For data partners
License your data for AI, on your terms.
Publishers, archives, production companies and software businesses hold material AI developers need. We assess what can be licensed, find the right buyer and structure a licence that protects you.
Who may be a suitable partner
The strongest candidates hold substantial archives they control. Size helps, but depth, context and clear authority to license matter more.
Publishers and archive owners
Specialist, academic and trade publishers, learned societies and archives that hold the digital rights to their content.
Production companies and footage owners
Producers, broadcasters, educational publishers and specialist operators with unused rushes or archive footage.
Software and technology companies
Teams with long histories of code, issue tracking, code review discussion, incident write-ups and documentation.
Companies closing or restructuring
Directors, administrators and liquidators looking to realise value from data assets, with the right checks done first.
Data AI developers look for
Typical sources, and what tends to make them useful. This is a guide, not a checklist.
| Area | Examples | What tends to make it useful |
|---|---|---|
| Publishing and text | Books, journals, trade press, specialist archives | Specialist depth, clean structure and a clear rights chain per title |
| Video and footage | Unused rushes, documentary and educational archives, drone footage | Original quality, shot metadata and people, places and music that can be cleared |
| Software and engineering | Codebases, issues, reviews, incident records | Discussion linked to changes, tests and outcomes |
| Workplace records | Team chat, email, tickets, project histories | Complete threads tied to real work, where employees' and clients' rights allow |
| Customer support | Tickets, chat transcripts, help-desk histories | Resolution paths, categories and outcomes over time |
| Documentation | Manuals, wikis, policies, runbooks | Well-maintained documents with revision history |
How a preliminary assessment works
Before anyone touches real records, we look at whether a dataset could plausibly be useful and licensable. That judgement rests on five factors, and every one of them has to hold.
Permissions are assessed with your team and your advisers: who controls the data, what employees and customers were told, which contracts apply, and which uses could be permitted.
Quality
Is the data coherent, work-focused and reasonably complete?
Context
Does it keep the structure that makes it useful, such as metadata, sequences, links and outcomes?
Volume and history
Is there enough of it, over a long enough period, to be meaningful?
Rights
Does your organisation have the authority to license it, and for which uses?
Intended use
Does it match what AI developers are currently looking for?
What happens after you get in touch
You can stop at any stage. Nothing is shared or licensed without your agreement.
Enquiry
You tell us about your organisation and the kinds of data you hold. No data is shared at this stage.
Introductory call
We discuss your systems, history and goals, and whether there is a realistic fit with current demand.
Preliminary assessment
We work from descriptions, schemas and volumes, under confidentiality terms where needed, rather than raw records.
Rights and permitted use
Together with your advisers, we review authority, notices, contracts and the uses that could be permitted.
Human review
Commercial terms
If the data is suitable, we discuss scope, licence structure and compensation.
Preparation and licence
The agreed dataset is prepared, checked with you and licensed under defined terms.
Commercial arrangements
Where a dataset is suitable, we discuss terms that reflect its scope, quality and the uses being licensed. Possible structures include a one-off licence fee, a share of licence revenue or an ongoing supply arrangement for organisations that generate relevant data continuously.
We don't quote figures before an assessment, and an enquiry doesn't guarantee acceptance or payment. What we can promise is a clear explanation of how any proposed arrangement works.
Please don't send raw data. The enquiry form below asks for a description only. Never upload employee messages, emails, customer records or other confidential material through this website.
Common questions
Do we need to share any data to start a conversation?
No. Early conversations rely on descriptions of your systems, data types and volumes. If a closer look is needed, we agree the scope and confidentiality terms first, and work from schemas or samples you approve rather than raw records.
Who decides whether our data can be licensed?
You do, with your own advisers. We help structure the assessment, covering authority, contracts with authors and contributors, notices given to employees and customers, other contractual restrictions and the legal basis for any new use, but the decision to proceed is always yours.
Does de-identification remove our legal obligations?
Not necessarily. De-identified data can still be personal data, for example under UK GDPR, depending on the residual risk of re-identification. We assess that risk for each dataset rather than treating it as solved.
Can we control how the data is used?
Yes. Licences define permitted uses and restrictions, such as evaluation only, no redistribution or no attempts at re-identification. You agree those terms before anything is released.
How are partners compensated?
It depends on the dataset and the arrangement. Possible structures include a one-off licence fee, a share of licence revenue or ongoing supply terms. We don't quote figures before an assessment, and we can't guarantee that any dataset will be licensed.
We are winding down a company. Can its data be licensed?
Often, in part. Codebases, documentation and other company-owned material are usually the simplest assets. Communications involving employees, customers or suppliers need a rights and data protection review first. We work alongside directors, administrators or liquidators and their advisers.
What if our data isn't suitable?
We'll tell you, and explain why where we can. Suitability depends on rights, quality and current demand, so a dataset that isn't a fit today may become one later.
Explore a partnership
Tell us about your organisation and the data you hold, at a high level. Every enquiry is read by a member of our team.
- Takes around five minutes
- No data is shared through this form
- Submitting doesn't commit you to anything