Run it on a real stack
Vibedata is your data engineering agent. This is a hosted environment seeded on DuckDB — no install, no tenant setup. Author a pipeline, watch it fail a gate, ship it when it passes.
Sign in now. Sandbox provisioning is live — sign in and you're provisioned.
You get your own demo domain and your own GitHub repository, with three worked examples already sitting in them.
Three things, on a seeded stack
Author. Hand the Build agent an intent and watch it generate a dlt pipeline and a dbt model — ingestion and transformation together.
Break it. Push a change that should not ship, and watch the gate stop it. The verification is structural: it runs against real data.
Operate. Open an issue from a detected problem and take it through the Fix agent's explain, investigate, remediate loop.
What it is
You get your own demo domain, seeded on DuckDB and running on shared capacity. The estate is brownfield and everything in it is synthetic, so keep production and sensitive data out of it. To evaluate against your own stack, come to a workshop or talk to us.
Before you sign in
You need a GitHub account with a verified email address.
Vibedata signs you in through GitHub and commits its work to a repository of your own. Sign-in fails if your account's primary email address is not verified.
Using the Sandbox means accepting the Sandbox Terms.
By signing in, you agree to the Sandbox Terms (version 1 August 2026). The Sandbox runs on synthetic data — keep production and personal data out of it.
What happens next. You'll land on Studio and sign in with GitHub. Signing in is all it takes. Studio then walks you through two steps: giving Vibedata permission to commit to a repository, which is a separate grant from the sign-in you have just done, and provisioning your demo domain.
Talk to the people who built it
The monthly community call is open, and the L1 workshop covers the harness end to end. If you run Microsoft Fabric, the L2 workshop goes deeper.