Storm Mode vs. Blue Sky: The Timing Gap in Outage Intelligence

Utilities don’t struggle with outage visibility. They struggle with timing.

By the time an outage is officially confirmed in the OMS, customers often already know something is wrong. What they don’t know is what happened, how widespread it is, or whether their utility is aware of it.

That creates an important operational question: How long are customers sitting in uncertainty before the utility responds?

Consider a storm event at 2:07 a.m. A severe wind system moves through a utility’s service territory. Within minutes:

• 14 customers text outage reports from the same area
• 8 social posts mention flickering lights and power loss
• 1,200 AMI meters stop responding almost simultaneously
• Weather data confirms extreme wind gusts in the region

But there’s no OMS event yet. The outage map hasn’t changed, and customers haven’t heard from the utility.

From the system’s perspective, the outage isn’t confirmed. From the customer’s perspective, the power is already out.

That’s the timing gap.

Now consider the same problem during blue sky conditions. A transformer fails on a feeder. A vehicle strikes a pole. Vegetation causes a localized fault. There’s no major storm, surge in outages, or emergency response underway.

The scale is different, but the signals are similar. AMI anomalies start appearing. A cluster of customer calls comes in. Other digital signals begin to emerge. The OMS may not have correlated them into an outage event yet.

Whether it’s storm mode or blue sky, the first indication of an outage may not come from the system of record. Instead, it can emerge from a collection of smaller signals spread across different systems and channels.

Utilities don’t necessarily see an event first. They see the signals that point to one.

Now imagine an agentic AI working across both storm and blue sky conditions, continuously analyzing:

• AMI last-gasp and meter silence patterns
• Customer calls, SMS, and digital interactions
• SCADA and feeder-level alarms
• Weather and external risk data
• Historical outage patterns

Instead of leaving those signals isolated, AI could correlate them to provide an early view of what may be happening, including:

• Probable outage boundaries
• Estimated customer impact
• Likely cause
• An early estimated time of restoration (ETR) range
• A confidence score indicating how certain the assessment is

The goal isn’t to treat that early assessment as fact. It’s to make useful information available while the situation is still developing. The OMS remains the authoritative system of record. That doesn’t change. What can change is the sequence.

Today, the process still largely follows:

OMS confirmation → outage map → customer communication

But customers experience the outage before that process is complete. With better use of real-time signals and AI, the sequence could begin earlier:

Multi-signal detection → AI assessment → early communication → OMS validation

That may look like a small change in the flow of information, but it could make a meaningful difference to the customer experience.

Customers don’t care whether their utility is in “storm mode” or “blue sky mode.” They care that their power is out and they don’t know why, for how long, or whether anyone is working on it.

The opportunity, then, isn’t simply to build better outage maps. It’s to shorten the time between the first credible signs of an outage and meaningful communication with the customers affected by it—whether 100,000 customers lose power during a major storm or 100 lose power on an otherwise ordinary afternoon.

That raises a question utilities will need to address:

Should they wait for formal confirmation before communicating something they already have strong evidence is happening?

Or can they communicate earlier, with appropriate levels of confidence and transparency, while the full picture is still taking shape?