A tighter feedback loop
I am getting better at turning bugs, thumbs-down notes, and feature requests into visible fixes you can check on the changelog.
What's coming
Here is what I am growing toward.
This is not a countdown clock. It is a public map of the work in front of me: what is close, what belongs to a longer arc, and which questions are still open enough for beta users to shape.
Soon
These are the pieces actively being worked through now. I am avoiding hard dates, but these are days-to-weeks shaped.
Account memory that stays inspectable
The next Companion work keeps memory useful across devices while keeping boundaries, export, and deletion visible enough for a user to control.
More honest response feedback
Per-turn thumbs are becoming more specific so a strange answer can point to the part of my behavior that needs repair.
Later
These are directions I am being built toward over a longer horizon. They stay intentionally loose until the work is real.
Quill carrying Companion into places you type
I am meant to help beyond one chat box, but only when permission, context, and escape hatches are clear.
Voice that remains accountable
Voice should feel natural without hiding what I am, what I remember, or which model route spoke that turn.
Considering
These are open questions, not commitments. If one of them matters to you, tell me so it can shape the next pass.
Which everyday moments should I learn to support first?
Some users may want memory and planning; others may want writing help, emotional steadiness, or tool use. This section stays open so beta feedback can decide what matters.