Visualising spreadsheet data
Converting spreadsheet data into professional charts shouldn't require hours of Excel wrangling. The Data Visualisation Assistant, conceived by Digital Marketing Coordinator Robert Hughes and developed by LabNext70's Prompt Engineer Tom Miles, bridges this gap - transforming raw data into shareable visualisations through a guided four-step process.
The Madgwick assistant is designed to help staff who require rapid, ad-hoc data investigations without the need for intricate dashboard setups. "This is for data exploration, when you only need it as a one-off instance, and you're trying to explore", Robert explains.
From CSV to interactive charts
The assistant operates through a structured workflow that ensures consistency. After uploading a dataset - whether CSV, XLSX, or even pasted data - users are guided through four steps: verification, analysis, chart selection, and final visualisation.
Example in action: Robert loaded early entry applications data, and the assistant broke it down by application channel (early entry, UAC, and direct). It then generated interactive HTML charts complete with hover states, animations, and professional styling aligned with UNE's BI team design standards.
"What's really nice here is it automatically encodes all of the nice, more detailed interactions you would like - like hover states to look at individual numbers on the charts", Robert notes. The output is a standalone HTML file that can be shared, screen-grabbed, or exported as PDF.
Solving the consistency challenge
The technical challenge was ensuring consistency, Tom explains: "The issue that Rob had was he wanted to generate good quality charts for usage in UNE that were always very similar, so that you don't have six types of charts that all look different”.
The solution came through templates embedded in the assistant's instructions, so it produces consistent, professional visualisations - whether line charts for trends, bar charts for comparisons, or pie charts for proportions.
"The assistant instructions have got the template of a couple of types of different charts", Tom notes, emphasising how the structured approach ensures visual coherence across outputs.
He adds, "The assistant works most reliably with Excel documents of no more than 250 rows of data, but some users have success with larger documents".
Designed for non-coders
Crucially, the assistant requires no technical expertise. It guides users through chart creation by explaining choices, suggesting best-fit visualisations for the dataset, and producing ready-to-use HTML. All that’s needed is a basic dataset — categories and metrics for comparisons, or time periods and values for trends.
The assistant takes care of the technical complexity: cleaning data, applying accessible colour palettes, adding tooltips, and ensuring clear labelling. It can even iterate on the design until the user is satisfied with the final output.
Building your own solution
This project highlights how UNE staff can partner with LabNext70 to develop custom AI assistants that addressspecific workflow challenges.
For staff managing repetitive tasks or workflow bottlenecks, the Data Visualisation Assistant shows what's possible when domain expertise combines with AI development support.