Start with one layer or build the whole program.
Tell us what you are building. We scope the data, experts and platform work it needs, let you review samples first, and deliver with the records your legal and security teams will ask for.
Four steps, agreed in writing.
Scope
You share the model or system, the data and expertise it needs, and your constraints.
Review
You assess samples, documentation and a proposed plan before committing.
Agree
License scope, permitted uses, responsibilities and limits go into a written agreement.
Deliver and run
We deliver data and expert work, and support deployment and governance where in scope.
What to bring
- The intended use, including commercial use.
- Which layers you need: data, experts, platform, or all three.
- Modalities, languages or regions, and rough volume.
- Your legal, privacy and security requirements.
- Your timeline.
What you receive
- Datasets and expert work matched to the specification.
- License terms and permitted uses.
- Provenance records: source, consent basis, processing, version.
- Review results against the specification, on request.
Common questions.
Can we train commercial models on this data?
Permitted uses are written into each license. We scope the dataset to the use you describe, and we tell you if a source does not allow it.
Can the source organization be named?
Only if the owner has agreed. We never name an owner without permission.
Can we review before we buy?
Yes. We expect buyers to review samples and documentation first. The exact process is agreed per engagement.
Can the platform run inside our environment?
Deployment options are agreed per engagement.
How is personal data handled?
Digital data is anonymized before delivery, and the method is documented per dataset. Physical data comes from contributors who have agreed in writing to how recordings are used.
What if a source withdraws consent?
Terms for withdrawal are set in each agreement.