The platform introduces a suite of tools aimed at automating the lifecycle of data used in physical AI, where sensor streams and video recordings present unique storage and labeling challenges. Central to this update is the Voxel51 Agent, a system capable of chaining over 45 tools to perform complex tasks like automated labeling and dataset curation. By integrating a hybrid search index and synchronized time-series visualization, the company seeks to reduce the manual labor that currently consumes most of a development team's project time.

Beyond internal tooling, Voxel51 is partnering with Nebius to address the infrastructure gap between raw data and computational power. By combining Voxel51’s data stack with Nebius’s cloud environment, the companies intend to streamline the development cycle for firms like Ford and Porsche. This collaboration focuses on active learning loops, where every failure or edge case encountered in the field can be systematically fed back into model training.