OS Orchestrates Contextual App Actions

The key shift is that the operating system becomes the main decision maker, and apps become specialized pipes for data and actions. In Wafer’s model, a user no longer opens LinkedIn, Uber, or Spotify as separate destinations first. The OS reads context across the device, decides what matters, then calls into apps when needed. That is closer to an API economy than the classic mobile home screen model.

• This matters because today’s assistant integrations are narrow by design. Apple App Intents and Android App Actions let developers expose specific tasks to the system, but only the tasks the developer has explicitly mapped. Wafer is aiming for a deeper layer where the OS can use broader device context before deciding which app action to trigger.
  1 apple 2 android 3 android
• The practical result is less app hopping. Wafer describes flows like comparing ride prices across multiple apps when a meeting appears on the calendar, or learning the exact steps a user takes in Spotify and replaying that pattern later. That turns apps into execution engines, while the OS owns context, ranking, and orchestration.
  4 sacra
• The closest comparables sit below the normal app layer. Granola listens at the microphone layer to capture meetings without joining them. Perplexity Assistant sits in the default assistant slot and can launch actions across apps. Wafer goes one step further by forking Android itself, which is why it frames OEMs, not consumers, as the core customer.
  4 sacra 5 sacra

If this model works, mobile software gets rebuilt around structured data feeds and callable actions instead of polished standalone interfaces. The winners will be operating systems and assistants that can aggregate context, route intent to the right service, and make app choice feel invisible. That would shift power upward from individual apps toward the OS layer and the OEMs that ship it.