How to Build Dashboards with Rill Developer

Rill Developer is a free, open-source BI tool that takes your data from a CSV file or a cloud source and turns it into an interactive, filterable dashboard in minutes. You install it with one command on macOS or Linux, point it at a file, and use a built-in AI button to generate charts and key metrics automatically. No warehouse required. No SQL expertise assumed. No server to maintain.

What Rill Does Differently

Most BI tools require data to already be in a warehouse. Tableau connects to Snowflake. Looker Studio connects to BigQuery. Metabase connects to Postgres. Before you see a single chart, you need a database, credentials, and a pipeline.

Rill skips that step entirely. It bundles an embedded copy of DuckDB, reads your data files directly, and exposes a full interactive dashboard in your browser. A 500,000-row CSV that takes minutes to sort in Excel loads and queries in Rill in under one second. The project has accumulated more than 5,700 GitHub stars since its 2023 open-source release, with teams using it to replace ad-hoc SQL notebooks and Retool dashboards for day-to-day operational monitoring.

The architecture separates three layers: data sourcing, metrics definition, and dashboard layout. That means you define business KPIs once and they stay consistent across every chart, rather than recalculating the same formula in thirty different places.

Installing Rill

Rill runs on macOS and Linux. Install it with a single shell command:

curl https://rill.sh | sh

Then create and launch your first project:

rill start my-dashboard

This opens Rill Developer in your browser at http://localhost:9009. You will see a blank project with a sidebar on the left and a prompt to add your first data source. On Windows, install WSL2 first, then run the same commands inside the WSL terminal.

Connecting Your Data

Rill connects to local files, cloud buckets, and databases. For a first project, the simplest option is a local CSV or Parquet file.

Click "Add Data" in the sidebar and choose "Local file." Navigate to your file, select it, and click Connect. Rill reads the schema automatically, shows a preview of the first 150 rows, and creates a source definition file in your project directory.

If your data is in cloud storage, the process is identical. For S3, enter the path as s3://bucket-name/path/file.parquet. For Google Cloud Storage, use gs://bucket-name/file.parquet. Rill reads authentication from your local AWS or GCP credential files. For BigQuery, Snowflake, ClickHouse, and Postgres, provide connection credentials once in a rill.yaml file and Rill stores them for the project.

Transforming Data with Models

Once your source is connected, you can write SQL to clean or reshape data before it reaches the dashboard. Rill calls this layer a "model."

For most use cases, no transformation is needed. If your CSV already has clean columns, skip directly to the dashboard step. If you need to filter rows, rename fields, or join two tables, click "Add Model" and write a query against your source:

select
  order_date,
  region,
  product_category,
  revenue,
  units_sold,
  revenue / units_sold as avg_order_value
from orders
where order_date >= '2024-01-01'

Rill runs the query through DuckDB and materializes the result. Every time you open the project, the model runs fresh against the source file. If the source is updated, the dashboard reflects the change automatically.

Generating a Dashboard with AI

This is where Rill separates from the standard BI setup flow. Instead of dragging chart elements onto a blank canvas, Rill generates the full metrics definition and dashboard structure from your data with one click.

In the sidebar, hover over your model or source, click the three-dot menu, and select "Generate dashboard with AI." Rill analyzes your column names and data types, infers which columns are dimensions (region, category, status) and which are measures (revenue, units, count), and writes a metrics_view.yaml file and an explore_dashboard.yaml file automatically.

Open the generated dashboard. You get a time-series chart at the top, a leaderboard of top dimension values on the left, and a comparison table on the right. Every element is interactive. Clicking any row in the leaderboard filters the entire dashboard to that segment. Changing the time range updates every chart simultaneously. Adding or removing measures from the comparison panel requires no code changes.

That kind of synchronized cross-filtering across all charts at once is the feature that typically costs thousands of dollars per month in tools like ThoughtSpot or Sigma Computing. In Rill, it is free and runs on your laptop.

Editing the Metrics Layer

Rill stores your dashboard definition as YAML files inside your project folder. You can edit them to add custom measures, rename dimensions, or set default filters.

A typical metrics view file looks like this:

model: orders
timeseries: order_date
measures:
  - label: Total Revenue
    expression: sum(revenue)
    format_preset: currency_usd
  - label: Orders
    expression: count(*)
  - label: Avg Order Value
    expression: sum(revenue) / count(*)
    format_preset: currency_usd
dimensions:
  - column: region
    label: Region
  - column: product_category
    label: Category

Change any label or expression, save the file, and Rill reloads the dashboard instantly. Because all definitions live in text files, you can version them in Git, copy them for different teams, or pass them to an AI coding agent like Claude Code or Cursor for programmatic edits.

Sharing with Rill Cloud

Rill Developer is a local development tool. To share a dashboard with teammates, deploy to Rill Cloud with one command from your project directory:

rill deploy

Rill pushes your project to the cloud, connects to your data source using cloud credentials, and gives you a shareable URL. Recipients get the same interactive Explore dashboard in their browser with no local setup required.

Rill Cloud adds an AI Chat feature that lets users type questions such as "show me revenue by region for Q1" and get a filtered dashboard view in response. It also adds scheduled email reports and metric threshold alerts. The free tier supports up to five users.

When Rill Is the Right Tool

Rill excels when your data is relatively clean and your primary need is exploration and operational monitoring rather than pixel-perfect report layouts or custom chart types. The Explore dashboard is opinionated: time-series chart, leaderboards, and comparison tables. You cannot build a freeform canvas or add a chart type from a third-party library.

For pure exploration on CSV or Parquet data, Rill is faster to set up than Metabase, Evidence.dev, or Apache Superset. A benchmark published by the Rill team in 2024 showed queries on a 100-million-row Parquet file completing in under two seconds on a standard laptop, versus 15 to 30 seconds for an equivalent Pandas-based notebook operation.

If your file is messy and you want instant statistical analysis, AI-generated charts, and plain-English query results from an upload prompt, VSLZ handles that from a file drop with no YAML or SQL configuration.

Next Steps

After your first working dashboard in Rill Developer, the most useful next actions are setting up metric alerts, which send an email when a value moves outside a defined range, and scheduling a daily report that delivers a snapshot of your key numbers each morning. Both are available in Rill Cloud under project settings.

For teams with larger datasets or real-time event streams, Rill supports ClickHouse as an alternative engine. Once connected, Rill can query billions of rows with sub-second response times while supporting multiple concurrent dashboard users.