The local-first data workbench.
Open any file or database. Transform it with full lineage. Package the whole workflow as one replayable .dat0 file — and push compute to the cloud only when you choose.
01 · file-native
Drop a 12GB parquet. Scroll it like it's 5MB.
Huge files open instantly, in place. No upload, no import — your data never leaves your machine.
02 · one query
Join a parquet to postgres. One query.
Files and live databases in a single query — no ETL, no load step. Point at them and ask.
03 · cloud on your terms
Too big for your laptop? Push it to the cloud — on your terms.
Local by default. When a compute is too big or too long, offload it to MotherDuck — only when you choose.
04 · reproducible .dat0
Email your whole workflow. They replay it on new data.
Data, transforms, queries, session, and lineage — bundled into one .dat0 file. Send it; they re-run it on their own data.
bundle.dat0
- data→ tables
- transforms→ steps
- queries→ sql
- session→ ui state
- lineage→ graph
This is the workbench we're building.
dat0 is in active development. Join the waitlist and we'll email you the moment there's something to run.
the .dat0 file
One file holds the whole workflow.
A .dat0 file is designed to bundle everything a workflow needs — the data, the transforms, the queries, the UI session, and the full lineage — into one replayable artifact. The idea: attach it to an email; whoever opens it can re-run the exact same workflow on their own data.
- data
- transforms
- queries
- session
- lineage
bundle.dat0
- data→ tables
- transforms→ steps
- queries→ sql
- session→ ui state
- lineage→ graph
what dat0 isn't
Defined as much by what it isn't.
Not a BI tool
dashboards are an output; dat0 is the workbench where you actually shape the data.
Not a notebook
no kernel to babysit, no hidden state — a reproducible file, not a scratchpad.
Not a DB client
it doesn't just query a database — it joins files and databases together, in place.
Not cloud-only
local by default. the cloud is a choice you make, not a requirement.
why we're building this
Your data should never have to leave your machine to be useful.
Every time we wanted to poke at a big file, the tools pushed us toward a cloud we didn't need — upload it, wait, pay, and hope the thing we shipped still matched the thing we explored. The data was already on our laptops. The compute could be too.
So dat0 is being built around a simple belief: local-first, reproducible, yours. Open any file or database where it lives. Shape it with full lineage so the path is never a mystery. Then package the whole workflow as one file someone else can replay — and reach for the cloud only when you actually want it.
None of it ships yet. We're in active development, building in the open, and writing this down so the vision is honest before the binary is. If that's the workbench you've been wanting, get on the list — we'll tell you the moment there's something to run.
Be there when it runs.
One email when there's something to run. No spam, no noise.