Notebooks that are not pretending to be a laptop.
A browser notebook wired to a real Apache Spark™ kernel running in your account, with the same login and the same permissions as the rest of the platform.
Separate the notebook from the compute
Start a workspace once and open several notebooks against it. Shut the browser and the kernel keeps its state.
Pick your runtime
Runtime 0 ships Spark 3. Runtime 1.0 ships Spark 4.0. Each bundles matching Hudi, Iceberg and OpenLineage builds, so version conflicts are somebody else's problem.
Publish a notebook as a pipeline
The assistant watches the lineage your notebook emits, works out what it reads from, and offers to turn it into a scheduled job triggered off those sources.
Development lineage stays out of the catalog
Exploratory runs do not pollute production lineage unless an admin turns that on.
On the roadmap, not built yet: instant snapshot sandboxes — an isolated environment built from a data snapshot, for debugging or writing code against something that behaves exactly like production without copying the underlying data.
See it against your own data
We will walk through your workloads and show what runs where.