Rerun is building the data layer for physical AI. Robots, drones, and self-driving cars all run on physical data: multimodal, multi-rate sensor streams tied to time, space, and embodiment. Data infrastructure built for web and business data breaks down under that load.
Rerun’s open source SDK has become a default way robotics teams visualize, log, and query this kind of data. On top of that open source foundation, Rerun is now building Rerun Hub, a commercial data catalog and storage engine, currently in private preview, that extends the same tools to the scale of object storage. The strategy mirrors the open source to commercial platform playbook of Confluent, MongoDB, and Elastic, except Rerun is building the category as well as the company.
Some of the most ambitious teams building physical AI already build on Rerun’s open source tools, including teams at Meta, Google, Hugging Face, and Unitree.