p03 · cloud canvas

bubblmind — how it works

touch
double-tap
new bubble — then speak or type it
hold (empty)
talk to the AI (release to keep listening; keep holding = push-to-talk)
hold (a bubble)
talk to the AI about that bubble
drag a bubble
move it (positions are yours — never moved for you)
lasso a loop
quick release → make a scope · hold → ask the AI about them
zigzag ×2
through a bubble pops it · through a membrane dissolves it
drop across a membrane
moves a bubble in or out of that scope
double-tap+drag a bubble
pull a link to another (or speak a child)
2 fingers
pan · pinch-zoom · twist-rotate · 2-finger tap = menu
drag a scope's name
move the whole scope; hold it to talk to the scope
mouse
right-drag
pan the canvas · right-click (no drag) = menu
wheel
zoom · shift+wheel rotates · trackpad two-finger pans, pinch zooms
anywhere it listens
type instead
a physical keyboard always works — Enter confirms, Esc cancels
⌖ / 1×
fit everything into view · reset the view
⚙ → BYO key
your own Gemini key unlocks unlimited AI, stored on this device
the idea

Bubbles hold thoughts; scopes (membranes) group them; the canvas is a markdown file underneath. Ask the AI to create, connect, or tidy — it acts only when you ask.

projects

voice pipelines

Voice output, dictation, and Gemini Live all run on your Gemini API key (🔑) — Cloud STT needs the Speech-to-Text API and a Cloud key. Live mode: hold empty space to start, tap empty space to end; canvas ops are disabled during live (v1).

name