Catch off-brand copy before it goes to the client.

Three tone wordsThree adjectives that define how the brand sounds. Pick words that pull in slightly different directions, so they capture real personality instead of just "professional." Drafts are scored on how closely their tone lands to these. Examples: warm · precise · sensory — or calm · grounded · honest.

Three words that pull against each other. Example: warm · precise · sensory.

Banned words and phrasesSpecific words or phrases this brand never uses — hype, clichés, filler. The checker scans every draft for them and flags them instantly, with no AI involved, so it never misses one. Add a short reason for each, and it shows up on the flag. Examples: "literally" (filler), "game-changer" (cliché), "amazing" (hollow).

Specific words/phrases the voice never uses. Used by the deterministic vocabulary check. Aim for 3+.

Cadence rulesHow the writing should feel — sentence length, rhythm, and stance. This is about structure, not word choice: short vs flowing, talking with the reader vs lecturing at them, concrete detail vs vague claims. The AI weighs drafts against these. Examples: "Short sentences, mostly very short."; "Prefer sensory detail over evaluative claims."

Sentence-level rhythm. Sentence length, stance, lecture-vs-converse, sensory-over-claims.

Voice-on examplesA few short passages — one to three sentences each — that sound exactly right for this brand. These are the gold standard the AI compares new drafts against, so the more real, on-brand samples you add, the sharper the tone score. Example: "A mat that holds steady through a full vinyasa. That's it. That's the post."

Short paragraphs (1–3 sentences) that sound exactly right. Pairwise tone scoring uses these as the positive anchor. Aim for 3+.

Voice-off examplesShort passages that sound wrong for this brand, each with one line on why. They teach the AI what off-brand looks like, so it can tell the two apart. Example: "CRUSH your practice with this game-changer mat!" — why: violent verb, hype register.

Short paragraphs that read off-brand, each with one line on why. The negative anchor for pairwise scoring. Aim for 2+.

Punctuation rulesHow the brand handles exclamation marks, emojis, em-dashes, and ellipses. The exclamation and emoji settings are checked automatically; em-dashes and ellipses are described in plain words for the AI to weigh. Example: exclamation marks "at most one per post"; emojis "never."

CTA voice rulesHow the brand asks people to act. List the calls-to-action it likes, the ones it avoids, and the overall tone of the ask. Examples: preferred — "Book a consultation"; avoid — "Buy now!!!"; tone of the ask — "measured, no urgency theater."

Industry-specific terminologyWord swaps the brand insists on. Use this is the word to keep — the correct, on-brand term. Not this is the word to avoid; the checker scans drafts for it and flags it. Why is an optional note that shows up as the reason on the flag. Order matters — keep-word first, avoid-word second. Example: keep "lashes", avoid "extensions" → a draft saying "extensions" gets flagged and "lashes" is suggested.

Word swaps the brand insists on. Keep the first word, avoid the second — the check flags the avoided word in a draft and suggests the right one.

Client-specific exceptionsPlaces where the brand deliberately breaks its own rules, and why. Most brands don't need any — leave it empty if so. Example: "Exclamation marks are fine in holiday posts — the warmth fits there."

When the brand breaks its own rules and why. Optional — most brands don't.

Examples galleryFive to ten real posts, captions, or emails that nailed the voice. This is extra fuel for tone scoring — the more genuine on-brand writing the AI can see, the better it judges new drafts. Paste real published pieces, not invented ones.

5–10 real captions, emails, or posts that nailed the voice. Pairwise scoring leans on these. Aim for 5+.