Agent Settings

Agent Settings is where you configure the key properties of your agent. These settings control how the agent identifies itself, which model it uses, and how it behaves in conversations.

Accessing settings

1. Select your agent from the Agents list.
2. Click Agent Settings in the sidebar.

Configuration options

Agent name

The display name for the agent in the console. This is only visible to your team — users see the chat widget branding, not the agent name.

Language

The default language for the agent. This influences the language of system-generated messages (like greetings and error fallbacks). The agent can still detect and respond in the user’s language regardless of this setting.

Model

The AI model used to generate responses. Different models offer different balances of quality, speed, and capability. See Model Selection for guidance on choosing the right one.

System prompt

The instructions that define your agent’s personality, rules, and boundaries. This is the most important setting — it shapes every response your agent gives. See Configure Behavior for a detailed guide.

Saving changes

Click Save after making changes. Updates take effect immediately for new conversations. Active conversations continue with the previous settings until a new message is sent.

Best practices

• Test after every change — use the Playground to verify that changes produce the results you expect.
• Change one thing at a time — if you adjust the model and prompt simultaneously, it’s hard to tell which change caused a difference in quality.
• Document your settings — keep a note of what works well for each agent so you can replicate it for new ones.

Next steps

• Configure Behavior — write an effective system prompt.
• Model Selection — choose the right model.
• Playground — test your settings in real time.
• Knowledge Base — improve answer quality by adding more documents.