Skip to content

Writing a Brand Brain

How one brand's tone, audiences, banned phrases and claim rules are recorded, and which of them are actually enforced.

Updated

A Brand Brain is how one brand sounds: its tone of voice, personality, audiences, the phrases it must never use and the rules about what it may claim. There is one per brand, and like Business DNA every save creates a new immutable version.

The important thing to understand before you write claim rules is that only some of them can be enforced automatically. Quoted wording is blocked; a free-form principle is passed to the writer as guidance and is reported as unenforceable rather than silently ignored.

BEFORE YOU START

  • The brand must already exist. Create it at /app/brands first.
  • You need the "edit" permission, and you need to be able to reach the workspace the brand lives in. If you cannot, the page returns you to the brands list without saying which of the two is missing.
  • It costs no credits and needs no AI provider.
  • Decide in advance which phrases you want hard-blocked. Those need to be written as quoted wording.

STEPS

  1. Open /app/brands and choose "Open Brand Brain" on the brand you want.
  2. Fill in "Tone of voice" and "Personality". Write how the brand speaks, not what it sells; what it sells belongs in Business DNA.
  3. Fill in "Audiences" with who the brand is actually addressing.
  4. Fill in "Banned phrases", one per line. Each line is capped at 200 characters. These are matched case-insensitively, on whole words for a single word and as a substring for a multi-word phrase.
  5. Fill in "Claim rules", one per line, each up to 500 characters. To get a rule enforced, quote the wording you want blocked, in the form: never say "guaranteed", or never claim "fastest".
  6. Choose "Save new version".

WHAT YOU SHOULD SEE

A confirmation reading "Saved as version 1." (or the next number up), and the brand's chip on /app/brands changes from "Brand Brain not written yet" to "Brand Brain v1".

WHAT THIS WILL NOT DO

  • It does not automatically block a free-form rule. "No unverified statistics" is given to generation as guidance, and Brand Guard reports it as unenforceable because it cannot tell which wording it forbids. Only quoted phrases are blocked.
  • It does not silently rewrite output that breaks a rule. A violation is recorded against the generation so you can see it, rather than being edited away where you cannot.
  • It does not cover every field the API accepts. Terminology, sample content and languages can be stored through the API but have no field on this page, so they cannot be set here.
  • It has no version list or restore button. To undo, save a corrected version on top.

IF IT DOES NOT WORK

  • "You do not have access to this brand." means the brand is in a workspace you cannot reach. Ask an administrator for access to that workspace.
  • "Edit permission required." means your role cannot save brand settings.
  • Opening the page bounces you to the brands list: same two causes, deliberately not distinguished so the page cannot be used to discover which brands exist.
  • "The request did not complete, so it is not known whether this version was saved." means reload first and read the current version number before saving again.

COMMON QUESTIONS

Why are my banned phrases still appearing?

Check the match. A single word is matched on word boundaries, so "cheap" will not match "cheaper". A multi-word phrase is matched as a substring. If it is genuinely in the output, the violation should be recorded against that generation.

Can one Brand Brain cover several brands?

No. It is per brand by design, because two brands in the same organisation are usually the case where a shared voice would be wrong.

Which wins, the Brand Brain or a writing profile?

They do different jobs. The Brand Brain is the brand's fixed voice and guardrails; a writing profile is a reusable steer you pick per generation. Prohibited words from both are combined.

  • Writing your Business DNA

    The eight facts about your business that generation is allowed to cite, and how saving a new version works.

  • Adding and managing brands

    How to create a brand, what your plan allows, and why the brand count on the page can differ from the number you can see.

  • Setting your brand voice

    Reusable steers for generation - tone, length, reading level, audience, words to avoid and words to include.

  • Writing posts with AI

    Generating a first draft from a brief, checking what it will cost first, and what the output is and is not.