Across the United States, a quiet shift is underway in the way local governments answer the phone. Non-emergency calls that once rang straight through to human dispatchers are increasingly being picked up first by artificial intelligence, which promises faster responses for routine problems and more time for people to handle real crises. The experiment is unfolding in counties from Idaho to Florida, and it is beginning to redefine what it feels like to dial for help when the situation is urgent, but not life or death.
I see a pattern emerging: understaffed emergency communication centers are turning to AI not as a novelty, but as a survival tool to keep up with demand, protect overworked staff, and keep the human focus on true emergencies. The question is no longer whether one county will let AI answer your non-emergency calls, but how many will follow and how they will balance efficiency with trust.
Why 911 centers are turning to AI for non-emergencies
Public safety call centers are under strain, and the pressure is not limited to big coastal cities. In places like Bonneville County and fast-growing regions of Brazos County, the same teams that answer life-threatening 911 calls are also fielding reports of loud parties, parking complaints, and questions about when to file a police report. Dispatch leaders argue that if AI can reliably handle those lower-stakes calls, human call takers can stay focused on the people who dial in because someone is hurt, missing, or in danger. That is the core logic behind the new systems: triage the phone lines so emergencies do not wait behind noise complaints.
At the same time, local officials are confronting a staffing problem that technology vendors are eager to solve. Emergency communication jobs are stressful, turnover is high, and training new dispatchers takes months. In Dakota County and other Midwestern hubs like Grand Traverse County, leaders describe AI as a way to fill gaps without lowering service standards. The pitch is straightforward: let software handle the repetitive intake questions, route callers to the right place, and free human dispatchers to do the complex, emotionally demanding work that machines cannot.
Bonneville County’s “AI Annie” and the Idaho test bed
In eastern Idaho, the Emergency Communications Center that serves Idaho Falls and Bonneville County has become one of the clearest examples of this shift. The center launched an AI assistant nicknamed “AI Annie” to answer non-emergency calls and help screen which issues truly require a dispatcher. Officials there emphasize that All AI-handled calls are recorded and generate logs, transcripts, and summaries that can be checked later, a design meant to make the system easier to audit and to verify that callers are getting consistent, clear, and accurate responses. That level of documentation is something many human-only centers struggle to match in real time.
Crucially, the Idaho Falls / Bonneville County ECC is not separating its workforce into “AI calls” and “real calls.” In Bonneville County, 911 calls and calls to the non-emergency dispatch number are all answered by the same team of Emergency Comm professionals, and the AI is layered in as a front-end helper rather than a replacement. When the system detects that a situation is urgent or a caller is struggling, it can hand the call back to a human dispatcher, preserving the safety net that residents expect when they dial for help.
From Texas to Florida, AI picks up the non-emergency line
Far from Idaho, The Brazos County 911 District is running a similar experiment in the heart of Texas. The Brazos County 911 center introduces AI to handle non-emergency calls, with local leaders arguing that the technology can improve service quality for emergency callers by keeping human call takers focused on the most serious situations. The system is designed to answer routine questions, collect basic information, and route people to the right agency, while still allowing callers to reach a person if they need to. In practice, that means someone reporting a minor fender bender or asking about a past incident might talk to software first, while a caller describing an active crash or medical issue is quickly escalated.
On the Atlantic coast, Inside Volusia County’s 911 communications center, a new voice is answering the phones and it is not a dispatcher. In VOLUSIA COUNTY, Fla, officials have deployed an AI system to handle non-emergency calls, repeatedly stressing that it “isn’t here to take your job” but to cut down on hold times and repetitive questioning. The AI screens calls that come in on the same infrastructure as 911, then passes more complex or urgent situations through to live dispatchers. That setup reflects a broader trend: in Volusia County and elsewhere, AI is being framed as a co-worker that takes the first pass at routine calls, not a wall between residents and human help.
Midwestern counties build a regional AI playbook
In the Upper Midwest, several counties are moving in parallel, creating an informal playbook for how AI can fit into regional emergency communications. In Summit County, the Summit County Police and Fire Departments decided to use AI to answer non-emergency calls 24/7, effectively giving residents a constant front door to local services even when human staffing is thin overnight. When someone calls, the AI captures the details and displays them on a dispatcher’s computer, so by the time a person steps in, they already have a structured summary of what is happening. That approach treats AI as a data-entry specialist that never sleeps.
Northwest of there, Two Wisconsin counties are now using artificial intelligence to handle non-emergency calls, and their early numbers are drawing attention. In Waukesha County and neighboring La Crosse County, officials report that the AI system has already handled over 40,000 calls so far, a volume that would otherwise have landed on human dispatchers’ desks. That scale is one reason other Wisconsin jurisdictions are paying attention, and why a separate report notes that Wisconsin agencies see AI as a practical response to chronic understaffing.
Dakota County and Dane County weigh the tradeoffs
When someone calls the non-emergency line in Dakota County, instead of a dispatcher answering, an AI assistant now does. Local leaders describe the system as a way to shave down wait times and standardize the intake process, while still giving callers a path to a human if they are uncomfortable or if the situation escalates. One official, identified in coverage as Hieserich, acknowledged that the technology might “creep people out a little bit,” but argued that the payoff is a more efficient pipeline for non-emergency reports that would otherwise tie up staff. The Minnesota rollout shows how counties are trying to be candid about the cultural adjustment that comes with talking to a machine about a stressful situation.
Elsewhere in the region, Dane County is not yet all-in, but it is moving in that direction. The Dane County 911 center is considering an AI system to answer non-emergency calls, with internal discussions focused on how AI would eliminate that waiting on hold that frustrates callers today. Officials there have talked about the possibility of having a system in place by the end of this year, but they are also weighing how to introduce it without undermining public trust. The debate in Dane County captures a broader tension: residents want shorter waits and better service, but they also want to know a person is there when it really matters.
Branding the bots: “AI Annie,” “Aurelian” and the human touch
One striking detail across these deployments is how often agencies give their AI systems human-style names, as if they were new hires rather than software. In Idaho Falls, the Emergency Communications Center talks about “AI Annie” as if she were a colleague, and in northern Michigan, Grand Traverse County adopted an AI system named Aurelian to assist with non-emergency calls. During a county commission meeting in TRAVERSE CITY, Mich, officials described how Aurelian would help staff manage call volume under a three-year agreement, with options for the subsequent two years, signaling that they see this as a long-term partnership rather than a short pilot. That kind of branding is not accidental; it is meant to make the technology feel approachable to residents who might otherwise hang up at the sound of a synthetic voice.
In my view, these names also reveal how agencies are trying to thread the needle between automation and accountability. When Grand Traverse County talks about Grand Traverse County’s Aurelian or Idaho officials describe All AI-handled calls being logged for later review, they are implicitly promising that someone is responsible for what the system says and does. The more these tools are treated like members of the team, the easier it may be for dispatchers and the public to hold the agencies, and by extension the vendors, accountable when something goes wrong.
Phoenix and big-city experiments in AI call triage
Large cities are also testing how far AI can go in sorting out non-emergency calls before they reach a human. In the Southwest, Phoenix is utilizing the CallTriage system from a company called Versaterm, which markets its AI-powered service as a solution for helping understaffed emergency call centers. Phoenix is using the system on a dedicated non-emergency line, where callers interact with AI that can collect details, categorize the issue, and decide whether to pass the call to a dispatcher or direct it elsewhere. For a sprawling metro area, that kind of automated sorting can be the difference between a dispatcher spending several minutes on a low-level complaint and being free when a more serious call comes in.
The Phoenix Police say, depending on the situation, the system can transfer callers to Silent Witness, Safe School Tips, the Office of the local Humane Society, or other City and community resources, instead of routing everything through a central dispatcher. That flexibility lets the AI act as a switchboard operator that knows the difference between a tip about a crime, a concern about a student, and a stray animal. It also shows how AI can be woven into a broader ecosystem of services, not just police and fire. For residents of Phoenix, the experience of calling for help may now involve a machine that can send them to Silent Witness or Safe School Tips faster than a human operator flipping through a directory.
Waukesha, La Crosse and the Wisconsin wave
Back in Wisconsin, the scale of adoption is turning the state into a bellwether for AI in public safety communications. The Waukesha County dispatch center announced that the next time you call the non-emergency line, artificial intelligence could be picking up your call, with local coverage noting that CBS 58 highlighted how the system will answer routine questions and free up staff. Leaders in Waukesha County frame the move as a way to modernize without sacrificing the option to talk to a person, promising that callers can still request a human dispatcher if they prefer.
Those assurances matter because the technology is spreading beyond a single county. Two Wisconsin counties are now using artificial intelligence to handle non-emergency calls, and officials say the systems have already processed over 40,000 calls so far, a figure that underscores how quickly AI can become embedded in daily operations. In Waukesha County and Two Wisconsin jurisdictions including La Crosse, the early data is being watched closely by neighboring areas that are considering similar systems. If the numbers continue to show shorter waits and fewer abandoned calls, it will be harder for other counties to argue that AI is too risky to try.
Summit County, Lyon County and the 24/7 virtual receptionist
Some counties are leaning into AI as a way to guarantee that someone, or something, always picks up. In Ohio, Summit County Police and Fire Departments decided to use AI to answer non-emergency calls 24/7, effectively turning the system into a virtual receptionist that never goes off duty. When a call comes in, the AI gathers key details and displays them on a dispatcher’s computer, so by the time a human steps in, they have a concise summary instead of a blank screen. For a place like Summit County, which includes urban and suburban communities, that can mean faster follow-up on everything from noise complaints to minor crashes.
In the Great Plains, a Kansas county is taking a similar approach. A report on how a Kansas county employs an AI virtual assistant to answer non-emergency calls describes how the system is being used in Lyon County, where the director of Lyon County 911, quoted in coverage by Aniyah Robinson, explained that the AI helps staff manage call volume. The story notes that the deployment is part of a broader trend in Kansas, where rural and semi-rural counties like Lyon County are looking for ways to stretch limited staff across large geographic areas. For residents, the experience may feel similar to calling a bank or airline, but with the stakes tied to public safety rather than travel plans.
How residents experience the shift from dispatcher to AI
For callers, the most immediate change is the voice on the other end of the line. In VOLUSIA COUNTY, Fla, people who dial the non-emergency number now hear an AI system that introduces itself and starts asking questions, before passing more complex issues through to live dispatchers. Inside Volusia County’s 911 communications center, staff say the goal is to reduce the time callers spend waiting on hold and to make sure that when they do reach a person, that dispatcher already has the basic facts. That mirrors the experience in places like Volusia County, where the AI is framed as a way to get people to the right help faster, not as a barrier.
In my conversations with dispatch leaders and based on the reporting from counties like Lyon County and Summit County, the biggest hurdle is not the technology itself but the perception that automation means indifference. Agencies are trying to counter that by emphasizing features like the ability to request a human at any time, the fact that All AI-handled calls are recorded and auditable, and the way AI summaries can actually help dispatchers respond more personally once they pick up. Whether that message lands will determine how quickly residents accept a future in which the first “hello” they hear when they call for help comes from a machine.
The next phase: from pilots to permanent infrastructure
What started as a handful of pilots is rapidly hardening into infrastructure. In Idaho Falls, the Emergency Communications Center’s decision to log every AI interaction and generate transcripts shows a commitment to long-term oversight, not a short-lived experiment. In All AI deployments like this, the ability to audit conversations after the fact is becoming a baseline expectation, both for quality control and for legal accountability. That is a sign that counties see AI as part of their core operations, subject to the same scrutiny as any other part of the emergency response chain.
At the same time, cities like Phoenix are betting on commercial platforms like Versaterm’s CallTriage, while counties from Brazos County 911 to In Bonneville County are integrating AI into existing centers that already juggle 911 and non-emergency calls. As more places follow Waukesha, La Crosse, Summit, Dakota, and Lyon into AI-assisted call handling, the novelty will fade and the focus will shift to performance: Are calls answered faster, are emergencies prioritized better, and do residents feel heard? Those are the metrics that will decide whether AI remains a front-line voice in public safety or retreats back into the lab.
Supporting sources: Local county now using AI 24/7 to answer non-emergency calls, Phoenix PD using AI to filter non-emergency calls; here’s how it works.
More from MorningOverview
