Personal AI is the direction we’re working toward. Memory, voice, privacy and user control are the starting points. This is a note about the questions behind KITTY AI, not a description of a finished release.
Useful memory
An assistant that forgets every conversation makes you repeat yourself. An assistant that remembers everything without a clear way to review it creates a different problem.
We’re interested in the space between those extremes: context that is useful, understandable and manageable. What should be remembered? How would you correct it? What should happen when you ask it to forget?
Voice without ceremony
We want voice interaction to feel less like filling in a form. That means considering continuity, interruptions and how an assistant makes its state clear.
Those are development questions. A smooth conversation in a demonstration isn’t enough to establish how it will behave in everyday use.
Control before claims
Privacy is part of the design direction. We’re considering local processing and explicit controls, but we’re not claiming an unreleased product has solved those problems.
As the work develops, we’ll share real interfaces, demonstrations and the decisions behind them. There isn’t a public release date yet.