Note.space ships an opt-in writing assistant. It is off by default. Nothing you type leaves the app until you turn it on under Settings → Preferences → Writing assistant.
What it does
The assistant lives behind the small pill at the bottom-right of the editor. Open it to:
- ask questions about the note you're in
- draft new sections from a brief
- summarise long material
- rewrite a paragraph in a different tone or shorter form
- run free-form chat against a chosen model
It is a companion, not a co-pilot. There is no inline ghost text and no auto-complete in the writing surface.
What gets sent
When the assistant is enabled and you send a message:
- the message text is always sent to the selected model provider
- the body of the current note is sent only when you explicitly attach it (via the Attach toggle) or ask the assistant to read or summarise it
- other notes in your workspace are never sent unless you attach them
- workspace metadata, file uploads, and meeting recordings are never sent to the assistant
Where it goes
Requests are routed through the Vercel AI Gateway to one of:
- Anthropic (Claude family)
- OpenAI (GPT family)
- Google (Gemini family)
We instruct the gateway to forward a no-prompt-training preference to each provider. We do not use your notes to train any model and we do not permit our sub-processors to do so.
What we log
We log message-level metadata — model name, token counts, request duration — so we can keep usage limits accurate and debug failures. We do not log full prompt or response text on our infrastructure.
CAUTION
AI outputs are probabilistic. Treat them as a draft, not a source of truth. Don't paste regulated data, secrets, or third-party personal information into prompts.
Turning it off
Flip the toggle in Settings → Preferences → Writing assistant. Past chats are kept until you delete them or your account.