Vocabulary Signature

notion image
A living dictionary of the words and phrases you use, so AI writes with your vocabulary instead of its own.
Category
Context & Identity
Return
Every draft in your words
Stack
Vocabulary database · Banned list

The Problem

Every writer has a fingerprint: the words they reach for, the phrases they repeat, the terms they refuse. AI has its own fingerprint, and it's the same one for millions of users. Drafts that use the model's vocabulary instead of yours read as almost-you, and almost-you is what audiences notice.

The Solution

A living dictionary of your language. The words and phrases you use most, the concepts you name in your own way, and the terms banned from your writing, stored where every AI tool reads it before drafting.

How It Works

  1. We extract your most frequent words, phrases and signature concepts from a body of your real writing.
  2. Each entry records how you use it, so "energy" or "system" carries your meaning, not the dictionary's.
  3. A banned list captures the words that never appear in your writing, including the AI favorites.
  4. The signature installs next to the Writing Style rules, and drafts pull vocabulary from it first.
  5. New coinages get added as your language evolves, because a voice is a moving target.

Built With

  • A vocabulary database extracted from your real writing
  • Usage notes that carry your meaning for each signature term
  • A banned-words list installed as a hard rule
It's the finest-grained layer of the voice stack, under the Writing Style and the Personality Definition, and it's what makes the difference between imitation and fingerprint.

What You Get

  • Drafts built from your words, including the ones only you use
  • A banned list that keeps the model's stock vocabulary out
  • A record of your language that grows with you

The Numbers

½

Day to the first version

~1

Week to pay for itself
Half a day to extract and structure the first version. The return shows less in hours and more in recognition, readers hear you in the first paragraph, and the small edits that used to catch borrowed words disappear.