We co-design the Aero Hand, the Aero UMI data-collection exoskeleton, and the GAIN3D data engine — turning human demonstrations into zero-embodiment-gap robot data.
The single biggest bottleneck in robotics is the hand — and the high-quality data to train it. We build the hand together with the data engine that captures human work and the data that feeds the foundation model behind a robot that performs on a real production line.
Sub-millimeter accuracy and force-controlled contact, so the hand can take on the tight-tolerance, fragile-part work fixed automation can't.
Production cycle times, not research demos. Every layer of the stack is designed to hit takt time on a real line.
Built for 24/7 uptime. Calibration that holds shift after shift, sensors that survive the factory floor, parts you can swap in minutes.
Engineered for production economics from day one. The price point that makes general-purpose deployment make sense at every cell, not just every plant.
Chestnut is a tightly-knit team of researchers, mechanical engineers, and operators who have shipped robots, semiconductor tools, and consumer hardware at scale. We work in tight loops, in person, on real hardware, and we hire people who do the same.
We prefer one working prototype to ten polished decks. Every Friday we put something new on the line.
The bottleneck for useful robots isn't AI alone. It's hands that can do real work without breaking the bank.
The best dataset is the one collected by the system that will use it. We design data and deployment together.
Robots that never leave the lab don't get better. We deploy early, instrument everything, and compound.
Truly dexterous robotics is held back by two problems: hardware reliable enough to run on a production line, and the high-quality manipulation data needed to post-train a robotic foundation model. We're building the dexterous hand, the Aero UMI data-collection exoskeleton, and the foundation model to break both at once.
18-DOF, sub-mm precision, 0.1 N tactile sensitivity, 20 kg payload, built for 3M+ cycles of real factory duty.
An ergonomic, URDF-matched exoskeleton plus a geometry-aware inpainting algorithm that turns human demonstrations into zero-embodiment-gap robot data.
We build on best-in-class foundation models and make them production-grade with our high-quality, zero-embodiment-gap data.
Tell us a little about yourself and we'll be in touch. We work with customers, investors, researchers, and engineers building toward the same goal.