> ## Documentation Index
> Fetch the complete documentation index at: https://www.plain.com/docs/llms.txt
> Use this file to discover all available pages before exploring further.

# Tone of voice

> Set the rules that control how AI sounds in customer-facing replies.

Tone of voice rules control how AI sounds in customer-facing replies. So any AI-generated responses stay consistent with your brand.

Tone rules are configured per workspace from **Settings → Plain AI → Configuration → Tone of voice**. They apply to all customer-facing AI in that workspace, including

* Suggested responses in the composer

* Ari responses

* Help Center AI chat

* Sidekick draft replies.

Plain AI automatically adjusts writing style based on the channel, like using a structured letter format for email and a concise, conversational style for chat. You don't need to set this yourself.

## How tone rules work

Rules are organized into **five rule categories**:

* **Empathy**: How much AI should acknowledge feelings and respond with understanding

* **Language**: Terminology, jargon, and level of technicality

* **Formality**: How formal or casual messages should feel

* **Personality**: How brand voice comes through

* **Warmth**: How friendly and positive messages should feel

You can fill in each category with your rule: short, free-text instructions that tell the model how to behave (e.g. "Use plain language and avoid jargon" or "Be warm and conversational").

## Generating rules from a description

If you already have a style guide or a short description of how you want to sound, you can use **Generate rules** instead of writing each rule by hand. In the popover, paste or type a general description of your brand / tone guidelines (e.g. a paragraph from your style guide). Plain tries to turn it into suggested rules and fill in the same five categories for you. Those suggestions appear as **drafts** which you can edit, and choose which to keep or reject.