Dashboard Detective BI navigation assistant  


UNE BI team builds assistant guide for our data landscape. 

When you're managing dozens of Power BI reports with hundreds of pages, finding the right information becomes a significant challenge. Dwayne Hurst, Manager of Analytics Insights and Reporting, recognised this pain point and built a solution that could transform how staff navigate the university's extensive data resources.

 

The challenge of data discovery
"We make lots of Power BI reports," Dwayne explains. "There's scores of them, and there's hundreds of pages, and there's lots of overlaying data - very hard to find what you're looking for." His solution is Dashboard Detective, an AI assistant designed to help all UNE staff quickly locate the specific reports and pages they need.

The assistant acts as an intelligent guide rather than a data analyst. "If you want to say how many students are studying nursing, it won't tell you the number - but it will go and find the report where you could find those numbers," Dwayne notes. Currently the assistant only finds reports from the Bi Team, but Dwayne is open to adding any reports from across UNE.

 

Page-level precision
What sets Dashboard Detective apart is its page-specific approach. Instead of simply linking users to entire reports, the assistant directs them to the exact page containing their information.

"This assistant is page driven, not report driven," Dwayne explains. "If you just link the person to the report, I think it still might be challenging for the person to find out where they're getting to go on what page to find it."

When staff query the assistant, it provides comprehensive guidance including the app name, report name, page name, recommended visualisations, available filters, and direct links to the relevant page.

 

Building knowledge with AI assistance
The creation process demonstrates how staff can leverage AI to build AI solutions. Dwayne has systematically documented approximately 160 Power BI pages (and still growing), using Madgwick itself to streamline the process.

"I open the assistant, share my screen, make it look at the Power BI report page, and I tell it to get the app name, get the report name, get the page name, get the filters, get the visuals," he describes. "Copy, paste, save, done… 15 seconds to do a page."

The assistant's instructions were refined during development sessions at AI Café, and then Dwayne used LabNext70’s custom AI Assistant creator - Madgwick Wizard - to continue improving iteratively over several weeks.

 

Future potential
While currently focused on navigation, Dashboard Detective represents a foundation for more advanced capabilities. "The ultimate goal for me and my team is like, don't just find me the report, find me the report and give me the analysis out of it," Dwayne says.

The assistant has recently been released on the DataHub with feedback being sought from colleagues before broader release. For staff who struggle to navigate UNE's comprehensive BI landscape, Dashboard Detective promises to transform data discovery from a time-consuming search into a simple conversation.

 
Previous
Previous

New AI architecture challenges large models on complex reasoning

Next
Next

How to convert tables into LLM friendly format