Lightdash is what happens when you treat BI dashboards like code and it is the right approach

If you work in data engineering or analytics and have ever dealt with the nightmare of undocumented, ungoverned dashboards that nobody is sure are pulling from the right source anymore, Lightdash is the answer to that problem.

The core idea is code-first BI. Your dashboards and charts are defined as YAML files rather than built through a GUI and saved in some internal database that only the tool understands. That means they live in version control alongside the rest of your data infrastructure. You can review changes, roll back, track who changed what and why, all the same governance practices you apply to code.

The dbt integration is the foundation of this. If you are already using dbt, Lightdash connects directly to your semantic layer so your metrics have a single source of truth across every dashboard rather than being recalculated differently in different places.

The AI integration is where it gets particularly interesting for teams using tools like Cursor. You can build charts and dashboards using natural language prompts through an AI editor, and the MCP integration gives those AI assistants full context about your data project so the suggestions are grounded in your actual schema rather than generic.

The CLI tools for syncing configurations between local environments and the cloud mean the whole thing fits into a proper engineering workflow rather than requiring a separate manual process for BI.