California-based Sunday Robotics is presenting Memo as a practical ally for households, able to clear a table and load a dishwasher in demonstrations that push aside the stereotype of clunky robots. The company says Memo’s capabilities extend to lifting wine glasses, folding socks, and even operating an espresso machine, signaling what its creators call a ‘step change in robotic AI.’
Tony Zhao, co-founder, has emphasized that Memo performed for more than 20 live sessions without breaking a single wine glass. Memo’s power stems from training on real human tasks rather than simulations or teleoperation. Rather than relying solely on synthetic data, the company built a glove-shaped system that captures how people grasp and apply force. Hundreds of people across the United States wear these gloves to record data, including the force needed to lift diverse objects. The hands themselves are Lego-like and designed to resemble, not merely mimic, human touch.
Cost considerations mark a clear departure from traditional teleoperation: a single glove costs about $200, versus tens of thousands of dollars for a teleoperated robot. Sunday Robotics says the glove approach is far more cost-efficient, enabling faster accumulation of practical training data. With more than 500 human-data collectors involved, Memo’s designers argue that relying on teleoperation would have delayed progress by decades.
The company insists the glove system lets Memo learn from real-world handling of everyday items. These technical strides bring home-robotic assistance closer to reality, but they also raise questions about privacy, safety, and expectations. While Memo has demonstrated reliability in controlled demos, broader adoption will depend on how well it handles the unpredictable variety of objects and layouts inside actual homes. Still, Memo’s team argues that this approach could accelerate a new era of capable, affordable domestic robots.
