Original LinkedIn Post
Recently, I took inspiration from James Goodall's excellent Tableau Public project, the "Tableau Advent Calendar," and used it as a benchmark for my open-source Tableau workbook generation tool, cwtwb.
To be honest, while trying to recreate these stunning visualizations with code, my biggest challenge was restraint. As we all know, Tableau's visual flexibility is virtually limitless. If you try hard enough, you can hack together almost any wild chart type imaginable. I found myself slipping into the endless rabbit hole of wanting to support every cool visualization out there.
To stop myself from burning out on this endless development cycle, and to ensure the tool actually delivers practical value, I decided to pull back the reins. I redefined the core focus of cwtwb: stripping away the flashy gimmicks and intensely focusing on efficiently building high-value, standard business dashboards.
Driven by this pragmatic focus, I completely overhauled the generation logic. cwtwb has moved away from hacky, one-off workarounds to a highly structured and extensible architecture. Additionally, I am thrilled to announce that it now officially supports connecting to Hyper data extracts (.hyper)!
Currently, cwtwb solidly supports 15 core business visualization charts, structured across clear routing policies: Core Primitives: Bar Chart, Line Chart, Area Chart, Pie Chart, and Scatter Plot. Text & KPIs: Text Tables (Crosstabs) and Measure Values (KPI Cards). Advanced Patterns & Dual Axis: Dual Combo Chart, Donut Chart, Lollipop Chart, and Butterfly Chart. Custom Recipes: Calendar Chart (featuring complex date filters and custom XML injection).
What I am most proud of with this milestone is not just the growing number of charts, but that the system no longer treats every new visual as a special case. Whether it is handling complex dual-axis synchronization, independent axis reversal, or fine-grained formatting control like hiding axes and zero lines, the underlying engine outputs clean, valid, and native Tableau XML structures. This also lays an incredibly solid foundation for seamlessly integrating AI agents in the future.
This is the kind of engineering work I enjoy most: drawing inspiration from a beautiful public project, restraining the urge to over-engineer, and turning that inspiration into a reusable system focused on real business value.
Southard Jones Elif Tutuk Matthew Miller Adam Mico Allan Folting Roy Raviv Would love your thoughts!
If you are exploring automated Tableau workbook generation or are interested in building business dashboards using code and AI, I highly recommend checking out the cwtwb open-source project. I welcome you to explore it and look forward to seeing what you build.
#Tableau #DataVisualization #ArtificialIntelligence #BusinessIntelligence #OpenSource #DataAnalytics #MCP #AIAgents #DataEngineering #datafam #gemini #google #ai #bi #agent #cwtwb #codex #chatgpt