Learn - Intro to Canvas/Analysis [Replaced on February 23, 2026]

The canvas is your analytical workspace. Querying, modeling, and presenting all happen here. You can begin with the AI agent and refine and extend the outputs or start from a blank canvas. On a canvas, you can place cells like SQL in Python, visuals, tables, and maps, and controls. Or you can place objects like stickies, text, or shapes, connectors, stamps, images, and rich media. The data panel shows your available database connections, catalogs in your semantic layer, and information about cells on your canvas. The result of one cell can be used by another. Everything is connected. Cells reference each other directly. If you change upstream logic, everything downstream updates. Add control cells to turn your analysis into something interactive. Visual cells are built using the visual editor. You don't have to run every query on your data warehouse. Count uses DuckDB to run analysis efficiently. That means you can run analysis in browser for fast local exploration or on the server for larger workloads. This speeds up iteration and can reduce warehouse costs. Your data can be cached for the duration you choose, with the option to schedule refreshes of the data if you wish. You can use templates to standardize outputs and create slides. Switch to presentation view for a clean, focused format. And finally, group analysis into frames to organize your canvas. Move seamlessly from working canvas to agentic AI insights to structured report to meeting ready presentation while collaborating at every stage.