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.

For a city (Town Hall pictured) with around 100 employees serving 30,000 residents, the incident illustrated how AI could help staff handle unexpected enquiries. The technology has since been deployed on a much larger volume of routine work.

“Usually whenever you get a question like that, you don’t really know where to even start to look, and it’ll probably take a longer time to do so,” says Kim Bui, Demand Generation Specialist at Polimorphic, the company behind the assistant.

Suisun City has extended the deployment into its finance department, where employees previously spent considerable time redirecting residents to an external utility-billing provider. The deployment has also given the city information about the questions residents ask and where its communications need improving.

Starting with a familiar problem

Suisun City’s challenge was the time employees spent helping residents find information. Questions about bills and municipal services repeatedly reached a small workforce managing competing responsibilities.

The city introduced its main AI telephone line in September 2025. Susie draws on information published by the municipality, allowing residents to obtain answers without waiting for an employee.

According to Bui, the initial deployment took approximately an hour to configure, without requiring extensive changes to existing systems.

Early results attracted interest from the finance department, which was receiving large numbers of utility-billing calls. Payments were handled by a third-party provider, leaving employees frequently explaining procedures and transferring callers.

“They told us these pain points of ‘we take so many calls, but once we get them, it’s just a switch, a transfer at the end of it’,” Bui says.

Polimorphic worked with finance employees to understand the enquiries they received and how calls were handled. A dedicated hotline followed in December 2025.

Getting the groundwork right

The initial deployment relied on information already available to the public rather than requiring the city to prepare extensive internal datasets.

Susie uses material from the city’s website and police website. It does not access internal databases or individual utility accounts, limiting the technical work required before launch.

“Setup usually doesn’t take that long, especially for our chatbot, because it’s already operating on the content that you already have,” Bui says.

The assistant is not currently connected to the city’s utility-billing software, supplied by Tyler Technologies. It therefore provides general billing guidance and directs residents to the billing provider rather than accessing individual account details. Bui says integration is planned but is not currently a priority.

Suisun City also had no formal AI policy before deployment and has not introduced one since. Instead, the company says security and privacy are addressed through safeguards built into the platform, restrictions on the information the assistant can access and analytics that allow interactions to be reviewed.

This approach has supported a relatively straightforward initial deployment, although applications involving individual records or transactions would introduce additional data access and governance considerations.

Measuring what changed

The finance department receives approximately 1,500 calls each month, with the assistant resolving around 70 percent without staff intervention.

Polimorphic counts a call as resolved when the resident receives the necessary information or direction without being transferred to an employee. This can include directing someone to the appropriate billing provider rather than completing a payment.

“If it doesn’t end in a transfer to a live person, then that’s how we know that the AI was successful at doing that,” Bui explains.

Initial figures suggested the service was saving approximately 40 staff hours a month, with an estimated annual value of US$25,000 in recovered staff capacity.

Updated figures supplied by Bui showed that the finance line had handled 11,659 calls since launch, saving an estimated 415.3 staff hours, equivalent to more than 50 eight-hour working days.

These figures represent estimated staff time saved, rather than confirmed reductions in expenditure. They also measure calls handled without staff intervention, not necessarily the completion of the underlying transaction.

Beyond the numbers

For employees, the reduction in telephone interruptions has implications beyond the estimated hours saved. Repeatedly stopping work to answer familiar questions can make it difficult to concentrate on other responsibilities.

The system records interactions and provides analytics showing what residents are asking, how calls are handled and whether enquiries are resolved.

Polimorphic’s customer success team reviews call logs each month with Michael Elm from the City Manager’s Office, identifying opportunities to improve responses and the information available to residents.

These reviews have highlighted gaps in the city’s published information. Repeated enquiries about a particular service can indicate that a webpage is incomplete or unclear.

“’Oh, okay, we need to update this page on the site because it doesn’t have the relevant information’,” Bui says, illustrating how staff can use the analytics.

The findings can inform updates to the website, potentially reducing confusion and the need for residents to contact the city in the first place.

Monthly reviews also allow Polimorphic and city employees to examine the assistant’s performance and identify recurring problems. The arrangement provides a continuing feedback process rather than treating deployment as a one-off exercise.

Building on the first deployment

Suisun City is considering extending AI into administrative workflows, including applications that currently require manual processing. The proposed next phase would move beyond answering questions to helping residents obtain responses more quickly.

Introducing workflows would require a different level of technical preparation. Unlike the existing telephone service, processing applications could involve accessing internal records, connecting to municipal systems and establishing controls over how information is handled.

For now, the main outcomes are the volume of enquiries handled without employee involvement, the estimated staff time recovered and the information collected about residents’ needs.

The next stage will test whether that approach can be extended to more complicated processes.

“The voice line was our proving ground, and it was incredibly successful. It showed the city that this is a viable solution for an understaffed, under-resourced city,” says Michael Elm from Suisun City’s City Manager’s Office. “We’re not using AI to replace jobs. We’re using it to supplement our team and bring residents a level of service we couldn’t provide before. The next phase is workflows, which will help residents get responses on applications faster and cut down processing time for staff.”

Main image: Kiddo27 / Wikimedia Commons / CC BY-SA 4.0