Forget personal assistants, The real battle is for voice UIs
Stacey Higginbotham

  1. Chat (text or voice) is just another user interface surface. Like GUIs, gestures, game controllers, and the rest.
  2. The fight for power starts at the surface. Your deconstruction of the architecture is right: for realtime experiences, especially aural ones, “don’t make me think” and the other sensory and cognitive limits force the marketplace to converge, to consolidate.
  3. Each surface fronts a platform. The engines adding smarts to chat UIs are starting as standalone services. Some, perhaps most, will offer themselves through more than one chat surface. One of the next struggles is for each chat surface to become the developers’ platform of choice. Does your platform offer more users? Better retention? A fairer shot at being discovered? More money?
  4. Agents will cooperate with each other. “Get me a sandwich from that place Harry wrote about last week.” You shouldn't know how many dozens of subsystems and data sets were orchestrated to get you your corned beef on rye. You shouldn't care. But new IfThisThenThat-style recipes will be crafted on the fly to get you your sandwich. And split the tip.
  5. Chat UI Opacity Needs Trust. We won’t see all the duck feet paddling furiously under the surface. We won’t know which companies or government agencies comprise the data and services that power our chat surfaces. Despite that ignorance, we will have to trust the Amazon or Google or Tencent that hosts our chat interfaces across our devices.
  6. Cortana will be our hostess. As our chat UIs become rich and powerful and flexible, they’ll self-facet personas. Cortana will don a separate personality just for body issues, with a tone and privacy sensibility suited for health, diet, fitness, and clinical self-care. Cortana or Siri or Home will remain front and center, but expect them to be ringmaster for a handful of regular personalities and a larger cast of guest personae.