Specialists Win Early Deployments
Figure AI
Narrow starting tasks are the fastest way to turn a humanoid robot from a demo into a usable labor product. In Agility's case, bin to conveyor work lets Digit use mitt like grippers, a 35 pound payload, and software tied tightly to warehouse systems, so customers can validate ROI on one repeatable job. Figure is making the opposite bet, building broader hands, language control, and household capable behavior that can widen use cases over time but takes longer to prove in production.
-
Agility's workflow is unusually concrete. Digit picks totes from shelves, walks them to conveyors, docks itself for charging, and is managed through the Arc cloud platform. That specificity is why it has already logged 100,000 plus tote handling cycles in live logistics deployments.
-
The tradeoff is market width. Agility can stretch from tote moving into palletizing, truck loading, line feeding, and returns sorting, but those are still nearby warehouse and factory motions. Figure and Apptronik are both designed to cross more environments, from automotive plants to homes or healthcare, because their product roadmaps start from broader task flexibility.
-
Apptronik sits between the two models. Apollo is sold as a general humanoid, but its modular body lets the same upper unit sit on legs, wheels, or pedestals for different jobs. That preserves more market scope than a single fixed workflow, while still giving customers a more optimized setup than a one shape fits all robot.
The market is likely to keep splitting into specialists and generalists. Specialists should win the first paid deployments because customers buy a worker for one painful task, not a vision of future versatility. Generalists should capture more of the long term value if they can turn early factory data into robots that reliably move across many workflows without custom redesign each time.