Resident Experience

When the Lights Go Out, the Phones Lines Flood: How AI Can Help

clock-icon
Read Time:
5 minutes
calendar-icon
Published:
October 20, 2026

Across Southern California, residents are checking outage maps and asking the same question,  “Will my power go out, and for how long?”

260,000 SoCal locals are on the cusp of a Public Safety Power Shutoff, or PSPS. (And, no, that’s not just a way to call over a cat.) A PSPS is when an electric company temporarily shuts off power to reduce the risk of causing a wildfire. High winds and dry conditions make downed or sparking lines an ignition risk, so utilities cut power to specific areas before the wind arrives. It’s an important and effective safety measure, but it can also be disruptive for the people in those areas as PSPS events can last over 24 hours. 

While the electric companies do a lot of outreach—customers are typically notified by text, email, and automated calls ahead of a shutoff—a notification is only a starting point. The moment people get notified, they have questions, and they take those questions to the number they trust—their city or county’s non-emergency line.

‍

The 311 Call Surge

Picture a 311 center right after a utility company issues a shutoff warning. Thousands of residents are calling at once, and almost all of them are asking the same handful of questions:

  • Is my neighborhood affected?
  • When is the power coming back?
  • Where can I charge my phone or get water and ice?
  • What do I do about my medical equipment, my refrigerator, my elderly parents?
  • Who do I call if I see a downed line?

The call center doesn’t have time to staff accordingly, causing hold times to climb. Callers who have been waiting twenty minutes to hear information (that, by the way, is already posted online) get frustrated. The agents spend their shifts reciting the same answer hundreds of times, unable to differentiate between callers with real emergencies and those looking for the same answers as the previous callers.

It’s nearly impossible for cities or counties to staff for the peak of a once-in-a-season event, but the cost of falling short is real. In cases like this, everyone loses. You end up with unhappy residents, overwhelmed staff, and a loss of trust at the moment the community most needs to rely on local government.

‍

More Than a Wildfire Problem

Every unpredictable event creates the same dynamic for local government—a sudden spike in demand, an onslaught of near-identical questions, and a fixed number of people to answer them.

‍

Hurricanes

Before landfall, residents want to know evacuation zones, shelter locations, and sandbagging sites. Afterward, they want to know when power and water will return and how to file a damage report. 

‍

Snowstorms and Extreme Cold 

Residents call about plowing schedules, road closures, trash pickup delays, and warming centers. These are high-volume, location-dependent questions that suit automation well. A targeted message about a street-by-street plowing schedule can head off hundreds of calls.

‍

Heat Waves

As temperatures climb, residents need cooling center hours and locations, and vulnerable residents need check-ins. Outbound messaging can reach the right neighborhoods, and human agents can follow up on the calls that signal real risk.

‍

Floods, Tornadoes, and Boil-Water Notices 

The details differ, but the need doesn’t. People want accurate, local information fast, and the people who need help want to reach a real person quickly.

In each case, the same three capabilities apply—answer the repeat calls, reach the right people proactively, and keep humans free for the true emergencies.

‍

Where Polimorphic Fits

Government service operations are fragmented across departments, systems, and channels. AI can bridge them without replacing the systems agencies already use. 

Polimorphic’s AI capabilities work alongside 311 centers’ human teams in three ways. 

1. Answer Repeat Calls Instantly, With Location-Specific Information
Most surge calls are repeats. Polimorphic can pick up every one of them immediately, with no hold time, and answer based on where the caller actually is. A resident in one neighborhood hears whether their area is under a shutoff warning, where the nearest community resource center is, and what the latest restoration information says. A resident in a neighborhood that isn’t affected can stop worrying. 

2. Send Mass Outbound Messages, Targeted by Neighborhood
The best way to reduce inbound calls is to answer the question before it’s asked. Polimorphic can send outbound messages to the people who need them. Everyone in a given neighborhood can get a message about the shutoff window, cooling centers, or an updated restoration estimate, while residents elsewhere aren’t flooded with irrelevant alerts. Targeting keeps the message relevant, and relevant messages get read. As a result, fewer people have to call 311 in the first place.

3. Hand Off to Humans When Needed
Automation should handle the routine and repetitive and leave critical judgment to people. A caller who relies on home medical equipment, or who is stranded or frightened, needs a human. Because the repeat calls are being handled, your agents are available for exactly those conversations. Polimorphic routes those callers to staff, along with the context already gathered, so the agent doesn’t start from scratch.

The result is a 311 center that stays responsive at the peak, where the human team’s time goes to the people who need it most.

Preparing Before the Next Disaster

The best time to set up is before you need it. A system that is already configured with your agency’s resources, neighborhood boundaries, and escalation rules can turn on the moment conditions change. If your team is trying to figure out the tools in the middle of a crisis, you are already behind.

Local governments can’t control the wind, the storm, or the heat, but you can control how well you respond to the people who call. Residents want an answer quickly, and when their situation is serious, they want a human. With the right tools behind your 311 team, both are possible.

If you’re thinking about how your city or county handles the next surge, we’d be glad to talk. Request a demo.

‍

Related Articles

October 20, 2026
Customer Stories

How a Californian City is Raising Productivity Through A Citizen Hotline

When a resident contacted Suisun City in California with an unusual question about salvage rights to a sunken vessel in the harbour, finding an answer would normally have involved an exhaustive search of municipal records alongside a time-consuming consultation with a range of colleagues. Instead, Susie, the city’s AI-powered telephone and web assistant, found the information in minutes.
3 Minutes
October 16, 2026

Governments Become More "Natural" With Search Tools

As natural language search gains traction in the public space, an associated technology is also finding more interest: the use of chatbots, including ones that offer responses in multiple languages. It’s not just what you say but how you say it that could become a guiding idea as state and local governments boost digital communications and services, often via artificial intelligence. Local governments such as Grand Rapids, Mich., and the borough of Prospect Park, N.J., are testing an increasingly popular form of search designed to connect residents and officials with ever more precise civic information. The idea, common in e-commerce, revolves around using natural language to unearth data often buried in legislation, meeting minutes, the deeper layers of municipal websites, or other digital hiding places. Natural language search, or NLS, means that people can go to a website and type in — or speak — a query that reflects how that person would ask that question in personal conversation. NLS, which takes into account the context of search terms, seeks to better understand the detailed intentions of users than the older methods of Boolean searches that rely on relatively static keywords. Here are examples of how NLS could work in local government. It comes from California-based Polimorphic, a three-year-old government technology provider that has put up its new AI-powered NLS search tool for testing in those two locations: “I’m opening a café to serve coffee by day and wine by night. What permits do I need?”‍ ‍“How do I make a public records request?”‍ ‍“I just moved to Grand Rapids. What do I need to do?” ‍The new tool goes “beyond just finding matches” to produce answers to such inquiries, Parth Shah, the company’s CEO, told Government Technology. The tool learns from the government’s website, using only sources from the site, puts it all together and provides answers, he said. He likened the process to transforming a digital maze into “something understandable.” That, in turn, can reduce 311 calls or angry complaints to city hall customer service workers — an attractive proposition for officials. “With better search results, residents need to make fewer calls to Borough Hall, freeing up our staff to dedicate time to resident issues that do require person-to-person interaction,” said Mayor Mohamed Khairullah of the borough of Prospect Park in a statement. “Meanwhile, residents' ability to quickly find essential information, regardless of language barriers, democratizes access to crucial information across our diverse community.” Officials, meanwhile, can use the natural language search and the tool’s AI to generate reports on a variety of issues, Shah said. Shah, who has a background in machine learning from MIT, said his grandfather’s experience as a utility director in India sparked his ongoing interest in workforce and customer service in government. Shah’s company has raised more than $3.7 million, according to Crunchbase, and was part of CivStart’s 2022 gov tech accelerator program, which just absorbed a fresh cohort. Natural language tools are gaining ground in government. One recent example comes from Illinois, where the state’s Department of Children and Family Services has started using natural language processing software from Augintel. The technology is meant to help agency employees gain a firmer grasp of data and narratives about cases. One analysis of the possible future of natural language tools in government foresees those tools encouraging more "one-stop shopping" in the public sector — that is, having a single portal through which residents can conduct such government business as tax payments and permit applications without having to navigate to or bookmark multiple pages. ‍
cta

Are you ready to become a Service-First Government?

Request Demo