Sunday's Managed Home Data Flywheel

The moat here is not just better robot hardware, it is faster accumulation of messy home data across many real apartments. Sunday is training Memo through Skill Capture Gloves used in 500 plus households, then turning those human demonstrations into robot actions with about 90% fidelity. That means every new kitchen, bedroom, and storage closet expands the model’s map of real edge cases before later entrants can collect an equivalent distribution.

• Managed apartments and short term rentals matter because they can supply many similar but not identical homes under one operator. That gives Sunday paid deployments and a steady stream of repeat chores, layouts, and failure cases, which is more useful than a small number of showcase demos.
  1 sacra 2 sacra
• The main alternative path is teleoperation. 1X sells NEO with scheduled remote expert guidance for tasks it does not yet know, and Figure is pursuing large scale residential video collection through Brookfield. Sunday is different because it can gather training data before full robot rollout, using cheaper wearables instead of putting expensive robots in every home.
  1 sacra 3 1x 4 sacra
• This is why data volume can outrun architecture elegance. 1X is betting on a humanoid form and a world model to generalize across homes, while Sunday is betting that more demonstrations across more households will teach a simpler robot the highest frequency chores first. In home robotics, the broader dataset often decides what actually works in production.
  1 sacra 5 sacra

The next phase is a race to lock in distribution before consumer demand fully opens. If Sunday can turn property portfolios into recurring training grounds, its model should improve on the exact tasks that matter most, and each deployment will make the next one easier to sell and harder for newcomers to match.