Skild AI

Proud to partner with Skild AI to deploy omni-bodied AI in real factory environments. Check out a live demo of UR arms powered by Skild Brain assembling GPU racks at #NVIDIAGTC 🦾 Read more on the collaboration here https://lnkd.in/gW-2Fa9c

We’re helping define the factories of the future - Reindustrial Revolution! Really excited about this partnership with ABB Robotics, Universal Robots, and NVIDIA Robotics. It is a big deal for us for two reasons: 1. AI-first reinvention of automation Unlike hand-engineered classical automation: - We deploy the Skild Brain (fully end-to-end neural network), fine-tuned with minimal robot data, and at times, none at all (more on this soon) - The system is inherently robust to real-world disturbances and can be set up for entirely new tasks in a fraction of the time. - In-context memory enables true long-horizon execution. No brittle, step-by-step programming required. - No custom tuning, no big metallic cages, no fancy sensors, just off-the-shelf robots and cameras. Why doesn’t classical automation scale? - Traditional systems rely on custom hardware and painstaking manual engineering, often costing multiples of the robot itself. Even then, they are fragile, and failures emerge if the environment shifts by as little as 0.1 mm. In dynamic & mixed assembly lines with humans and robots, that level of precision is unrealistic. - In mixed production lines, disturbances are inevitable and prohibitively expensive to eliminate. Classical approaches break under this variability unless you over-engineer the environment and spend lots of money (design for automation - DFA, rigid enclosures, heavy sensorization) to make sure everything is 0.1mm level precise, which defeats the point if the setup is going to change quickly. E.g., NVIDIA changes GPU designs every 6months! Long-term vision Imagine a world where anyone with no expertise in controls (a factory worker, a technician, a small business owner) can automate complex factory stations in days. No specialized infrastructure. Just intelligence that adapts. This is more than incremental progress. This is going to revolutionize traditional automation, what we call the reindustrial revolution, where automation becomes accessible, flexible, and universal! 2. Data flywheel for Physical AI Industries are the backbone of society and an ideal proving ground for early AI deployments. They offer just enough structure to operate, and real, immediate demand to drive impact. Every deployment across tasks and sectors generates data that improves the omni-bodied Skild Brain, thus fueling a compounding loop where each improvement unlocks the next frontier. This is how we aim to build the largest data flywheel for physical AI. Industrial deployment is only the beginning. It opens the door to semi-structured environments (hospitals, hotels, grocery stores, etc.) where complexity rises, and capability deepens. From there, we move into fully unstructured consumer settings (homes). Step by step, this progression lays the foundation for something much bigger: the emergence of true general autonomy.