Nicknamed “the Hawkeye State,” Iowa is a midwestern state with a population of over 3 million residents. The State of Iowa is run by a series of agencies and departments that are all connected to the state website, Iowa.gov. This central site — along with the individual agency sites — are charged with providing residents with an enhanced yet intuitive digital experience.
Challenge
“We had several hurdles to overcome,” said Dawn Connet, eGovernment Services Coordinator for the State of Iowa. “One challenge is that while we have the central Iowa.gov site, all the agencies have their own websites. As a visitor, you don’t want to have to figure out which department you need to visit. And our old-school chat option on Iowa.gov was with state librarians and only allowed one chat at a time. Additionally, the organization of each agency’s site was different, so it provided an inconsistent experience — making information hard to find. And many sites had lengthy PDFs that visitors had to search to find answers.”
To provide a more comprehensive and less cumbersome experience for residents and visitors, the Office of the Chief Information Officer decided to update the central website self-service features. It chose a platform that could extend out to the individual agency sites over time to provide that consistent experience.
“We opted to take the leap to an AI-based environment,” added Connet. “We needed a platform that could work with our legacy systems and numerous stakeholders. When we coupled this with our live service the state reference librarians provide, we felt residents would have an enhanced experience.”
Solution
The State of Iowa selected Genesys DX™ to implement a live chat and an artificial intelligence (AI)-enabled chatbot. It began the implementation with the central Iowa.gov site, along with several agencies, including insurance and transportation. In building the knowledge base for Iowa.gov, agencies were asked to provide Connet’s team with the top five questions and answers to start. Following the launch, the team monitored new questions and phrases to continue building out the database.
“As an example, residents were asking the Department of Corrections what they could wear while visiting an inmate or what they could bring,” said Connet. “The information couldn’t be found via a search on the agency website as we were using the word ‘offender’ and not ‘inmate.’ This would then cause people to call the department for information. Once we understood these differences, we were able to adjust so the chatbot’s natural language processing would understand and serve up the right answer — even suggest other relevant information.”