Catalogue
Five lines on the shelf, licensed from the companies that own them. Four more sourced on demand from the firms and experts who hold them. Every one ships with documentation and starts with a test sample.
On the shelf
Sourced, cleaned and ready to sample.
Real engineering, not tutorial code.
Private repositories licensed from software companies with their history intact: how the code changed, why it changed, and who pushed back in review.
Inside
Ships as
Git bundles · JSONL
Starts with
A test sample your team runs before signing
How work actually gets done.
The day-to-day record of real businesses: the tickets, approvals, hand-offs and internal documents that show how decisions move through a company.
Inside
Ships as
JSONL · Parquet
Starts with
A test sample your team runs before signing
Hands-on footage from real shop floors.
First-person video of skilled work in garment, metal, pharma and other factories, with actions annotated. Captured on our rigs, or on yours if you have a setup you trust.
Inside
Ships as
MP4 + annotations · WebDataset
Starts with
A test sample your team runs before signing
Patient journeys, measured over years.
Longitudinal patient records across oncology, cardiology and other specialties: labs, notes and treatment timelines that follow the same people over time. De-identified before anything moves.
Inside
Ships as
Parquet · JSONL
Starts with
A test sample your team runs before signing
Speech in the languages models miss.
Studio-quality speech in Indic and other low-resource languages, transcribed and aligned, with both conversational and read material.
Inside
Ships as
WAV / FLAC + transcripts
Starts with
A test sample your team runs before signing
Sourced on demand
Found to your spec, at the firms and experts who hold it.
Engineering models with the design history.
Parametric parts and assemblies sourced from engineering firms and design teams, with the feature tree that shows how each model was made, not just the final shape.
Inside
Ships as
STEP · native files
Starts with
A pilot batch against your quality bar
Slides with a real argument.
Consulting, banking, board and sales decks sourced from the teams that write them for a living, delivered as editable native files with the structure intact.
Inside
Ships as
PPTX · PDF + JSON
Starts with
A pilot batch against your quality bar
Spreadsheets that actually calculate.
Three-statement, DCF, LBO and operating models sourced from finance and advisory teams, with live formulas and linked tabs rather than pasted values.
Inside
Ships as
XLSX
Starts with
A pilot batch against your quality bar
Judgement from people who know the domain.
Preference data, rubric grading and written critiques sourced from vetted domain experts, plus multi-turn trajectories for teams training agents.
Inside
Ships as
JSONL
Starts with
A pilot batch against your quality bar
If a company keeps it or a professional makes it, we can probably source it. Tell us what the model needs.
Pick a line, tell us your use case, and we’ll put representative data in front of your team.