Today, Lightdash natively integrates with MotherDuck. Connect your MotherDuck warehouse to Lightdash in five minutes and get governed, dbt-native analytics running at DuckDB speed, all the way through to your dashboards.

A warehouse built for the way modern data teams work

MotherDuck is a serverless data warehouse built on DuckDB.  There are no clusters to size, no warehouses to suspend, and no servers to provision. Compute spins up in under a second and you pay per second of actual work.

Under the hood, MotherDuck's hypertenancy architecture gives every user their own dedicated compute instance against shared data. That means the nightly dbt job, the board report, and the analyst running ad-hoc queries don't fight for the same resources — they each get their own. 

This matters even more when agents are doing the querying. Agents don't query like humans. They fire thousands of small, iterative queries until they land on the right answer. On a traditional warehouse, that's expensive and risky; you’re effectively running the warehouse 24/7. With hypertenancy, each agent runs in full isolation: sub-second startup, per-second billing, completely separate from your team's workloads.

"Lightdash and MotherDuck are both building for the agentic era. Hypertenancy gives every agent its own isolated compute — the right architecture for analytics when agents, not just people, are running the queries. Lightdash brings the governed, code-first layer on top of it.”
— Nouras Haddad, VP Partnerships & Business Development, MotherDuck

A BI layer built for code-first teams

Lightdash keeps your whole BI layer in code. It isn't only your metrics that live in code — your charts, your dashboards, and your access permissions do too. All of it version-controlled, reviewed in pull requests, and deployed through Git and CI like any other part of your stack. You can build and update it from Cursor or Claude Code without ever opening a UI.

Your metrics are defined once, in a governed semantic layer that becomes the single source of truth. For a MotherDuck project that definition lives in your dbt models, so what your engineers built and tested is exactly what shows up in the dashboard. Because that layer is governed, your AI agents query through it, not around it. Every query runs against the same tested metrics, whether a human writes it or an agent fires off a hundred. You get the speed of DuckDB with the governance your data team actually built.

The whole analytics stack stays in code, reviewed, and version-controlled.

"Lightdash customers who move to MotherDuck immediately notice two things: queries are faster, and there's nothing to manage. DuckDB does the heavy lifting; Lightdash reads the dbt project directly. For a data team that's tired of babysitting infrastructure, it's a meaningful difference."
— Ian Ahuja, Head of Sales & Partnerships, Lightdash

What you get

• Your BI layer in code. Metrics, charts, dashboards, and permissions version-controlled and reviewed in pull requests, not rebuilt by hand in a UI.

• A governed semantic layer. Agents query through, not around, so every query hits the same tested metrics whether a person or an agent runs it.

• Per-user and per-agent compute isolation. Heavy queries don't slow your warehouse, and your warehouse doesn't slow your dashboards.

• No infrastructure to manage. MotherDuck handles compute; Lightdash handles the semantic layer.

Set up in 5 minutes

In Lightdash, go to Settings → Connections → Add warehouse, select MotherDuck, paste your access token and database name, and point Lightdash at your dbt project. 

Note: MotherDuck connections require dbt v1.8 or later. 

Full setup docs here.

Running on DuckDB and want a BI layer that keeps up? Start your free trial.