Datayield helps companies license their operational data to AI labs. Your accounting ledgers, CRM history, support tickets and project trackers hold real records of how work gets done, and labs pay for that because they cannot generate it themselves. We check that you have the rights to sell, remove anything that identifies a person, find buyers, and pay you each time a license pays.
These docs are for three groups of readers:
- Sellers who want to know what happens to their data, what they need to do, and how they get paid.
- Engineers on the seller side who install the SDK, connect it to their systems and schedule it.
- AI labs who license datasets and want to know the formats, terms and evidence that come with each one.
What we sell
Every dataset comes from one connected system and is sold as one of three products.
| Product | What a buyer receives |
|---|---|
| Records | De-identified rows from one system, such as invoices, deals, tickets or samples. |
| Episodes | Ordered sequences of a situation, the action a person took, and the outcome, built from audit trails and change histories. |
| Environment | A working replica of a system: its structure plus synthetic data drawn from the real distributions, for agents to practice in. |
Records and episodes come out of the SDK. The SDK does not build environments, so ask us if you want to offer one.
How the pieces fit
- You apply, we sign a mutual NDA, and we review your contracts for the right to sell. See How it works.
- You send a sample, and we return a readiness report that shows what is sellable and what has to go. See Readiness report.
- When a buyer is interested, you install the SDK on a machine you control. It pulls records from your systems, scrubs them on that machine and uploads only the scrubbed output. See Quickstart for sellers.
- The first license sells your history. After that, the SDK sends new records every month, and each license with a refresh term pays for each batch. See How recurring revenue works.
Where to start
- New seller: read How it works, then What leaves your machine.
- Engineer setting up the SDK: go to Install and then the page for your connector.
- Lab evaluating a dataset: start with Data formats and Licensing terms.
- Building against the API directly: see the API reference.