Obsvra markObsvra
Dark data corridor lit by warm lamps
$ spicegrinder model init \
  --domain healthcare-claims \
  --scale 50M \
  --privacy synthetic-only

dataset generate --r
Obsvra · SpiceGrinder · Coming Soon

Synthetic data ground fresh from your own domain model.

SpiceGrinder lets your teams describe the shape of their real data, then generate synthetic datasets at any scale and complexity — without ever touching PII, PHI, or a compliance review queue.

Capabilities

Model your domain once, grind data whenever you need it.

SpiceGrinder is being built for teams whose test, demo, and training work keeps stalling on access to real records. You describe entities, relationships, and distributions — the grinder produces data that behaves like production and carries none of its risk.

Your domain, your model

You know the structure of your own data. SpiceGrinder gives you the tools to describe it in a model. Then the model produces data for you that is indistinguishable from real world data.

Your domain, your model
entity claim {
  id        uuid
  member    ref(member)
  amount    money(20..48000)
  status    enum(open, paid)
}

Any scale, any complexity

Generate a hundred rows for a unit test or fifty million for a load run, keeping referential integrity and edge cases intact at every size.

Any scale, any complexity
rows        50,000,000
relations   preserved
skew        production-like
runtime     11m 42s

No PII, PHI, or waiting

Nothing is copied from production, so datasets can move freely between environments, vendors, and demos without a privacy review blocking the work.

No PII, PHI, or waiting
source      model-only
pii         none
phi         none
export      unrestricted
How it will work

Three moves from real-world shape to safe data.

You understand your own data better than anyone. Model it with SpiceGrinder's built-in tools, generate as much or as little data as you need, and use the outputs however you like.

  1. 01

    Describe the domain

    Sketch your entities and the rules that connect them, from primary keys to the ugly edge cases your production data actually contains.

  2. 02

    Grind the dataset

    Choose scale, fidelity, and skew, then generate as often as you like — every run is reproducible from the model, never from a copy of production.

  3. 03

    Ship it anywhere

    You own the data, use it wherever you like. Export into environments, pipelines, sales demos, or partner sandboxes with no sensitive records to guard and no approval to chase.

Early access

Be first in line for the grind.

Tell us a little about the data you wish you could hand your engineers, and we will reach out as SpiceGrinder opens up its first private builds.

Launch updates only · no newsletter

We use your details only to contact you about SpiceGrinder availability.