Robot Skills as the Next App Ecosystem

An app is only a useful analogy up to a point

A phone app usually owns a screen and a user session. A robot capability acts in a shared physical environment, may move mass, interact with people and must obey hardware and safety constraints. The more useful software unit is therefore not a screen-based app but a skill with a defined contract.

Skills turn embodiment into callable tools

A higher-level agent should not need to command joints or servo loops. It should reason in terms of capabilities exposed by the platform.

navigate(location)
inspect(asset)
pick(object)
place(object, location)
dock(charger)
handover(object, person)
report(exception)

A real skill requires trust metadata

Physical skills need declared hardware compatibility, confidence, failure modes, required sensing, operating envelope and safety conditions. A marketplace that distributes code without validating these boundaries would be far more dangerous than a conventional app store. Simulation evidence, runtime permissions and certification may become part of the platform itself.

This is where software reuse begins to compound

A reliable inspection or manipulation skill can potentially be reused across customers and compatible robot bodies. Reuse turns one-off integration work into software IP. The more that domain knowledge, telemetry and recovery logic are captured in the skill layer, the more difficult that layer becomes to commoditize.

The application economy begins when skills become portable

The decisive moment for Physical AI may not be when humanoids look impressive. It may be when a developer can build a useful physical workflow once and deploy it across multiple compatible machines. That is the point at which robotics starts behaving less like custom automation and more like a software platform economy.

The explicit-skill thesis is weakening, but the interface layer is not disappearing

Confirmed evidence. Universal Robots' Gen 7 platform still exposes an explicit application surface: PolyScope X provides open APIs and an SDK, UR+ distributes validated third-party software and hardware, and the platform includes PolyScope X Smart Skills. At the same time, Reward AI says OM-1 can learn a new task from human demonstration data and run one policy across different robot bodies, while Figure reports that Helix 2.5 performed three long-horizon whole-body behaviors across 30 previously unseen homes without environment-specific data collection or fine-tuning. Reward AI and Figure results are company-reported rather than independently reproduced.

FlyPig interpretation. The counter-thesis is now structural rather than anecdotal. Skild, Reward AI and Figure each point toward general models absorbing more task behavior internally, reducing the need for every capability to be installed as a separate software skill. But production systems still need stable boundaries for permissions, safety, validation, orchestration, billing and compatibility. The more durable primitive may therefore be an external capability contract or tool interface, while a general model generates the underlying behavior dynamically.

FlyPig prediction. FlyPig prediction FP-PAI-004 moves from active to weakening, with confidence falling from 0.60 to 0.56. Explicit skills remain visible in AmbiOS, Qualcomm-NEURA and Universal Robots, so the thesis is not invalidated. The decisive test is whether third-party ecosystems standardize capability interfaces above general models, or whether end-to-end policies make most named skill packages unnecessary.