Outgoing Chancellor Jennifer Mnookin praised the University of Wisconsin-Madison’s artificial intelligence initiatives in her address to the Board of Regents last month, including one new tool researchers have been using to partner with industry partners and collaborators.
The Office of Business Engagement partnered with the Data Science Institute at the University of Wisconsin-Madison to release the Research and Business-Bridging Intelligence Tool (RABBIT) in January 2025, and since then, researchers say it has helped faculty and staff more easily find colleagues with relevant expertise across campus.
As federal funding cuts and the loss of grants continue to shape today’s research landscape, innovation has depended on connection, not just discovery. RABBIT, and the researchers behind it, hope to solve that. For years, connecting the right researcher with the right industry partner relied on tedious emails and web searches, and as a result, researchers frequently connected with partners who weren’t the best fit for their expertise.
At its core, RABBIT is an AI-powered faculty discovery platform that seeks to match researchers on campus with industry partners and collaborators more strategically than ever before.
Associate Director of the Office of Business Engagement Sara Braas has been the face of the project, though RABBIT was developed by a collaborative team across campus. She envisioned RABBIT as a tool to help connect industry partners with researchers.
“It is difficult to keep up on what all researchers on campus are working on and new faculty are being added under the RISE initiative, so we needed a tool that would help us identify faculty who were working on specific challenges that industry cares about,” she said in an email to The Daily Cardinal. “Those faculty that are already actively partnering with industry or have patents or grants in the specific research areas that companies are interested in are in high demand. We built RABBIT to meet that need.”
The search tool is fueled by research publications, as well as grant and industry partnership data. It also factors in patents and other information in its searches.
As UW-Madison continues to expand its research faculty and maintain its position as a national research leader — ranking fifth in the nation for research expenditures — it's increasingly difficult to track the evolving expertise across campus, Braas said. RABBIT's goal is to make the process faster and more efficient.
A key factor in that efficiency lies in the platform’s use of grant data. RABBIT analyzes grant data records of externally-funded research and identifies faculty members who have successfully secured funding in specific subject areas. Because grants are competitive and peer-reviewed, they signal credibility, active research engagement and proven demand in a particular field.
“Federal grants increasingly require applied science and industry partnerships,” Braas said. “They are also focused on grand challenges that require multidisciplinary solutions. RABBIT allows us to build internal, cross-campus teams that can jointly apply for these grants thereby strengthening UW-Madison's competitiveness.”
RABBIT also relies on a combination of research publications, industry partnership data and patents to form more advanced partnerships than simple keyword matching tools.
While much of RABBIT’s strength lies in the actual data used, its promptness and efficiency also depends on how the data is structured, in this case, around individual faculty members.
“We organized the data around each faculty member as the central profile,” Data Science Institute Developer Jason Lo said. “All of their related information, such as publications, grants and industry collaborations, was linked to that profile. This gave the system a complete view of each person in one place, making it easier to see connections across their work and understand how different activities relate to each other.”
Unlike traditional search engines such as Google and Yahoo, RABBIT uses semantic search technology that relies on the organization of the data to give the most accurate matches. Semantic search technology analyzes the meaning of the actual content, and the AI search system understands exactly what the user is searching for rather than focusing on keywords.
“A great feature of the AI-powered semantic search that RABBIT uses is that it helps you find people even if you aren’t familiar with the technical terminology or jargon that would be used in traditional keyword search,” Data Science Institute and RISE-AI Director Kyle Cranmer said. “You can even drag-and-drop a scientific paper that you might not understand and RABBIT will perform the search based on the concepts in that paper.”
RABBIT’s use of semantic AI makes it unique amongst its peers. When UW-Madison’s Office of Business Engagement was initially searching for a system to connect industry partners to campus research, they considered buying commercial software from outside companies, but they lacked some of the tools desired for RABBIT.
“While we explored some interesting software solutions in the market, in the end they were mostly able to tell us what companies were working on,” Braas said. “We were looking for a thorough ‘who on campus’ tool. In the end we hold that data internally, and it was a matter of building a system that could allow us to access it in a more convenient way.”
Before RABBIT, Braas said the process was far more tedious, manual and sometimes inaccurate.
“Before RABBIT, we used Google and lab websites and our personal networks. We'd call or email faculty and if they weren't the right person for the project, we'd sometimes ask them to recommend who else we needed to talk to,” she said. “We naturally over time develop relationships with the faculty we work with routinely, but there was always the chance we would miss talking to those we didn't know.”
The team knew in order to connect data already held, custom software was necessary to ensure as accurate results as possible. As technology improves, new methods of incorporating data sources and further refining semantic AI capabilities remain a goal for RABBIT developers.
“One improvement I am most excited about is adding an AI-driven query refinement feature,” Lo said. “We plan to introduce an interactive assistant that helps users clarify and structure their request through guided prompts. By eliciting intent more effectively, the tool can generate more precise and relevant matches.”
With continued innovation and updates, RABBIT hopes to not only strengthen collaboration at UW-Madison, but further redefine how research connections are built.
