Software engineering records
What's in it
Repositories with full commit history, linked to tickets, reviews and the chat around them
Typical systems
GitHub or GitLab, Jira or Linear, Slack or Teams
Datasets of real company work, licensed and de-identified.
KetData acquires the rights to company records and builds them into datasets for training and evaluating AI that does real work. Every dataset is de-identified to our standard before delivery and comes with a clean rights chain.
What we build
Operations records
What's in it
How work moves through a company: requests, decisions, exceptions, approvals
Typical systems
Email, chat, support desk, CRM, finance and ERP
AI tool logs
What's in it
Sessions of AI tools used on real tasks, with outcomes
Typical systems
Coding assistants, support chatbots
Every dataset includes
- a manifest (systems, date ranges, volumes, links between records);
- a de-identification certificate (what was removed, how, and how it was tested);
- the licence terms (permitted uses and exclusions);
- a rights record showing the licence chain from the company.
How licensing works
Non-exclusive by default; exclusive terms on request. Delivered as a de-identified copy, so there is no extraction or integration work on your side.
About KetData
KetData is an AI data company. We build licensed, de-identified datasets of real company work for AI training: we acquire the rights from companies, de-identify the records to our standard, and license the datasets to AI developers. Companies keep ownership of their records throughout.