service.
• Planning assumption: tech lead plus two senior engineers, three to four months: $35,000 to $45,000 a month, $105,000 to $180,000 in total.

Planning and regulatory work

11. Interconnection application intake

Developers of solar, storage and other projects submit interconnection requests with site control, one-line diagrams and technical data. The system checks completeness against the tariff's requirements, extracts the technical parameters into study models and drafts deficiency notices for engineers to review. FERC Order No. 2023 moved transmission providers to cluster studies with firm deadlines, which raised the cost of slow intake.

• Inputs: applications and attachments, tariff requirements, past deficiency letters.
• Connects to: interconnection queue software, study tools, document management.
• The hard part: one-line diagrams as images, and deadlines that make every missed deficiency costly.
• Planning assumption: tech lead plus two senior engineers, three to five months: $35,000 to $45,000 a month, $105,000 to $225,000 in total.

12. Rate case and regulatory research assistant

Regulatory staff ask what the commission decided on a cost recovery issue in past cases, or how a peer utility answered a data request. The assistant searches commission dockets, orders, testimony and the utility's own data request responses and cites each source.

• Inputs: docket filings, orders, testimony, discovery responses.
• Connects to: document management, regulatory case tracking.
• The hard part: very long documents and the need to cite exact pages. Testimony is still written and sponsored by a person.
• Planning assumption: tech lead plus one senior engineer, two to three months: $25,000 to $30,000 a month, $50,000 to $90,000 in total.

How to pick your first use case

• Stay outside the CIP perimeter at first. Forecasting, imagery, records and customer systems can run on the corporate side with data copied out. That avoids months of security review for a first project.
• Use the storm calendar. If outage prediction or outage communications is the goal, plan the pilot to be live before the next storm season, and test on replays of past storms.
• Pick data you already trust. AMI and OMS data are usually cleaner than old asset records. Records cleanup is valuable, but it is a slower first win.
• Bring in the regulatory view early. Anything that changes capital plans, vegetation cycles or customer bills may show up in a filing. Write the methodology as you build.
• Check readiness. The AI readiness scorecard takes about ten minutes and covers data access and governance.

What it costs

Total cost equals engineers times months times each person's monthly rate. Ryz's rates are $7,000 to $10,000 for mid-level, $10,000 to $15,000 for senior and $15,000 and up for tech leads, quoted per team.

• Regulatory or records pilot (tech lead plus one senior engineer) for three months: $25,000 to $30,000 a month, $75,000 to $90,000 in total.
• Imagery and vegetation pod (tech lead plus four senior engineers) for eight months: $55,000 to $75,000 a month, $440,000 to $600,000.
• Full AI pod of about seven senior engineers including a tech lead and an ML engineer: $75,000 to $105,000 a month.

Imagery licenses, drone flights, telephony, cloud and model usage are paid in your accounts and are not included. Our AI development cost guide shows how to budget the rest. You meet the named engineers first, and the team can usually start in under two weeks once you say go. If you want a packaged ADMS or vegetation platform with vendor support, license one; Ryz pods build systems on your own data and integrate them with what you already run.

Related

• Energy engineering teams
• Hire data engineers
• Data engineering services
• AI agents for IT operations
• AI use cases in manufacturing
• AI use cases in the public sector

FAQ

What are the main AI use cases for utilities?

Load and renewable forecasting, storm outage prediction, transformer health scoring, vegetation risk from imagery and LiDAR, inspection image review, AMI analytics, outage communications and interconnection intake are the systems utilities most often put into production.

Does NERC CIP prevent using cloud AI?

Not for most use cases. CIP applies to BES cyber systems and the information about them. Many AI systems use operational data copied to the corporate network or cloud through approved paths, and BES Cyber System Information needs the access and protection controls CIP-011 describes.

Can AI control the grid directly?

Utilities generally keep switching and dispatch with operators and certified control systems. AI advises: forecasts, risk scores, suggested actions. Closing the loop would bring the system into the most demanding reliability and security requirements.

How accurate is AI outage prediction?

It depends on the utility's outage history and the storm type. The useful measure is whether staging decisions improved on past storms, so pilots are usually tested on replays of several historical events before a live season.

Can a Ryz pod work with our GIS, OMS and CIS?

Yes. Pods build against Esri, Oracle Utilities, SAP, GE Vernova, Schneider Electric, Itron and similar systems through their APIs and integration layers, in your cloud and repos. You control access at all times and own all code and IP.

Questions we didn't answer? Email info@ryzlabs.com.