Reading material for teams working on documented AI data. Written for the people who actually have to sign off — not for the conference circuit.
Drag, scroll, or use the catalog below. Each piece is tagged by audience so you can skip what isn't for you.
Guide
A look at the documentation layer that turns synthetic data from a file into a product — what gets checked, what doesn't, and why the dossier is the deliverable.
Template
Reference fields for documenting origin, generation method, quality, ethical posture and legal status — the structure waiser dossiers follow.
Explainer
Why downloadable does not mean usable, and what a licence should actually say — written for the legal reviewer who has to sign off, not for the model trainer.
Use case
How teams use synthetic complaints and rights-request data to evaluate AI systems against regulatory patterns without exposing real customer records.
Checklist
The evidence we look for at initial screening — so your first submission lands in 'move to review' instead of 'we need more'.
Walkthrough
A guided tour of a real dossier — what every section means, where to find the licence posture, and which fields settle the 'can we use this for X?' question fastest.
Pick the role you're playing today. Each path stays under thirty minutes of reading.
Start with 'Reading a waiser dossier', then 'Permitted use vs. raw availability'. You'll know how to triage a listing in under five minutes.
Start with the 'Provider checklist before submission', then 'Documentation templates'. You'll arrive at screening with the evidence we actually need.
Start with 'Permitted use vs. raw availability', then 'What makes a synthetic dataset verified?'. Both are written for the reviewer who has to sign off.
If the library is missing something useful for your role, tell us. We'd rather publish what teams actually need than what looks good on a content calendar.