Your entire analyticsas code

Analytics as code means your analytics layer is defined in version-controlled files instead of locked inside a UI. In Lightdash, metrics, charts, and dashboards are YAML in Git, so you can build them with AI coding agents like Claude Code or Cursor, review changes in pull requests, and roll back mistakes.

No. Business users explore data, filter dashboards, and ask questions through the UI and AI agents, same as any analytics tool. Analytics as code is for your data team and your coding agents: they build and maintain content in Git while everyone else consumes it without knowing the YAML exists.

Changes go through the same controls as your software. Spin up a preview environment from the CLI, run lightdash validate in CI to catch charts a dbt change would break, and review every edit as a pull request before it merges. Nothing lands in production without a human approving it.

Great! We're always around to help. You can chat to us through the live chat on this webpage, or talk to us in our Slack community.

No. Lightdash builds its semantic layer from your dbt project or from standalone Lightdash YAML files pointed at your warehouse, with the same syntax. If you adopt dbt later, migration is straightforward. Either way, your metrics are defined once, in code, and every chart is built on them.

Run lightdash install-skills and your coding agent learns Lightdash conventions: how to write valid chart and dashboard YAML on your defined metrics. The Lightdash MCP server does the same for Claude and your internal agents. Because agents build on the semantic layer, they use governed definitions instead of guessing at SQL.

No. Lightdash supports traditional exploration and dashboards alongside AI. Teams can adopt analytics as code gradually.