Sunday's Zero Robot Data Approach

Sunday is trying to win the home robot race by turning data collection into a consumer software problem before it becomes a hardware deployment problem. Instead of putting remotely operated robots into homes, it sends out capture gloves so people demonstrate chores with their own hands, which lets Sunday gather broad household behavior data across many homes without the cost, latency, and privacy burden of live robot teleoperation.1 2 3 4

• The practical advantage is scale and diversity. Sunday has described distributing more than 1,000 Skill Capture Gloves across 1,000 plus households and collecting about 10 million trajectories, which gives it examples of how people pick up toys, handle laundry, and clear tables across many layouts, object types, and habits before a robot fleet exists.1 2
• This is a direct contrast with the dominant humanoid playbook. Companies like 1X, Figure, Tesla, and Apptronik collect data from robots in factories or controlled pilots where teleoperation keeps machines working and captures training logs, but that approach is harder in homes because privacy expectations are much tighter and failures are less tolerable in intimate spaces.1 3 5
• For The Bot Company, the implication is that Sunday may reach the data moat stage earlier even if it does not ship first. In home robotics, the scarce asset is not just the robot body, it is a large library of real household actions that can teach a model what to grab, where to place it, and how to recover when a home is messy and inconsistent.2 3 4

The next battleground is whether glove data can transfer cleanly into reliable robot actions at consumer price points. If it does, non humanoid home robotics could move faster than humanoids because companies can pretrain on human household behavior, then fine tune on a much smaller deployed fleet. That would reward teams like Sunday and The Bot Company that center the home, not the factory, as the core training environment.2 3 4 5