User Manual
Dashboards & Datasets
Turn your quality data into datasets and dashboards. Build a pipeline once, chart it in a click, and share the result by schedule, export, or plain-English question.
Build a dataset
A dataset is the shaped, reusable view of your data that everything else charts from. You build one from a source using a step-by-step pipeline, so each transformation is an explicit step you can review rather than a black box.
- •
Pick a source
Start the pipeline from a source and add the steps that shape the data. - •
Choose a display report and preview
Pick a display report and preview the data so you can confirm it looks right before you chart anything. - •
Refine the SQL if needed
Open the power mode to view and edit the generated SQL directly when you need finer control than the visual steps provide. - •
Set a refresh schedule
Refresh the dataset on a schedule so downstream charts stay current on their own.
Chart onto a dashboard
Once a dataset exists, you can chart it onto a dashboard in one click. Dashboards are designed for exploration, not just static display, and support a wide range of chart types so you can match the visual to the question.
- Many chart types to suit different data and comparisons.
- Global filters that apply across the dashboard, plus per-widget filters.
- Cross-filtering, so selecting in one widget filters the others.
- Click-to-drill-down, to move from a summary into its detail.
◳ Screenshot
A dashboard with global and per-widget filters, cross-filtering, and drill-down.
Ask the dashboard AI agent
You can ask the dashboard AI agent a question in plain English instead of building a chart by hand. It works against your data, so the answers describe your workspace rather than a generic example.
Go deeper with scripts and the Data Lab
For analysis beyond charting, Kanso supports Python and R script widgets that operate on curated datasets, and a Data Lab for deeper, exploratory work on the same curated data.
When to reach for scripts
Use a script widget when a standard chart can't express the calculation you need — the script runs on the curated dataset so it inherits the same governed, refreshed data as the rest of your dashboard.