Sullivan County: Using virtual agents to serve more citizens faster, at any hour of the day

The county offices for Sullivan County, New York, support more than 40 departments, ranging from the County Clerk's Office to Emergency Services and Parks and Recreation. Their busiest offices were handling hundreds of phone calls and emails every day about the same basic questions: "What are your hours?," "Where can I pay my taxes?," and "How do I renew my driver's license?" Serving more than 78,000 citizens with limited staff and budget made it difficult to keep pace with call volume demands and maintain responsiveness to citizens.

The county offices wanted to provide citizens with a higher level of service and reduce the growing workload on their staff, but didn't have the resources to expand their team or stand up costly call centers. Instead, the county focused on how they could answer simple questions quickly and reduce the amount of calls coming into the county so their staff could devote their time to help citizens with more complex questions that required their attention.

Sullivan County's offices turned to Google Cloud's tool Dialogflow to introduce the Sullivan County virtual agent, which immediately helps constituents find the information they seek. Egen teamed with Sullivan County and Google Cloud to bring the deep AI, data, and technology expertise required to develop and launch the solution. They embedded the agent into the chatbot on the county website to support those seeking information about our County Clerk’s and Treasurer’s offices.

Although more than 40 offices in Sullivan County were excited to implement a chatbot, the team was initially concerned about the time and effort it would take to tackle a project of that scale. However, they discovered Vertex AI Agent Builder, a tool that simplifies chatbot creation and allows for the training of machine learning models. By scraping information from office websites, the chatbots dynamically determine answers to questions without manual intervention. Partnering with Quantiphi, they successfully implemented a generative AI-powered virtual agent named Saige, connecting every department in just ten weeks.

The Sullivan County virtual agent is an artificial intelligence-driven program designed to immediately answer common questions on the County website. When a person clicks on the chatbot logo in the lower right of their screen, the virtual agent provides a list of topics and a spot to type in questions.

"We were particularly impressed with how they, as large tech companies, strove to understand our needs, concerns and limitations as a small municipality with a limited budget and staff, yet a large geographical region to cover," says Dan Hust, Sullivan County Director of Communications.

The Sullivan County chatbot is also helping the county DMV to continue serving customers outside of normal operating hours, such as nights and weekends, so they can have 24/7 access to the information they need. "If it's midnight on Saturday, and someone is figuring out when they can go to the DMV on Monday, now they don't have to wait until we're open on Monday morning to find an answer," says Potosek.

And as people access resources via the chatbot, call volumes decline. "The DMV experienced a more than 56% reduction in call volume in the last four months compared to the same time last year. It blew past our expectations," adds Hust.

As Saige gathered information from various department sites, it helped identify sources that needed to be updated. Saige is able to search for and find new information immediately. This is especially valuable because information changes; holidays change operating hours or a file is updated, and Sullivan County’s small teams don’t have time to manually update every chatbot workflow.

As Saige’s knowledge repository grows, offices continue to get more and more time back to manage more complex issues that require the face-to-face, human touch. Offices get more efficient, people get answers faster, and every interaction we have with a constituent helps inform the next.