Getting Started

Quickstart

From sign-in to your first chart in a few minutes. This walks the golden path; every step has a dedicated deep-dive linked at the end.

1 · Sign in

Open Kaveon and sign in. Authentication is provider-based (no local passwords by default) — GitHub, Google, or Microsoft Entra ID light up automatically when their credentials are present. The API’s authorization layer defines a four-role ladder (Viewer → Analyst → Editor → Admin), but the NextAuth sign-in resolves you to Admin (if your email is in AUTH_ADMIN_EMAILS) or Viewer.

Running it yourself? See Auth & RBAC for wiring GitHub/Google/Microsoft and the role model. The rest of this page assumes you’re signed in.

2 · Connect a data source

Go to Data Sources → + Add Data Source. Kaveon speaks to Microsoft Fabric SQL, Azure SQL, PostgreSQL, MySQL, and StarRocks. Fill in the connection, click Test, and save — no.env edits, no restart.

No warehouse handy? Skip this step. Kaveon ships with demo datasets (COVID & NYC taxi) in its metadata database, so every feature below works out of the box.

3 · Run a query

Open Lab, pick your database in the toolbar, and run something. The schema browser on the left loads tables and columns to power autocomplete.

sql

SELECT region, SUM(total) AS revenue
FROM   orders
GROUP  BY region
ORDER  BY revenue DESC;

Press Ctrl/Cmd + Enter to run. Full editor reference: SQL Lab.

4 · Define a dataset

A semantic dataset names your dimensions, metrics, and filter columns once so charts can be built without re-writing SQL. Go to Datasets → + New Dataset, pick a table (or join a few), and mark which columns are dimensions vs. metrics.

5 · Build a chart

Charts → + New Chart. Choose the dataset, pick a chart type (37, from bars to a 3D globe), and drag dimensions and metrics into place. Kaveon generates the JOINs and aggregation for you and renders live.

6 · Assemble a dashboard

Dashboards → + New Dashboard gives you a drag-and-drop canvas. Drop charts into resizable tiles, add markdown and section headers, and wire up cross-chart filtering — click a bar to filter every related chart at once.

7 · Just ask

On the home page, type a question in plain English — “top regions by revenue this quarter” — and Kaveon detects the right dataset, builds the SQL, runs it, and renders a chart inline. How that works: AI · NL→SQL.