Sourcing and delivery

We source licensed training data for AI teams.
We do not operate a public marketplace.
Start by sending a data request: /source-data/.

1. scope definition

Client provides

• Model requirements (modality, language, dialect).
• Intended use (training vs. retrieval).
• Scale targets and refresh cadence.

You receive

• A scoped brief that lists required data attributes.
• A checklist of deliverables for the next stage.

2. data inventory & sample

Client provides

• An inventory of available assets (speech, text, video).
• Preliminary metadata (size, source, format).

You receive

• A curated sample package that reflects the inventory.
• Data dictionary and initial quality notes.

3. rights verification

Client provides

• Documentation of ownership, licenses, and consents.
• Any existing rights registers.

You receive

• A rights register for each data group.
• A list of any gaps that must be resolved before delivery.

4. quality and suitability assessment

Client provides

• Access to the sample under agreed security terms.

You receive

• Objective quality metrics (signal‑to‑noise, transcription accuracy, annotation completeness).
• A suitability report that maps data to your model requirements.

5. delivery package

Client provides

• Final acceptance of the rights and quality reports.

You receive

• Fully licensed data packaged per your specifications (audio, video, or text).
• Documentation of usage rights, refresh schedule, and support contacts.

6. ongoing refresh & support

Client provides

• Updated model requirements or new language targets.

You receive

• A plan for periodic data refreshes.
• Continued rights monitoring and compliance updates.

For more detail on each service, see our offerings: /specifications/speech-and-audio-data/, /specifications/video-training-data/, /specifications/domain-specific-text-data/.

Explore industry insights at /guides/.