dbt Labs open-sources dbt Charts for agents
dbt Labs has introduced dbt Charts, an open-source (Apache 2.0) declarative language that brings dashboard creation out of monolithic BI tools and directly into version-controlled code. Built specifically for human and AI collaboration, dbt Charts combines SQL data queries with structured YAML configurations for layout, Jinja templating, and cascading styles across 16 chart types. It integrates directly into existing dbt repositories, allowing charts to be validated alongside data models in CI pipelines and rendered via CLI into formats like SVG, HTML, PDF, and terminal output. Alongside the open-source specification, dbt Labs launched a public beta of dbtCharts.com to serve as the hosted BI platform handling user permissions, hosting, and conversational analytics.
Declarative dashboards as code are the missing link for enterprise AI analytics, replacing brittle point-and-click BI UIs with verifiable, agent-friendly Git workflows. Because LLMs struggle with visual BI user interfaces but excel at structured text, dbt Charts gives agents a concise YAML syntax and compile-time visualization warnings to iteratively refine charts. Storing dashboard definitions in a charts directory alongside data models allows data teams to run automated checks via dct validate and catch breaking schema changes before production deployment. This strips visualization logic out of the proprietary BI tier into an open standard, leaving hosted platforms to focus primarily on governance and access management, though long-term adoption against established code-first alternatives like Evidence.dev will depend on how well non-technical stakeholders adapt to code-driven reports.
DISCOVERED
2h ago
2026-09-15
PUBLISHED
5h ago
2026-09-14
RELEVANCE
AUTHOR
thingsilearned