CIO Christiaan Polet explains the advantages

UZ Gent is one of the largest hospitals in Flanders, employing 7,000 staff members and 1,200 physicians, and handling more than 3,000 consultations per day. As a university hospital, it not only provides care but also plays an essential role in education and research. To optimally support all these activities, the hospital uses roughly 800 IT applications, ranging from organisation-wide administrative systems such as ERP to highly specialised healthcare applications.

Almost all physicians, care staff and other employees interact with IT in their daily work. When they have questions, they can contact the helpdesk teams for IT infrastructure, healthcare applications and business applications. Over the years, however, the number of questions steadily increased to around five thousand per month, including approximately 1,500 by phone. This resulted in ever-longer waiting times. UZ Gent therefore sought a solution with a dual objective: reducing the pressure on the helpdesk teams while resolving tickets more quickly and thus increasing employee satisfaction.

A strong case

“As a university hospital, we attach great importance to innovation,” says CIO Christiaan Polet. “We have long been familiar with AI, including in medical imaging.” The advent of generative AI opened the door to new applications. “We immediately recognised the potential of GenAI as a supporting tool, but we also know that many projects fail. Only with a strong case can you be successful.” Through its innovation funnel, which the hospital uses to proactively identify opportunities for innovation, UZ Gent found such a case in a project addressing the overburdened helpdesk.

UZ Gent engaged Inetum as its partner. “We have worked together long enough to know that Inetum has the necessary expertise in GenAI,” says Christiaan Polet. Inetum built a tailored chatbot using the components available within Microsoft Azure AI Foundry. The chatbot answers employees’ questions based on the information available on the hospital’s intranet. “It concerns roughly 500 pages,” says Bram van Groenigen, head of ICT infrastructure at UZ Gent. Employees can consult this information directly on the intranet as well. “But in practice, they usually reached for the phone straight away. Now they ask the chatbot, which immediately provides an answer and shows where the requested source information is located.”

Faster answers

The impact of the chatbot became clear shortly after its introduction. The chatbot resolves most questions immediately. For at most 5 to 8 percent of issues, the user is still directed from the chatbot to a helpdesk employee. “Password reset requests are a classic example,” says Bram van Groenigen. “These are not the most interesting questions for employees to handle. A chatbot resolves such queries very easily.” The advantage for the user is that they bypass the waiting line. They also receive an immediate answer outside the helpdesk’s standard opening hours.

By handling the most common and simple questions, the chatbot has reduced the high workload at the helpdesk. “This allows the team to focus on more challenging questions,” says Bram van Groenigen, “which also contributes to the team’s motivation and job satisfaction.” At the same time, this work will make the chatbot increasingly effective over time. “The tickets that employees resolve manually can be added to our information source. As a result, the chatbot will be able to handle even more questions.”

New domains

“We deliberately kept the project very simple,” says Christiaan Polet. “The chatbot only handles IT-related questions and refers exclusively to information found on our own intranet. This keeps the solution secure. We avoid hallucinations.” At the same time, the hospital sees potential for using the chatbot in other domains. “We have other helpdesks as well,” continues Christiaan Polet. “For example, HR and facilities.”

A chatbot could potentially provide faster answers and direct employees to the right information there as well. Once such a decision is made, the implementation process does not need to take long. “Inetum built the chatbot for the IT helpdesk in six weeks,” says Christiaan Polet. “If you want to innovate with new technology, you have to move quickly. An experienced partner makes all the difference. If you try to do everything yourself, you risk getting stuck in the experimental phase.”