A draft architecture brief for the WHY AI IDE private beta.
What the brief covers
- The learning signal pipeline: capture · curate · apply
- Local vs cloud personalization tradeoffs
- The four classes of signal we track (decisions, style, patterns, refusals)
- The control surface: every signal is editable, exportable, deletable
The lesson (preview)
Cursor reads your code. A private-first assistant is designed to learn how you think - and to let you delete that learning when you want to. The architectural difference is small. The product difference is large.