Hosted
Try the sandbox
A hosted environment seeded on DuckDB, running on our infrastructure. Sign in with GitHub and start. It is shared and seeded — evaluating against your own data is the other door.
Get started
One is hosted and seeded, and you sign in with GitHub. The other runs inside your own tenant, on your own platform. Pick the one that matches how your team needs to evaluate.
Hosted
A hosted environment seeded on DuckDB, running on our infrastructure. Sign in with GitHub and start. It is shared and seeded — evaluating against your own data is the other door.
Your tenant
Studio runs as containers on one machine under local Docker, or on an existing AKS cluster under Kubernetes on Azure, against DuckDB (local), Microsoft Fabric, and MotherDuck. The starter pack is the operator document for whichever combination you pick.
| Sandbox | Self-host | |
|---|---|---|
| Time to first result | Your first session begins as soon as you sign in. | One installer and one command, once the prerequisites are met. Local Docker on DuckDB needs one admin request; Kubernetes on Azure with Microsoft Fabric can reach four admin pages and up to eight separate right-holders. |
| What you need | A GitHub account. Sign-in goes through GitHub. | Local Docker: Docker Engine with the Compose plugin, the GitHub CLI, and port 5173 free. Kubernetes on Azure: an existing AKS cluster sized for it, with kubectl and helm. Every combination needs an Azure AI Foundry resource. |
| What you get | The hosted environment, seeded on DuckDB. Author a pipeline, watch it fail a gate, ship it when it passes. | Studio inside your own perimeter, on the platform your team already runs, using your own keys and your own model endpoints. |
| Your data | Seeded sample data only. It is a shared environment, so production, personal and confidential data stay out of it. | Your own data, in your own perimeter, under your own keys. Vibedata reads metadata, code and execution traces, never your rows. |
| What it costs | Nothing. It runs on our infrastructure. | The starter pack is free. Your platform capacity, your cluster and your model calls run on your own accounts; what a licence costs is a conversation. |
The self-host column comes from the deployment prerequisites in the starter pack, which is where each of those asks is written out in full.