Maxinsights

Maxinsights builds foundational data infrastructure for physical AI and robotics. The company specialises in large-scale multimodal data collection, annotation and curation, and operates as a vendor of physical AI and annotated egocentric data. Its technical work spans hand tracking, upper-body and whole-body motion capture, tactile force signals, simulation, and data evaluation and curation - the inputs required to train Vision-Language-Action (VLA) models and world action models (WAM) for generalist robotics.

The scale of the operation is substantial. Maxinsights produces up to 450,000 hours of video per month through a distributed data collection and processing system that gathers and refines training data across 50,000 different scenes. Collection settings range from homes, kitchens, workshops and labs to retail, warehousing and outdoor environments, giving models exposure to the diversity of real-world conditions they must handle. The company has accumulated 2 million hours of real-world experience data.

Beyond collection, Maxinsights curates, searches and assembles the appropriate mix of data for each model training requirement, supporting robotics foundation models, VLA models and world action models. The company is based in the San Francisco Bay Area. Further information is available at https://maxinsights.ai/.