AI brand voice guidelines: the layer under everything.

I use written voice and brand guidelines as the thing every other skill checks itself against. They are not a document I read - they are loaded into the skills, so when Claude writes an email, a post, a newsletter, a sales room or a lesson, it already knows my tone and the meritt brand rules and holds them by default. I reach for them on literally everything that represents meritt: the moment AI writes or designs anything outward-facing, the voice and brand layer is on. The reason the other 15 skills produce on-brand output instead of generic AI is that this layer sits under all of them.

How I use this, and what for

I use written voice and brand guidelines as the thing every other skill checks itself against. They are not a document I read - they are loaded into the skills, so when Claude writes an email, a post, a newsletter, a sales room or a lesson, it already knows my tone and the meritt brand rules and holds them by default. I reach for them on literally everything that represents meritt: the moment AI writes or designs anything outward-facing, the voice and brand layer is on. The reason the other 15 skills produce on-brand output instead of generic AI is that this layer sits under all of them.

What the AI actually does

It applies your voice and brand rules to everything it writes, automatically, so you do not have to. Without a defined voice, an AI writes a hundred slightly-off things a week and the brand blurs. With the guidelines wired in, every output - across email, social, content, decks, pages - comes out in one voice and one look. You get the speed of AI without the cost of it sounding like AI.

This is a real, supported pattern, not a workaround. Claude lets you save a brand voice as a reusable style or project so the same tone, rules and vocabulary apply every time, instead of being re-explained per chat. Teams that configure a brand style this way cut content revision time by a reported 40-60% - because the output starts on-brand rather than getting dragged there in editing.

The tools and APIs you need

• Claude Code
• Wispr Flow

The right prompts

Apply voice and brand to a piece

```
Write this in my voice and brand-check it against the meritt V3 guidelines
before you show me - no em dashes, lowercase meritt, plain English,
candidate-first, no buzzwords. [paste or describe the piece]
```

Audit something for brand drift

```
/meritt-core-meritt-brand
Check this against our voice and brand guidelines and flag anything off -
tone, em dashes, casing, jargon, visual rules. Then fix it. [paste]
```

How to do it

Keep the guidelines written down and current

They only work if they are explicit - tone, the rules, the palette, the do-nots. Vague guidelines cannot be enforced.

Wire them into the skills, do not paste them each time

The voice and brand skills carry the rules, so every other skill inherits them by naming the skill.

Name the voice on anything outward-facing

"In my voice" pulls will-voice; Lauren's work pulls lauren-voice.

Run the brand check before publishing

- email, post, newsletter, page, lesson. It catches the tells (see below) before they go out.

Update the guidelines, and everything updates

Change the rule once in the source of truth and every skill that reads it follows the new rule.

See the difference

Let AI carry the heavy lifting; you keep the judgement and the final say.

How you'll know it's working

• Everything meritt puts out sounds like one company, in one voice, with one look.
• Off-brand output gets caught by the gate, not by a customer.
• "In my voice" means something repeatable, because the voice is written down and wired in.
• The other 15 skills come out professional by default - because this layer is under all of them.

Questions people ask