dbt Labs launches dbt Charts, an open-source YAML language for dashboards
dbt Labs has launched dbt Charts, an open-source language that lets teams define dashboards in simple YAML files stored alongside their dbt projects. The company introduced it on the Monday of dbt Summit 2026 at dbtcharts.com.
- dbt Labs launched dbt Charts, an open-source YAML language for dashboards, at dbt Summit 2026.
- A charts folder sits next to the models folder, so dashboards follow the same git tests and deploys as dbt models.
- The language is licensed under Apache 2.0 and runs locally through a CLI called dct.
- A hosted product at dbtcharts.com adds permissioning, sharing, a visual editor, version history and a conversational AI interface.
- Design defaults by RJ Andrews include direct labels instead of legends and decimal-aligned support tables.
The problem dbt Charts is meant to solve
Dave Fowler, general manager and head of the dbt Charts project, described two problems with how dashboards are built today. According to dbt Labs, AI tools have become good at vibe-coding a dashboard from a prompt, but each result comes out differently. Each uses a different mix of JavaScript, Python, CSS and charting libraries, which makes them hard to keep consistent. They are harder still for an analytics engineer to trace back to the underlying query when a business user asks for a double-check.
Traditional BI tools have the opposite issue, Fowler argued. They are UI-first products with AI added afterward, which limits how much an AI agent can actually change.
How the language works
dbt Charts is a simple YAML syntax built around variables, queries, charts, and rows for layout. Fowler compared it to a markdown for dashboards. It adds a charts folder that sits next to a dbt project's models folder. As a result, renaming a column or updating a model runs through the same git tests and deploys in the same branch as everything else in the project.
The language is open source under Apache 2.0, the same license as dbt, and can run locally through a CLI called dct. dbt Labs also offers a hosted product at dbtcharts.com. It adds:
- permissioning and sharing
- a visual editor
- version history
- a conversational AI interface for building and editing dashboards
In a live demo, Fowler edited chart titles, colors and markdown text directly on a dashboard. He then reverted the whole page to an earlier version using its built-in git history.
Design defaults from a data graphics historian
RJ Andrews, the designer of dbt Charts, describes himself as a chartographer and is a historian of data graphics. He walked through the defaults built into the language:
- direct labels instead of legends
- decimal-aligned support tables, so numbers scan cleanly down a column
- axis labels placed on whichever side keeps a reader's eye on the most recent data
The goal, according to the source, is that a dashboard built entirely on dbt Charts defaults, with no custom styling, already looks like something worth spending time reading.
What it means for analytics teams
For analytics engineers, the main change is where dashboards live. Because chart definitions sit in the same repository as models, they follow the same review, testing and deployment path as the rest of the dbt project. A dashboard that an AI agent generates or edits is written in one consistent, readable syntax rather than a varying mix of code, which is the traceability problem Fowler raised.
Teams that prefer to stay local can use the open-source language with the dct CLI. Those who want sharing, permissions and a visual or conversational editing experience can use the hosted product. Recordings of every keynote and breakout from dbt Summit 2026 are available on demand.
