Maxinsights is a data vendor for physical AI and robotics, providing large-scale multimodal collection, motion capture, tactile sensing, and annotated egocentric data for VLA and world action model training.
Maxinsights builds data infrastructure for physical AI and robotics. The company operates as a vendor of large-scale multimodal and annotated egocentric data, covering collection, processing, curation, and evaluation for teams training Vision-Language-Action (VLA) models and world action models (WAM).
On the collection side, the multimodal system combines hand tracking, upper-body and whole-body motion capture with tactile force signal capture, alongside egocentric video. Stated capacity is up to 450,000 hours of video per month across roughly 50,000 scenes, with about 2 million hours of accumulated real-world experience data. Scene coverage spans domestic environments, kitchens, workshops, labs, retail, warehouses, and outdoor settings. Curation and search tooling is used to assemble data mixes matched to specific training requirements, and the company also offers simulation and data evaluation services.
Work is based in the San Francisco Bay Area. Roles touching the collection rigs, motion capture and tactile sensing pipelines, annotation tooling, or data evaluation and curation systems sit at the hardware/software boundary of a data stack built for generalist robotics training.
