xAGI LABS / DATA SERVICES
AI Dataset Sourcing
A sourcing brief built around the model, the task and the coverage you need.
Feasibility, availability and permitted use are established for each project.
Start with a model requirement
Specify the model capability, training or evaluation task, modality, languages and environments. A precise gap is more useful than a request for more data.
Choose the sourcing route
Consider existing licensed assets, newly commissioned collection or a combination. Each route depends on access, source evidence, quality requirements and permitted use. Availability is established per project, not implied by this page.
Scope a representative pilot
Define the sample composition, metadata, formats, rejection rules and acceptance checks before discussing larger quantities. Agree what happens when a source does not meet the brief.