AI-driven predictive maintenance and AR field tools are the two most practical technology overlays on utility asset management today. They address different problems, one predicts which assets need attention before failure, the other helps the technician who shows up to do the work. Both depend on the ERP as the system of record.
AI for Predictive Asset Maintenance
The operating model for AI-driven predictive maintenance in utilities is well-established at the concept level, but results vary significantly based on data quality.
The AI system consumes sensor data (thermal, vibration, partial discharge, pressure), maintenance history from the ERP, asset age and specification data from GIS and the asset master, and sometimes AMI-derived load data to infer thermal stress on distribution transformers. Machine learning models then score each asset by estimated failure probability within a time window.
This scoring is useful when it feeds directly into the maintenance planning workflow in the ERP. In SAP IS-U or S/4HANA Utilities, that means creating maintenance notifications or planned orders in the PM module, prioritized by the AI score. In Oracle CC&B environments with integrated EAM, it means populating work order queues. Without that ERP integration, the AI output sits in a separate dashboard and does not change how maintenance is actually scheduled.
The cases where this works well: distribution transformers with sensor-equipped bushings, underground cable circuits with partial discharge monitoring, and gas mains in systems that have already deployed acoustic leak sensors. The cases where it works poorly: assets with sparse maintenance history, aging infrastructure where the ERP records were never consistently populated, and asset classes where sensor deployment is limited.
AR for Field Asset Inspection
AR field tools for asset inspection have a different value proposition from predictive maintenance: they help the technician who is already at an asset do the job more accurately and capture better data back to the ERP.
The core workflow: a technician arrives at an asset, scans a barcode or QR code (or, in more sophisticated systems, uses visual recognition), and the AR application surfaces the asset record. This includes installation date, nameplate specs, last maintenance date, any open work orders, and prior inspection notes. The technician works through a structured checklist tied to the asset class. Findings, measurements, photos, condition ratings, write back to the ERP through the same API connection.
This matters for data quality. Inspection findings captured in a structured AR workflow are more consistently recorded and more reliably tied to the correct asset record than findings entered from memory after returning to the truck or the office.
For the specific case of SAP IS-U and S/4HANA Utilities, how AR platforms integrate through SAP APIs and what the PM module data model looks like, see AR integration with SAP utilities.
Remote Expert Assist
A scenario that has proven genuinely useful in utility asset management is remote expert assist via AR. A field technician encountering an unfamiliar fault condition or a non-standard equipment configuration connects to a specialist via AR video. The specialist sees the technician’s camera feed and can annotate the physical environment, highlighting a specific terminal, drawing a connection path, flagging a clearance concern.
This reduces specialist dispatch for non-routine situations without removing expert knowledge from the equation. It also creates a video record of the interaction that can be attached to the work order. Enel Group’s deployment of AR for field maintenance is a publicized example of this pattern at utility scale.
Drones in the Inspection Stack
Drone inspection is a distinct but complementary technology to AR field tools. Drones with thermal cameras and visual inspection payloads are used for:
- Transmission line patrol and hotspot identification
- Substation yard surveys for thermal anomalies on high-voltage equipment
- Solar array inspection
- Pipeline right-of-way surveillance for encroachment and vegetation
The AI analysis of drone imagery, identifying thermal hotspots, classifying conductor condition, flagging vegetation proximity, is maturing and reducing the manual review burden. The findings need to flow to the ERP asset record and work order system, same as AR-captured inspection data.
Drones do not replace technician-level AR for equipment that requires close-in inspection and hands-on assessment. The two tools address different points in the inspection workflow.
The ERP Dependency
Every AI and AR tool in this stack depends on ERP data quality to deliver its claimed value. Predictive maintenance needs accurate maintenance history and asset specifications. AR field tools need a populated asset master with correct equipment records. Remote assist records need work orders to attach to.
Utilities that get value from these tools have typically invested in ERP data quality improvement as a prerequisite, cleaning up asset records in SAP PM or Oracle EAM, standardizing equipment classifications in GIS, and ensuring maintenance records are captured consistently. The technology layers amplify whatever data foundation exists beneath them.
For the broader AI-in-operations picture, see AI for utility operations.