Corpus
Records
De-identified ledgers, lab results, process data and tickets from one field, covering years and many companies.
Used for Pretraining and domain adaptation
For AI labs and data teams
Business data you can't scrape.
We supply real records from accounting firms, labs, chemistry and chip companies. Every dataset is de-identified, cleared for licensing and documented.
- Rights reviewed for every seller
- De-identification report included
- Sample before you buy
de-identifying · 18,420 records
Your systems
- Xero
- ELN
- LIMS
- MES
- ERP
Hamster · Brief · BRF-1182
raw exportAna Ruiz chose webhook retries over polling · 6 tasks → PR #4417 · merged, rolled back once · for Brightline Health · STRIPE_KEY=sk_live_51H…
- 4 identifiers replaced
- 1 secret removed
- Re-identification check passed
- Waiting for your approval
Packaged for labs
Records62%
Episodes38%
Environments18%
Estimated value
$165,780
Sample records. You review every dataset before it's offered.
What we supply
Choose the format you need.
State
Action
Outcome
Situation, action, outcome
Episodes
Step-by-step records rebuilt from audit logs, showing what a person saw, what they did in the software and what happened next.
Used for Post-training and agent RL
Working copies
Environments
Copies of business software built from real database structures and filled with realistic synthetic data, so agents can practice in them.
Used for Agent training and evaluation
Included
What comes with every dataset
✓
Rights statement
A signed statement that the seller can license the data, backed by our contract review.
✓
De-identification report
Every change we made, counted by type, with notes on remaining risk.
✓
Re-identification testing
We check that no record can be singled out from a combination of fields and remove any that can.
✓
Source and field notes
The source system, version and date range, and a description of every field.
✓
Sample
A representative sample to evaluate before you commit.
✓
License terms
Permitted uses, length, exclusivity and deletion, agreed before you buy.
Request data
Tell us what data you need.
We look for sellers based on real requests. Tell us the field, format and timing, and we'll reply with what's available and a plan for a sample.
Questions
What people ask us
Established mid-size companies in accounting, chemistry, life sciences, semiconductors and manufacturing. Each one signs an agreement confirming it can license the data, after we review its customer contracts.