Live sandbox
Run it on a real stack
A hosted environment seeded on DuckDB: author a pipeline, watch it fail a gate, ship it when it passes. Sign in with GitHub and start.
The sandbox runs on our infrastructure, on a DuckDB stack we seed. It is the inner loop and nothing more than that: say what you want, let the agent author the pipeline, watch a gate stop a change that does not hold up on real rows, and ship it when it passes.
If your evaluation has to happen on your own platform against your own data, the self-host starter pack is the other door.
Tell us what you run
Shared environment Use the sample data we seed. Do not put production, personal or confidential data into the sandbox: content you put in may be visible to our staff and, depending on configuration, to other users. The Sandbox Terms and the privacy policy set this out in full.
Not wired The fields, both attestations and the GitHub button are markup only in this prototype. Nothing is submitted and no account is created. Until this door is wired, hello@acceleratedata.ai reaches a founder directly.
Signing in through GitHub is one grant, and it is all that getting in takes. Letting Vibedata commit its work to a repository is a second, separate grant, asked for inside Studio once you are there — signing in does not give it, and you see what it covers before you agree.