Charts built for Chat
dbt Labs open-sourced a YAML-based charting language that lets AI agents generate dashboards as code instead of UI-bound reports, solving the friction between messy code generation and restrictive BI tool interfaces. The accompanying dbtCharts.com platform adds hosting, access control, and conversational analytics on top of the open standard.
- Current AI-generated reports create sprawling file structures across multiple languages, making audits slow and token-expensive; dbt Charts consolidates everything into a single readable YAML file that agents can modify efficiently
- Charts defined in code live in Git alongside dbt models, so schema changes and chart updates ship together in one CI run and fail before reaching production
- The language includes 1,100+ config options across 16 chart types with cascading styles and inheritance, designed to look intentionally crafted rather than dashboard-grid generic
dbt Labs just open-sourced dbtCharts, a YAML-based language for building dashboards that lives in code rather than behind a UI. The problem they're solving is real: AI agents are terrible at clicking through interfaces but fluent in code and Git. When you ask an agent to build you a report, it currently spits out scattered HTML, CSS, JavaScript, and whatever charting library it grabbed—a mess to audit and expensive to iterate on. Traditional BI tools locked charts into their UIs to keep things controlled, but that control cuts both ways: agents can only do what the interface exposes. dbtCharts splits the difference by putting charts into a single, auditable YAML file that declares what data you want (SQL) and how to visualize it (YAML configuration). You can render it locally to SVG or HTML, serve it as a site, or hand the whole thing to an agent to modify.
The language has real depth under a readable surface. It supports sixteen chart types with over 1,100 config options, but you rarely touch most of them because styles cascade from themes and boards can inherit from other boards. When you use it with a dbt project—charts living in a `/charts` folder next to `/models`—the integration gets tight: model changes and chart changes ship in one Git branch, breaking before they hit production. Validation is strict. Run `dct render` and you get warnings like "BAR-BAND-WIDTH-TOO-NARROW: 182 bands x 2 series across 640px" before anyone sees a broken dashboard. The visual design came from RJ Andrews, a data graphic designer, and ditches the typical dashboard grid aesthetic in favor of something that actually thinks about spacing and alignment.
Alongside the open-source language (Apache 2.0), dbt Labs launched dbtCharts.com in public beta—a hosted platform that adds hosting, permissions, version history, and conversational analytics on top of the language. You can chat with an AI analyst that has permissioned read-only access to your warehouse, tweak results in a visual editor, and everything syncs back to the same YAML in your Git repo. Nothing gets locked in. The same board runs on your laptop, in CI, and on the platform. You can install it yourself with `uv tool install dbt-charts` or just hand that line to your coding agent.