Partnering with Mosaico: Building the Data Infrastructure Powering Physical AI

We’re excited to announce our partnership with Mosaico, following their €1M pre-seed raise led by Connect Ventures, alongside Vento, Eden Ventures, and leading Deep Tech angels. It’s a strong signal that the infrastructure challenges they’re tackling in robotics and Physical AI are both real and urgent.

At JOS TV Plus, we focus on technologies that unlock what’s next—not incremental improvements, but foundational shifts. Mosaico is building exactly that: the missing data infrastructure layer required to take Physical AI from research labs into production.

Robotics is fundamentally a data problem, yet today’s systems still rely on fragile pipelines, massive binary files, and chaotic high-frequency sensor streams. These limitations slow development and make scaling painful. Mosaico addresses this head-on with an open-source-first approach, grounded in the belief that robotics infrastructure must be transparent, extensible, and built in the open.

Since open-sourcing in December, they’ve attracted a fast-growing community—300 contributors in just three weeks—and shipped practical tools like a native ROSbag ↔ LeRobot bridge. Their upcoming release tackles one of the hardest problems in robotics: converting messy, asynchronous ROS data into structured, AI-ready datasets. Powered by the Mosaico Ontology, the system synchronizes multi-rate ROS streams into fixed-frequency, high-fidelity data suitable for Foundation Models and Physical AI agents. With Mosaico Core v1.0 planned for 2026 and an expanding engineering team, the groundwork for production-grade Physical AI is being built now. We’re proud to partner with Mosaico and support a team shaping the future of robotics infrastructure.