AI Can Close the Information Gap Between Outage Occurrence and Customer Awareness

For years, utility outage maps have followed a familiar pattern: an outage occurs, the OMS identifies it, and the outage map displays it. Simple. Logical.

But what if we’ve been thinking about outage communications all wrong? What if the outage map of the future isn’t just a reflection of what the utility already knows? What if it’s an intelligent prediction of what the utility is discovering in real time?

Imagine an agentic AI continuously analyzing information from:

• Customer phone calls
• SMS outage reports
• Web chats
• Mobile app submissions
• Social media posts
• AMI last-gasp messages
• Meter ping failures
• SCADA alarms
• Weather feeds
• Lightning strike data
• Historical outage patterns

Now,  consider this scenario.

At 2:07 AM:

• 14 customers send outage texts from the same neighborhood
• 8 social media posts mention a power outage
• 1,200 AMI meters stop responding
• Severe winds are moving through the area
• No OMS event has yet been created—because if you’re a small or medium-sized utility, there may be no operator actively monitoring the system.

Could AI identify the outage before the OMS? I believe the answer is yes.

An agentic AI could immediately correlate these signals, estimate the affected area, calculate the probable customer count, suggest a likely cause, generate an initial ETR, and publish a confidence-scored outage notification.

For example: “Potential outage detected affecting approximately 1,000–1,500 customers in the northeast service territory. Cause likely weather-related. Investigation underway. Confidence: 87%.” Hours later, utility staff validate the event and create the official OMS record.

The AI-generated information doesn’t replace the OMS. Nor should it. The OMS remains the authoritative operational system.

But there is room for a second layer—an AI-driven outage intelligence layer that helps utilities identify, communicate, and even anticipate outages before they are officially confirmed.

And this is where the conversation gets interesting.

The greatest value may not be the outage map itself. The greatest value may be getting credible information into the hands of customers as quickly as possible.

Today, utilities often wait until an outage has been confirmed before communicating. Meanwhile, customers are left wondering: “Is it just me?” “Has the utility noticed?” “Has anyone reported it?” “When will I know what’s happening?”

AI has the potential to dramatically reduce that uncertainty. Traditional outage maps are descriptive. They tell customers what the utility already knows. AI-enabled outage intelligence could make outage communications predictive. They could tell customers what the utility believes is happening before the investigation is complete.

The goal isn’t to replace operational systems. The goal is to reduce the gap between outage occurrence and customer awareness. In an era of AMI, advanced analytics, social media monitoring, and AI, should customers really be waiting for an OMS event before hearing from their utility? Or should utilities be communicating outage intelligence the moment credible evidence begins to emerge?

Is your utility preparing for a future where outage communications become predictive rather than descriptive? And if not, why not?