Original LinkedIn Post
Give AI a dashboard screenshot, and it hands back a perfectly structured Tableau .twb file.
While testing my open-source MCP server, cwtwb, I decided to push it to the limit with a bold experiment: Image-to-Layout.
I fed an existing dashboard screenshot to the LLM along with this highly specific prompt: "Use the cwtwb MCP tool to completely replicate this dashboard layout. Generate the layout JSON and save it locally, then use this JSON to generate the final dashboard. I don't need you to perfectly replicate the calculations; you can just mock up a few. Please use text objects when creating the dashboard, and save the final generated .twb file to the current directory."
The result was incredibly exciting.
As you can see in the screenshot, the final generated dashboard might look like an "ugly" skeleton filled with text placeholders, but that is exactly the magic of this workflow. If you look closely at the Tableau container Item Hierarchy on the right, you will see that the AI perfectly parsed the highly complex horizontal and vertical nesting logic from the original image, achieving a 100% precise structural replication in the local .twb file.
Any heavy Tableau user knows that manually dragging, dropping, and debugging multi-layered nested containers is tedious and exhausting. Now, the most frustrating part of "building the skeleton" can be completed in seconds using natural language. Once that foundational architecture is perfectly built, tweaking styles and populating data becomes a breeze.
Say goodbye to manual dragging and dropping. Building the underlying logic of a dashboard directly through conversation and screenshots?this is the future cwtwb is exploring.
If you are interested in the Model Context Protocol (MCP), AI-assisted development, or data visualization automation, let's connect and discuss.
#Tableau #DataVisualization #ArtificialIntelligence #BusinessIntelligence #OpenSource #DataAnalytics #MCP #AIAgents #DataEngineering #datafam #gemini #google #ai #bi #agent #cwtwb #text2dashboard #autolayout #dataviz