Skip to article

Around here

How Virginia Beach is using AI to help callers reach the right services

Virginia Beach built its AI phone service around the words callers actually use. A look at the trial, the practical workflows and what businesses can learn from it.

Listen · Along the CorridorListen · 9 min · AI-narrated

Transcript

How Virginia Beach uses AI to understand callers, route nonemergency requests and send useful links—and what local businesses can learn from its approach.

0:00 / 8:43

About 9 minutes · AI-generated voice · Read the transcript

On this page

A person calling about a lost dog knows what they need. They may not know which department handles it, what the service is called or which option on a phone menu will get them there. Studying the organization chart shouldn’t be the first step in asking for help.

That gap between a person’s question and the right service is a useful place to put AI to work. Virginia Beach has been doing that with Amazon Connect since April 2024, and the interesting part of its approach began before the system was ready to answer anyone.

The team first gathered the words callers actually used. It’s a modest detail in a story about artificial intelligence, but one I like very much: before deciding how the system should respond, they paid attention to what people were asking.

The work began with callers’ own words

An Amazon Web Services account of the project describes the underlying pressure: staff responsible for emergencies were also fielding routine inquiries through a phone system with limited menu choices. The aim was to get ordinary questions answered or directed appropriately while giving telecommunicators more room for urgent work.

The city’s September 2024 account explains the prelaunch trial. Callers stated their reason for calling before reaching staff. The team used those statements, together with historical records, to identify common requests and develop spoken responses and relevant website links. AWS describes that trial as lasting two weeks.

That preparation matters because an organization’s labels and a caller’s vocabulary don’t always match. A perfectly accurate department name can be useless to someone who doesn’t know what that department does. Starting with real requests gives the team a way to connect the language people bring with the services the city can provide.

It also gives the AI a specific job. Instead of asking a general-purpose assistant to sound helpful about everything, the team can connect familiar kinds of requests with useful answers and destinations. The substance comes from understanding the service, as well as understanding the sentence.

An answer, a connection or a useful link

AWS describes several parts working together: Amazon Connect manages the call flow; Amazon Lex recognizes speech and identifies what the caller is asking; Amazon Bedrock summarizes information from FAQ documents. Messaging services can then send information or links to the caller’s phone. Its examples include lost-pet inquiries and questions about eviction procedures.

There’s a useful distinction inside that sequence. AI helps interpret the request and formulate information from the supplied material. Moving the call or sending a text puts the result somewhere the caller can use it. A good answer is more valuable when it arrives with the next step attached.

A texted form can be a small mercy. It saves someone from remembering a long web address, spelling it back or searching through several pages after hanging up. The link doesn’t have to be technologically impressive. It has to be the right one.

The city’s current Amazon Connect FAQ also confirms that callers whose questions remain unanswered can choose to speak with a live agent. That leaves space for requests that need explanation or judgment, and for people who need help beyond an automated response.

A practical use during the July fireworks rush

The city’s original account gives a concrete example from around July 4, 2024. A special greeting option directed fireworks reports through emergency screening, then sent eligible callers a link to an online reporting form. The city reports hundreds of submissions and many hours of staff talk time saved.

What interests me is how closely the response fitted the task. These callers needed a way to report a particular problem. The city had a form for collecting it. Connecting those two things allowed the phone service to move the report along without making every caller describe it to a telecommunicator first.

That’s a historical example of how the city adapted the service, rather than a current instruction for reporting fireworks. It shows why useful automation often depends on the work surrounding the AI: a suitable destination for the request and a process for getting it there.

What the city reported after launch

In its first month, the city reported that 36% of incoming nonemergency-line calls were either completed by Amazon Connect or transferred to a more appropriate department. That combines automated handling and routing; it isn’t a claim that AI independently solved 36% of callers’ problems.

Both outcomes can be useful. Some callers need information the system can supply; others benefit from reaching the staff who can help them. The figure describes an early operational result, not a present-day satisfaction score or a measure of emergency response times.

The work has continued beyond that first month. Virginia Beach’s FY2026–27 adopted budget says the department still uses Connect for automated responses and to divert nonemergency call volume. That makes this an established local use of AI, with a clearer purpose than simply putting an AI voice on the phone.

A useful starting point for a business

For an owner considering an automated receptionist, the part of this project I’d borrow first is the listening. Look at the requests customers actually make before choosing what the assistant should handle. The words on your website may describe your services perfectly while missing the way someone asks for them on a busy afternoon.

Suppose a customer says they’re checking whether an order is ready. Recognizing an order-status question is useful, but it doesn’t tell the system whether that particular order is ready for collection. To answer, it needs access to the right order information. Without that access, directing the customer to someone who has it is the useful action.

That example helps separate a plausible-sounding reply from a service you can build. You can decide whether the assistant has enough information to answer, whether it can send the customer something useful, or whether it should connect them with the team. Virginia Beach’s combination of spoken answers, links and routing offers a concrete model for thinking those choices through.

I’m enthusiastic about this kind of AI use because the benefit is easy to recognize: someone asks in their own words and gets closer to the help they need. The preparation behind that experience deserves as much attention as the voice answering the phone.

If you need to contact Virginia Beach

The city’s September 2026 calling guide distinguishes 911 for emergencies requiring an immediate public-safety response; 311 or 757-385-3111 for general city information and service requests; and 757-385-5000 for nonemergency public-safety assistance. The short 311 number works within city limits; the full number also serves callers outside the city.

The guide includes an important exception: people outside Virginia Beach reporting an emergency inside the city should call 757-385-5000. Amazon Connect itself is described as a nonemergency service. Use the city’s current guidance for the situation you’re reporting.

Sources checked October 5, 2026. This account draws on published city and supplier records; it is not a test of the phone service.

Sources

Sources support the reported facts in this article. Practical recommendations are analysis by Alasdair for New Kent Digital Works.

  1. Virginia Beach: Award-Winning Non-Emergency Call Routing System Reduces Hold Times, Prioritizes Emergency Calls · Published · Accessed
  2. AWS: How the City of Virginia Beach modernized its emergency and citizen services with Amazon Connect · Published · Accessed
  3. Virginia Beach ECCS: Amazon Connect and frequently asked questions · Accessed
  4. Virginia Beach FY2026–27 adopted operating budget, printed page 135 · Accessed
  5. Virginia Beach: Make the Right Call · Published · Accessed