‍

🤖 🤖 🤖 🤖 🤖 🤖 🤖 🤖 🤖

The broken infrastructure of robotics data

After spending time with founders and engineers who are building robotics and physical AI companies, one issue comes up again and again: data infrastructure is broken. And it’s slowing the whole industry down.

Robotics systems generate huge volumes of messy, multi-sensor data: cameras, LiDAR, radar, IMUs. Most teams still rely on DIY pipelines that are fragile, hard to debug, and almost impossible to audit. Tracing where a data point came from, or reproducing a result, is painful… Until now.

Introducing Mosaico

We’re excited to announce our investment in Mosaico, an open-source, next-generation data management platform built specifically for robotics and physical AI. We are excited to have led the pre-seed, alongside Vento and high-value Deep Tech angels.

We believe Mosaico is building a critical and sticky layer at the heart of the physical AI stack. Once this layer exists, everything else accelerates. Mosaico is built from the ground up for robotics data. It handles large-scale, multi-sensor data, making it easily repeatable: Simplify testing and debugging by recreating the exact data conditions of any past event. The goal is simple - let dev teams focus on autonomy, not infrastructure.

Product founders have deep insights and are opinionated about the problem

Gabriele (CEO), Federico (CTO), and Francesco (CSO) spent over 12 years building autonomous driving systems between Parma and Silicon Valley, from VisLab to Ambarella. They worked on real-world autonomy, at scale, with petabyte-sized, high-frequency sensor data. They lived and breathed the problem daily; endless DIY data pipelines, painful debugging, time taken away from algorithms. Mosaico is the platform they wish had always existed.

The Parmesan connection

The Mosaico team is based in Parma and Bologna - the corner of Italy where I am from, which is famous for its cheese 🧀 and ham 🍖. But from now on, thanks to Mosaico, this region will also be home to robotics 🤖

Traction

Mosaico launched successfully on GitHub on 15 December (300+ stars, 100+ clones). Mosaico is already working with 15+ organisations globally, from startups to large enterprises across industrial, deep tech, and defence—confirming the problem is horizontal and universal: time-series data from heterogeneous sensors needs a shared infrastructure, not vertical, fragmented solutions.

What’s next?

We are proud to partner with Gabriele, Federico, and Francesco as they execute their mission to unlock the next wave of innovation in physical AI to become the standard data management layer for every robotics company in the world

Let's get physical.