ROLE
Design Lead
TIMELINE
1 week
WHAT I DID
Product requirements
Workflow and interface design
AI-assisted Python development
Testing and chart exports
Challenge
Updating Chaptrs’ recurring performance charts required manual adjustments in Figma. Each update took around 30 minutes, while alternative tools introduced styling inconsistencies or additional cost.
All metric names, dates and figures shown in this case study have been replaced with representative data for confidentiality.
A routine monthly update had become a repetitive 30-minute task.
At the end of each month, I update Chaptrs’ performance graphs for the official pitch deck, which has always been designed in Figma for easy collaboration. However, every chart required manual adjustments as it was originally designed from scratch. I tested Excel, which introduced visual inconsistencies and issues with quality, while tools such as Graphy required another paid subscription.
Examples of Chaptrs’ monthly performance graphs, manually updated in Figma. All metric names, dates and figures have been replaced with representative data for confidentiality.
AI made chart creation faster, but the outputs remained inconsistent and required repeated edits.
Before testing, I removed all references to Chaptrs and any labels that could identify the metrics, sharing only decontextualised figures and chart specifications. Despite providing a visual reference and clear design constraints, the outputs remained inconsistent, revealing the need for a more controlled solution.
Edited excerpt from early ChatGPT testing. Company identifiers and metric context were removed before testing; the names and figures shown here have also been replaced for confidentiality.
Instead of prompting ChatGPT each time, could I build a tool that produced charts at the press of a button?
Through further discussions with ChatGPT, I explored a more reliable approach: building a dedicated internal tool with Python that had Chaptrs’ visual design rules embedded into it. This would produce consistent charts from new data without manually redesigning them each time.
The conversation that introduced the idea of replacing repeated chart prompts with a reusable internal tool.
Without coding experience, I used AI to turn my product requirements into a working MVP.
1
I defined the MVP
Specified the design rules, data handling and required export formats.
Select a chart
→
Add or amend data
→
Preview the output
→
Export
2
ChatGPT supported the build
Translated the requirements into a Python script and provided setup guidance.
3
I installed and tested it
Ran the script locally with real data and checked the exported charts.
4
The MVP proved the idea
The charts worked, but the workflow needed easier data updates, live previews and additional controls for visual customisation.
Testing the MVP with real data revealed the features needed for a more usable and flexible workflow.
All dates and figures have been replaced with representative data for confidentiality.
Features needed to improve usability
Modify existing data and add new data
See live previews of the chart
Modify font sizing and spacing if needed
A faster, repeatable way to create every chart.
Chart Studio turns updated performance data into consistently styled, presentation-ready charts. Users can manage data, preview changes and export high-resolution PNG, PDF and SVG files through one streamlined workflow.
Disclaimer: All metric names, dates and figures have been replaced with representative data for confidentiality.
Chart Studio's default screen when opened.
Chart Settings in Chart Studio, where users can adjust spacing and font size, and choose whether to include a background.
Manage Data window within Chart Studio. This window allows the user to edit data for an existing month or add data for a new month.
Impact and reflections
01
A 30-minute task became a five-minute task
Chart Studio significantly reduced the time needed to create our monthly performance charts.
02
Detailed prompts still leave room for mistakes
I learned that AI outputs still need to be carefully tested, no matter how clear the instructions are.
03
Building software felt more achievable than I expected
Creating my first working tool without coding experience was completely new and genuinely satisfying.
04
One successful build opened up new possibilities
Chart Studio gave me renewed confidence to explore other ways AI could improve my workflows.