The Bennett School of Public Policy is a new, flagship department within the University of Cambridge – in fact, it’s the first new department to be set up within the 800-year-old University in three decades. And it’s committed to doing things differently.
“We’re really taking a fresh look at how we operate,” explains Department Administrator Sarah Rosella. “Rather than doing things because they’ve always been done that way, we’re trying to think outside the box and review all of our systems and processes so they can be as streamlined as possible.”
That mindset made the Bennett School a natural fit for ai@cam’s AI for Ops prgoramme – a University-wide initiative supporting colleagues to use AI to tackle real operational challenges. Sarah and her team successfully secured support to explore how AI could help manage recruitment enquiries more efficiently.
The need for the project became clear as the new department expanded. With around 12 roles advertised each year, the Bennett School has been running an intensive recruitment programme while also building many of its systems and processes from scratch.
“The team was experiencing very high workloads, and the pressure was beginning to mount,” Sarah says. “The real crunch point came in spring 2025, when we were advertising multiple posts at the same time.”
Across four assistant professor vacancies alone, the team received more than 100 queries from prospective applicants. “At the height of the recruitment campaign, I had one member of the team focused almost entirely on answering and managing those queries,” Sarah says.
Most of the questions were relatively straightforward, covering issues such as where to upload a CV, hybrid-working arrangements or information already available in University HR policies. Responding to each enquiry individually, however, was time-consuming.
“We started recording the queries so we could identify patterns and think about how to address them more systematically,” Sarah says. “When the AI for Ops call for proposals opened, we saw an opportunity to tackle the problem.”
“Only a small number of the queries required a detailed or bespoke response – most were routine,” she explains. “We began to think that a chatbot could handle many of them so the team could spend more time on the questions and tasks that genuinely needed a human response.”
Since securing support through AI for Ops, Sarah and her team have been working with programme mentors and computer scientists to develop an AI-powered chatbot and automated email reply tool. The system is being designed to provide consistent answers to frequently asked questions about job opportunities at any time of day, regardless of applicants’ time zones.
The 24/7 chatbot and replybot are being designed to integrate within existing University systems so that it can be rolled out without requiring major system changes. Before wider implementation, the tools will be piloted during a single recruitment round to ensure they work as intended.
To prepare for this robust testing process, Sarah’s team has been busy compiling a spreadsheet of frequently asked questions and responses to ensure the chatbot is supplied with accurate, up-to-date information.
“Our goal is to make the process as straightforward as possible for applicants,” Sarah says. “If the chatbot can answer the routine questions quickly and consistently, it means the team can spend more time supporting candidates with other enquiries.”
Sarah hopes the new system will save up to ten hours of administrative time each week, freeing up capacity for less routine work such as coordinating interviews, supporting candidates and welcoming new recruits.
Laying the groundwork for the future
For Sarah, the combination of funding and expert support provided through the AI for Ops programme has been instrumental. “The programme has enabled us to get the project off the ground, but the mentoring has been just as valuable,” she says. “Because we’re scaling up so quickly as a new department, having that expertise to draw on has given us the confidence to move the project forward.”
The project has already attracted interest from colleagues in other parts of the University, with several departments approaching Sarah’s team to learn more. “We’ve had others get in touch who are really interested to find out how we are designing the tool, so we’re going to stay connected and share how it works for us,” she says. “Once the infrastructure is in place, other departments could deploy their own versions with minor adjustments, saving significant time and resources compared to individuals developing tools from scratch.”
Looking ahead, Sarah sees potential for her department beyond HR recruitment and is hoping that the chatbot can be rolled out to respond to student recruitment queries in future too.
“More broadly, AI could help us with streamlining other routine processes” she says. “As we build our confidence, we hope to look at other systems too – for example minute-taking – and explore what else may be possible.”
“As a new department we want to set best practice from the beginning,” says Sarah. “This has been an incredible opportunity to put the right foundations in place rather than relying on legacy ways of working. We’re excited to see where this goes next.”