Glonce

dbt Labs open sources dbt Charts, a YAML language for dashboards

toolsThe Glonce editorsHacker News🔥70Glonce rating70Impact60Novelty55Relevance85Surprise40Players55Credibility85
dbt Labs open sources dbt Charts, a YAML language for dashboards

dbt Labs, the company behind the dbt data transformation tool, has open sourced dbt Charts under the Apache 2.0 license: a structured YAML language that declares a full interactive dashboard in one auditable file. SQL still defines what data to query, while the YAML around it declares how to display it, and Markdown and Jinja carry prose and variables inside the same file. The language covers over 1,100 config options across sixteen chart types, and its CLI (dct) can render a board file to static SVG, HTML, PNG, PDF or the terminal, or serve a folder of boards as a site.

The company frames the move as the next step in an unbundling of business intelligence tools that dbt Labs founder Tristan Handy described in an essay called BI's Second Unbundling. He wrote that BI tools used to be full-stack products handling ingestion, transformation, compute, caching, semantics and visualization all at once, quoting MicroStrategy and Cognos as examples: "The BI tool was the data stack. MicroStrategy, Cognos, etc: they're not just visualization tools, they're integrated data platforms." From roughly 2015 to 2022, according to Handy, the infrastructure layers of that bundle were pulled into separate purpose-built tools: compute to the major cloud providers, ingestion to Fivetran, transformation to dbt. What was left in BI tools was visualization, interactive interfaces, semantic definitions, identity and access management, and web hosting. dbt Labs argues charts are the next piece to leave the UI, because AI agents write code, SQL and Git natively but are clumsy operating someone else's user interface.

When used inside a dbt project, a charts/ directory sits next to the existing models/ directory in the same Git repository, so a chart and the model that feeds it change on the same branch and CI run. Chart queries reference models through dbt's ref() function, resolved from the project's manifest, so a renamed or missing column fails a pull request before a warehouse rebuild runs. Support for dbt's Semantic Layer is planned so a board could reuse a metric's existing definition instead of restating it in SQL, though no date is given for that. The renderer also runs a set of visualization checks and prints warnings before a board is shown, such as flagging bar charts with bands packed too tightly for the available width or tables wider than their allotted space.

Alongside the open-source language, dbt Labs is launching dbtCharts.com in public beta, a hosted platform built on top of dbt Charts. It adds warehouse connections, permissioned sharing so viewers do not need warehouse logins, a visual editor, version history, and conversational analytics: a chat interface with read-only access to a connected warehouse that can build and edit boards, with every change from chat, the visual editor or hand-written code landing in the same YAML files in the user's own Git repo. dbt Labs recruited data graphic designer, author and historian RJ Andrews to design the chart visuals. dbt Charts is described as pre-1.0 and still changing; the company says that when its grammar changes, existing boards migrate automatically as they are parsed, so boards written today keep rendering. The language can be installed and used locally, including with dbtCharts.com, without creating an account, via uv tool install dbt-charts.

Key facts

  • dbt Labs open sourced dbt Charts under Apache 2.0, a YAML language that declares a full interactive dashboard in one file, with SQL for the data and YAML for its presentation.
  • The language spans over 1,100 config options across sixteen chart types, and its dct CLI renders boards to SVG, HTML, PNG, PDF, the terminal, or a served site.
  • Inside a dbt project, charts/ sits beside models/ in the same repo, and chart queries resolve through dbt's ref(), so a broken model reference fails the pull request before a warehouse rebuild.
  • dbt Labs simultaneously launched dbtCharts.com in public beta, a hosted BI platform adding warehouse connections, sharing, a visual editor and a chat-based conversational analytics interface on top of dbt Charts.
  • dbt Charts is pre-1.0 and still changing; the company says boards automatically migrate as the grammar evolves, so files written today keep rendering.

Why it matters

dbt Labs positions this as the next stage of what it calls BI's ongoing unbundling: after ingestion, compute and transformation were pulled out of monolithic BI tools between roughly 2015 and 2022 into products like Fivetran and dbt itself, visualization is what remains bundled into BI tools' user interfaces. dbt Labs argues that as chat agents become the primary way people interact with software, the fact that agents are fluent in code and SQL but clumsy inside someone else's UI makes it worth moving charts out of the BI tool and into a text file an agent can read, write and version like any other code.

Who it affects

The open-source language targets data teams and developers already using dbt, or anyone building dashboards with an AI coding agent, since it is designed to be generated and edited by chat as easily as by hand. The hosted dbtCharts.com platform targets teams that want warehouse connections, access control and a shareable UI on top of the same open format, including people who need to view a board without holding a warehouse login themselves.

How to use it

dbt Charts is free and open source under the Apache 2.0 license; it can be installed, and boards authored, rendered and served locally without creating an account, via uv tool install dbt-charts. Inside a dbt project, board files live in a new charts/ directory next to models/, and dbt parse && dct validate charts/ checks that chart queries still resolve against the project's models before anything is deployed. dbtCharts.com, the separately hosted platform built on the same language, is currently in public beta; no pricing or specific launch date for it is given in the source.

How solid is it

The claim rests on dbt Labs' own blog post announcing its product, plus an essay by its founder, Tristan Handy, describing the BI market's evolution; both are self-published by the company with an obvious commercial interest in the framing. The company states dbt Charts is pre-1.0 and still changing, but says its renderer migrates existing boards automatically when the grammar changes, so files written now should keep working across updates. It also built in validation and layout warnings, such as a check that flags bar charts with bands too narrow for the display width, aimed at catching visualization problems automatically.

Risks and caveats

As a pre-1.0 project, the language's grammar is still in flux, and while dbt Labs says boards migrate automatically, that guarantee is the vendor's own claim rather than an independently verified track record. Deeper dbt integration, including planned support for reusing metrics from the dbt Semantic Layer, is not yet built and carries no stated timeline. The source gives no pricing, company size, funding, or customer names for either dbt Charts or dbtCharts.com, and the hosted platform is explicitly a beta rather than a finished product.

“The BI tool was the data stack. MicroStrategy, Cognos, etc: they're not just visualization tools, they're integrated data platforms.”

— Tristan Handy, dbt Labs founder