xAGI LABS / DATA PROGRAMS
Video and multimodal training data
Define the scenes, actions, sensors and annotations a model needs before planning capture.
Describe the behavior in context
A useful capture brief names the model task, actions, environment and failure conditions. Distinguish a video understanding task from embodied or robotics work that needs synchronized action or sensor information. Specify camera viewpoint, frame rate, lighting, scene variation and any temporal alignment requirements. Ordinary video footage should not be described as robot trajectory data without the required signals.
Plan permissions and exclusion zones
Identify location permissions, participant consent and restrictions around faces, screens, voices, bystanders and confidential objects. Agree what must be excluded at capture time and what can be reviewed or redacted later. A realistic scene is not automatically suitable for training. Access conditions and permitted downstream uses need to be established for each collection.
Evaluate sequence quality
Review a representative set of sequences for task completion, label consistency, temporal coverage, missing frames and metadata alignment. Define acceptance at the clip or episode level rather than relying only on total recording hours. A pilot can reveal whether the requested conditions can be captured reliably before a larger program is commissioned.
A multimodal pilot specification
Illustrative planning template, version 1.0 · 17 October 2026. These examples are not a dataset listing, measured project results or a delivery commitment.
- Signals and alignment
- State which video, audio, sensor or action streams exist; define timestamps, synchronization tolerance and what happens when a stream is missing.
- Annotation unit
- Choose clip, frame, event or episode; document the label schema, time intervals and examples of ambiguous cases.
- Capture boundaries
- Define permitted locations, participants and excluded screens, faces or confidential objects. Agree redaction and review responsibilities.
- Pilot acceptance
- Review representative hard conditions, missing frames, alignment, labels and rights evidence. Proceed to scale only after agreeing acceptance; dates depend on source access and pilot results.
Prepared by the xAGI Labs team for project scoping. A project-specific domain reviewer and acceptance criteria are agreed before collection or labeling.
Start with a useful conversation.
Tell us the model task or source you want to explore. We establish feasibility, quality and rights before discussing delivery.
Book a scoping call ↗