An agent's personality lives in three places: how it presents visually, how it sounds in conversation, and (optionally) how it sounds out loud. This guide covers all three.
Display name
Set under Settings → General. The display name is what shows in the chat widget header and inside conversations on every channel.
Use a real name ("Maya from Acme") or a clear function ("Acme Support"). Avoid robotic names like "AI Bot" — customers respond better to something human.
Profile picture and icon
Two images, two places:
- Profile picture — large avatar shown inside the chat (next to messages). Use a friendly photo or a high-quality illustration.
- Icon — small mark used in the launcher bubble and headers. Use a simple, recognisable shape — your logo mark works well.
Upload both from Settings → Interface. PNG or JPG, ideally square, at least 256×256px.
Tip
Customers trust agents with photos more than agents with logos. If you don't have a designated team photo, a stylised illustration of a person beats a logo every time.
Voice
Voice is two different things in Chatzuri:
- Written voice — defined entirely by your system prompt. See the AI configuration guide.
- Spoken voice — only matters if you've enabled multimodal voice. Pick a voice from the dropdown under Settings → Multimodal.
Initial message and suggested replies
These set the tone of every new conversation:
- Initial messages — what the agent says first, before the customer types anything. One sentence is usually best.
- Suggested messages — quick-reply buttons shown underneath the initial message. Three to five options. Each should be something a customer would actually click.
Examples that work
Initial: "Hi! I'm Maya from Acme. What can I help with today?"
Suggested:
- See pricing
- Book a demo
- Talk to a humanNote
Suggested replies massively boost engagement. Customers click rather than type, and your agent immediately has structured intent to work with.