Voice Profiles
A voice profile is a description of how you write. It's the single biggest lever you have for making Robin's output sound like your prose, not generic prose.
What goes into one
On the web, a voice has a name, a description, a target audience, Style and Tone fields, and Writing Samples. A complete voice profile covers three layers.
- Identity. Tone (warm, ironic, clinical), point of view (first person past, close third, omniscient), tense, register.
- Mechanics. Sentence-length tendencies, paragraph rhythm, dialogue style, common punctuation choices, words you use a lot, words you never use.
- Samples. Pasted passages of your actual prose: 3 to 20 samples, about 5,000 words in total. The richer the samples, the better the match.
You can keep multiple voice profiles (one per book, or one per persona) and switch between them. Each book points at the voice profile that best fits it.
Why it matters
When Robin is asked to generate or rewrite text, Robin reads the voice profile before your request. That's why the same prompt produces dramatically different output across two writers using the same app: their voice profiles are different.
If a reply from Robin feels generic, the cure is almost always to add more detail and more samples to the voice profile, then try again.
Sharing voices in a workspace
Voice profiles live inside a workspace, just like books. In a shared workspace, every member can read and use the workspace's voices, but only editors and above can change them.
When you copy a book to a new workspace, the attached voice comes with it: a personal voice is deep-copied, a shared/global voice is referenced.
Where to set them up
- Web: see Web → Voice Profiles.
- Mobile: see Writing Voices on Mobile for picking, creating, editing, and trashing voices.