How can we help?

Set your brand voice

Updated

Make every AI reply sound like your brand, not a generic bot. A brand-voice profile is applied to the AI agent's answers and to the copilot's drafts.

Setting it up

  1. Go to Settings → General → Brand voice.
  2. Pick a tone — Default, Friendly, Professional, Playful, Formal or Concise — and add any house style and rules: do's and don'ts, words to avoid, a sign-off.
  3. Save. From then on, AI replies and drafted responses follow it.

Shown with our demo workspace.

Voice versus facts

The brand voice shapes how the AI writes. Your knowledge base controls what it says. They are separate on purpose: however you describe your tone, the agent still answers from the same sources (your published articles, plus any knowledge files and snippets you add) and still cites the help articles it used.

That means a brand voice cannot make the agent more confident, more willing to guess, or more generous with a policy. If you want it to say something different, change the article, not the voice.

What to put in it

Specific instructions beat adjectives. Useful things to state:

  • The greeting and sign-off you want, if any.
  • Whether to use the customer's first name.
  • Words or phrases to avoid — including any you are contractually required to avoid.
  • How formal to be, with an example sentence in your preferred register.
  • Whether to use contractions, exclamation marks, or emoji.

"Warm but efficient; no exclamation marks; never say 'unfortunately'; sign off as the Support team" is far more actionable than "friendly and professional".

Where it applies

Both the customer-facing AI agent and the operator-side copilot drafts. A draft your team edits before sending still starts in your voice, which is usually less work than rewriting a generic one.

Testing it

Ask your own widget a question you know an article covers, and read the answer as a customer. Tone problems are much easier to spot in a real answer than in the abstract.

Was this article helpful?