The boring robot already shipped, and in 2023 it was flipping food. US11744403B2 ("Automated bin system for accepting food items in robotic kitchen workspace") and US11577401B2 ("Modular robotic food preparation system") from Miso Robotics describe the unglamorous opposite of a general-purpose humanoid.
Classified under B25J 11/0045 (robots for specific applications) and B25J 9/1697, these patents win by narrowing. A general humanoid has to handle any kitchen; Miso's robot handles one fryer station, in a fixed workspace, with food arriving through a defined bin system. Every constraint the patents impose is a problem the robot no longer has to solve.
ROI per square foot, not per keynote, and the kitchen math is favorable precisely because it is constrained. A robot bolted over a fry station does one repetitive, hot, undesirable job all shift, in a space designed around it. It does not need to walk, does not need general dexterity, and does not need to understand a novel environment. That is why it deploys while the humanoids pilot.
The honest read is that this is automation, not robotics-as-spectacle, and that is the point. The modular food-prep system patent even mirrors the FedEx modularity thesis — swappable stations for serviceability — because real deployments are won on uptime and maintenance, not on capability demos.
Backlog is the only honest demo. Miso's relevance is not a viral video; it is units installed over fry stations doing a job no human wants in the heat. The patents are the IP behind a deployable, narrow product, and narrow-and-deployed beats general-and-piloted every quarter that the bills come due.
For readers tempted by humanoid hype, the kitchen robot is the corrective. The fastest path to a robot that actually earns money is not a machine that can do everything; it is a machine that does one tedious thing reliably, in a workspace built for it. Miso patented exactly that.