Sunday Robotics Unveils Memo: A Household Robot Trained on 10 Million Real-World Chores

Revolutionary Training Methodology

Sunday Robotics has developed an innovative approach to training household robots that diverges significantly from traditional methods. The company employs a network of over 500 human data collectors across the United States who wear specially designed $400 sensor-equipped gloves that mirror the movements of Memo's mechanical hands 3

3

4

4

. This glove-based system captures direct measurements of human handling strategies, including grip strength, finger placement, and motion trajectories as operators perform real household chores.

The data acquisition process feeds into Memo's training pipeline, which combines glove telemetry, vision input, and proprioceptive sensor readings to refine manipulation models 1

1

. This methodology has enabled the company to compile a dataset of approximately 10 million genuine household routines, representing massive data diversity that allows Memo to adapt to unpredictable home environments 4

4

.

Technical Design and Capabilities

Memo's physical design deliberately diverges from the current trend toward humanoid robots. The machine, roughly the size of a compact dishwasher, features a wheeled base for mobility and balance, eliminating the complex balance modeling required for legged robots 1

1

2

2

. The robot includes two articulated arms with pincer-like grippers, a height-adjustable central column, and a display featuring a cartoon face topped with a stylized red cap.

Demonstrations show Memo performing complex household tasks with measured precision. In coffee-making demonstrations, the robot positions itself at countertops, fills portafilters with ground coffee, tamps using force-control algorithms, and delivers finished espresso 1

1

. The robot has successfully lifted two wine glasses with one hand, folded socks, cleared dinner tables, and loaded dishwashers without breaking delicate items during over 20 live demonstration sessions 3

3

.

Founding Team and Investment

Sunday Robotics was founded by Stanford Ph.D. roboticists Tony Zhao (CEO) and Cheng Chi (CTO), who started the company in a garage working around the clock with 3D printers to develop the robot's hardware and appearance 4

4

. The founding team brings experience from Tesla and Google DeepMind, with their previous research including ALOHA (a low-cost open-source hardware system) and UMI (universal manipulation interface) frameworks that laid groundwork for the skill capture glove technology.

The company launched with significant financial backing, securing $35 million in funding from notable investors including Benchmark and Conviction 4

4

. Sarah Guo, founder of Conviction, highlighted the team's blend of industry veterans, while Benchmark's Eric Vishria emphasized Sunday's focus on practical, real-world deployment over demonstration-focused robotics.

Market Position and Beta Testing

Sunday Robotics differentiates itself in the competitive domestic robotics landscape through vertical integration, developing both hardware and AI models required for nuanced control in ever-changing environments 1

1

. The company competes with startups like Physical Intelligence, Skild, and Generalist, which pursue flexible training approaches, and 1x, which has introduced teleoperation-assisted humanoid home robots.

The company opened applications for its Founding Family Beta program on November 19, 2025, with plans to select 50 households as early adopters who will receive numbered units and direct support 2

2

4

4

. This pilot phase, scheduled for late 2026, will provide critical evaluation of technical reliability and user satisfaction as robots navigate real-world challenges including children, pets, clutter, and incomplete household instructions.