dbt Charts, from dbt Labs, gives an AI-generated dashboard a structured source file that people can inspect. SQL specifies the data, while YAML specifies its presentation, bringing queries, charts, and layout into one dashboard definition.
The project's command-line tools validate the definition and flag visualization problems such as overcrowded bars or overflowing tables. The post presents this as a tight feedback loop for agents, which otherwise work blind, and the YAML lives in the team's Git repository, where changes can be reviewed. The open-source language can run locally; the separately launched hosted beta adds conversational analytics, editing, and sharing.
