Curtin University: Turning “Who Do I Even Talk To?” Into a Plain-English Search That Finds the Right Expert in Seconds

Brief

We built an AI-powered search chatbot that knows Curtin’s schools and has deep knowledge of every staff profile loaded into it: career history, research interests, publications, awards and more. Under the hood, an agentic search system answers simple queries instantly and digs deeper on hard ones, drawing on staff career histories, research interests, publications and awards. The whole system runs on Azure infrastructure, with no data leaving the environment, and was built to scale to any number of staff profiles.

Problem

When a company decides it wants to work with a university, the clock starts ticking. Someone in that business has a problem needing deep expertise, and they know Curtin probably has the right person, they just have no idea who. The information existed, but it lived everywhere and nowhere, spread across school pages, staff profiles and separate portals, each organised for the university rather than for the visitor. Finding the right expert meant manually trawling a complicated website, guessing at school names, and hoping the right profile surfaced. For the Curtin Industry Exchange team, whose whole job is connecting industry with researchers, that friction was the bottleneck, and promising engagements went cold simply because getting started was too hard. They needed a single entry point where a corporate stakeholder could describe their problem in their own words and be pointed to the right school and the right people, with a reason why, in seconds rather than hours.

Solution

We built an AI search agent that knows Curtin’s schools and has deep knowledge of every staff profile loaded into it: career history, research interests, publications, awards and more. A stakeholder types what they need in plain English, and the concierge recommends the right school and a shortlist of staff, each with a tailored explanation of why they’re a strong match.

• Understands intent, not just keywords: a query like “we need help automating quality inspection on a production line” finds the right robotics and computer vision researchers, even if no profile uses those exact words
• Fast when it can be, thorough when it must be: the agentic search design answers straightforward queries instantly and searches longer and harder for difficult ones, so quality never drops on the tricky questions
• Every recommendation comes with a why: instead of a ranked list of names, users get a personalised explanation for each suggested expert, making the next step obvious
• Secure by design: a privately hosted Azure model and vector database keep every byte of Curtin data inside a controlled infrastructure
• Built to scale from day one: automated scrapers collect and refresh profile data, with no ceiling on how many staff or schools the system can cover

Figure 1: Curtin AI Concierge recommending the appropriate school and staff.

Result

Corporate stakeholders can now describe what they need in plain English and receive a recommended school and a shortlist of relevant experts in seconds, work that previously meant manually hunting through a complicated website. Unlike a traditional search, every result is personalised to the query, with a clear explanation of why each researcher is a fit, so the path from “we have a problem” to “we’re talking to the right person” is dramatically shorter. For the Curtin Industry Exchange team, the build proved the concept end to end, that an AI concierge over their existing public data could become the front door for all industry engagement. The system was deliberately architected for scale and simple deployment within Curtin’s own Azure environment, and their internal team is now positioned to extend it across the wider university.

Under the Hood