Origami Robotics launched S1 for teams collecting human demonstrations to train dexterous robot hands. The glove records 20 finger joint angles with encoders, pressure at the fingertips and palm, and video from a wrist-mounted global-shutter camera, all on one timeline. It arrives as labs try to scale those hours, and it targets what video leaves out: fingers hidden by an object or the other hand, and pressure that changes while the pose barely moves. S1 was designed with Origami's robot hand and shares its joint arrangement. After fitting and calibration, each measured angle is a position target for the matching robot joint, so those finger movements do not go through a geometric retargeting solver. Recordings from the launch include violin, piano, and percussion, along with card shuffling, fast Rubik's Cube solves, and butterfly-knife tricks, used to check that practiced motion still fits in the glove. Origami reports millimeter-level fingertip accuracy for the index, middle, and ring fingers against OptiTrack in one recording, measured in the glove's wrist frame, and says the encoders kept measuring when fingers were fully out of camera view. Co-founders Quanting (Daniel) Xie, who worked on manipulation as a PhD student at Carnegie Mellon's Robotics Institute, and his brother Ryan Xie are building the hardware in San Francisco as a Y Combinator Winter 2026 company. An earlier rigid glove, deployed in a couple of factories seven months before this launch, fit only some workers. The current soft shell is meant for a wider range of hands, with each wearer still fitted and calibrated before recording. Origami says mass production has started and is opening S1 to a small group of early beta testers. Hand size still varies, the robot fingertips have different shapes, and the company says a learned model for that contact gap is still in progress. The performances document what a person can do in the glove. They do not show the robot reproducing those skills.