> For clean Markdown of any page, append .md to the page URL.
> For a complete documentation index, see https://docs.aura.konceptservices.com/llms.txt.
> For AI client integration (Claude Code, Cursor, etc.), connect to the MCP server at https://docs.aura.konceptservices.com/_mcp/server.

# Configure agent behavior

> Adjust how your agent replies to messages: style, emojis, message splitting, and response policy.

## Where to configure

All of these options are found in **Agent Settings > Agent > Messages**.

## Response style

The **Response style** field defines the length of the agent's replies:

* **Concise**: Short and direct replies, straight to the point.
* **Detailed**: Longer replies with explanations and additional context.

> **Tip**
>
> For channels like WhatsApp, where messages tend to be quick, the concise style works best. For email, a detailed style may be more appropriate.

## Emoji usage

Control how often the agent uses emojis in its replies:

| Option         | Description                                                 |
| -------------- | ----------------------------------------------------------- |
| **Never**      | The agent will not use emojis under any circumstances.      |
| **Rarely**     | The agent will use emojis very occasionally.                |
| **Sometimes**  | The agent will use emojis from time to time, in moderation. |
| **Frequently** | The agent will use emojis often in its replies.             |

## Message splitting

The **Message splitting** field controls whether the agent sends its entire reply in a single message or breaks it into several shorter messages. The value ranges from 0 to 5:

* **0**: The reply is sent as a single message.
* **5**: The reply is split into the maximum number of messages possible.

> **Info**
>
> Splitting messages does not affect cost. It is simply a matter of style: short, separate messages look more like how a real person writes in a chat.

## Fallback message

The **Fallback message** is the text the agent sends when it cannot generate an appropriate reply. For example:

> "I could not process your message. Could you rephrase it?"

Customize this message to match the tone of your business.

## Response policy

The response policy lets you define clear rules about when the agent should or should not reply:

### Reply only when

In the **Reply only when** field, describe the situations in which you want the agent to reply. If you leave it empty, the agent will reply to all messages.

Example: *"The customer asks about reservations or has a question about the service"*

### Do not reply when

In the **Do not reply when** field, describe the situations in which the agent should ignore the message and not reply.

Example: *"The message is a complaint or the customer asks to speak with a human"*

> **Info**
>
> You can also set a specific response policy for each individual conversation. If a conversation has its own policy, it will take priority over the general settings.