Questions we expect you to ask

It is voice for your whole computer, not one assistant's chat box - and it runs locally. Your voice is processed on your machine, your words go into whatever app has focus, and nothing is uploaded. The cloud assistants are someone else's helper on your computer; this one is yours.

For typing words - that is the mission, and dictation into any app works today: double-tap Ctrl, speak, and the words land where your cursor is. For controlling the machine by voice, we ship it step by step through early access and never claim a step before it works.

Yes - it types at the level of the machine, not inside any one program, so there is no per-app setup and no supported-apps list. If you can put a cursor in it, you can speak into it.

No. Recall keeps months of searchable history; MyQuietShadow's memory holds a rolling 24 hours and overwrites it - forgetting is the default, keeping is your explicit act. Everything stays on your machine, encrypted, with visible indicators and a one-key pause. And it is optional.

It is not a meeting bot, but yes - if you enable ambient audio, your machine's memory includes the meeting you forgot to record. Many places require everyone's consent to record a call: tell people, or pause it (one key). Meeting-aware auto-pause is on the roadmap.

No. Dictation is push-to-talk - the mic engages when you double-tap Ctrl. Ambient listening for transcripts and memory is off until you explicitly enable it during setup, and visible indicators show whenever anything is listening or recording.

Nowhere. Speech recognition, the buffer, clips, and transcripts live on your machine, encrypted. There is no cloud copy and no account required.

Two ways. Dictation: speak your prompts instead of typing them, in any AI app. Memory: instead of typing out context by hand, hand your AI the actual moment - what the screen showed and what was said. Stop typing what your computer already saw.

The buffer is capped: you choose how much disk to give it, and the oldest footage is overwritten to stay inside the cap. Hardware encoding keeps a full day surprisingly small. Exact numbers will be published with early access. And if you never enable the memory, it uses none.

Windows first. Phones are planned - the same helper following you off the desktop. Tell us what you need on the waitlist.

The waitlist is free. Pricing is not final; early-access users will hear it first and get the best terms we ever offer.

In small waves, in waitlist order. Dictation ships first because it already works; voice control follows step by step. Joining the list is the only way in.