Request a dataset
Tell us the data your model needs.
Describe the dataset you are looking for. We identify and approach organizations that generate matching data — including data that isn't packaged as a product yet. Specifications are kept confidential.
Request received.
Thank you. We'll review your specification and follow up to clarify requirements before we start sourcing. We reply within 24 hours. Your request is kept confidential.
What should a dataset request include?
You do not need every detail. A clear goal, and the outcome your model must predict or check, is enough to start. These points help us search faster:
- The domain and the task, such as defect detection or claim review.
- Record types and formats: tables, images, documents or signals.
- Scale and time span: how many records, over how many years.
- The labels or verified outcomes you need.
- Intended use: training, fine-tuning or evaluation.
- The rights you need: exclusive or not, duration and field of use.
What happens after you send it?
If you want to buy AI training data that does not exist as a product yet, this form is the place to start. Not every request can be met; if we cannot find a fit, we tell you.
- 01We reply within 24 hours to confirm your request and ask any open questions.
- 02We work out which kinds of organizations are likely to hold matching data.
- 03We approach data owners confidentially. Your specification stays confidential.
- 04If an owner can license matching data, you review descriptions and documentation first. Samples may follow under confidentiality terms.
- 05Price and scope are agreed in a license between you and the data owner.