For the complete documentation index, see llms.txt. This page is also available as Markdown.

Advanced layout settings

In addition to changing the appearance and interface of a layout, you can configure advanced settings that control how the layout behaves when it is published or embedded.

The Advanced settings area is divided into three sections:

  • SEO settings (search engine optimizations);

  • Widget advanced parameters;

  • PII detection.

These settings apply to the specific layout you are editing.

SEO settings (search engine optimizations)

Use SEO settings to configure the information associated with the layout when it is accessed as a web page or shared through compatible platforms.

  • Title: defines the title associated with the page;

  • Description: defines the description associated with the page. This information can be used by search engines and when the page is shared;

  • Canonical URL: defines the preferred URL associated with the page;

  • Preview image: select an image to use as the preview image when the page is shared through compatible platforms;

  • Show avatar as preview image: enable this option to use the Agent's avatar as the preview image instead.

The section also shows the corresponding HTML metadata generated from the configured values.

For example:

<meta property="og:title" content="File Piattaforme 2026" />

You can use the copy button next to the generated metadata to copy it.

💡 If the Agent is embedded inside an existing website, also review the SEO configuration of the page that contains the Agent.

Widget advanced parameters

Use Widget advanced parameters to control how conversations started through this layout behave and how the layout interacts with an embedded widget.

  • Initial context: provide contextual information that should be available when a conversation starts.

    The initial context only works when embedding the widget in your application or website. For more information about how to use context variables, please click here.

  • Initial question: define a question or instruction that is automatically sent to the Agent when a new conversation starts.

    The initial question only works when embedding the widget in your application or website.

    The instruction itself is not shown to the user. Instead, the Agent processes it when the conversation starts and the user sees the resulting response.

    For example:

    Introduce yourself and briefly explain how you can help the user.

    The Initial question is different from the Agent's general welcome message because it belongs to this specific layout and can therefore be different for each embedded experience.

  • Timeout: define the timeout used by the layout.

    The value is expressed in seconds.

    Use this setting when the embedded experience requires specific behavior after a period of inactivity.

  • Role: choose the role that should be impersonated when a conversation is started through this layout.

    This allows the same Agent to start conversations using a different role depending on the layout through which the user accesses it.

  • Agent Language: select the Agent language to use for this layout.

  • Whitelisted domains: restrict the domains that are allowed to embed and use the widget.

    Enter one domain URL per line.

    Leave the field empty to allow all domains.

    Add localhost to the list when you need to allow local development.

  • Ignore client attributes: enable this option when the values configured in the layout should take precedence over attributes provided by the client implementation.

    When enabled, client attributes passed through the React component or Web Component are ignored and only the values configured in the layout are used.

    Leave this option disabled when the embedding application should be able to provide or override supported configuration values through client attributes.

PII detection

Use PII detection to prevent users from sending messages that contain personal or sensitive information.

  • Enable PII detection: enable this option to check messages for personal or sensitive data before they are sent to the Agent.

    PII detection can block messages containing information such as email addresses, fiscal codes, or other configured sensitive data.

    When PII detection is enabled, configure the detection rules and the message that should be shown to the user when content is blocked.

This allows sensitive information to be intercepted before the message is submitted to the conversation.

Advanced settings are configured independently for each layout. This means the same Agent can use different SEO information, embedding parameters, roles, domain restrictions, and PII detection settings depending on where and how each layout is published.

To learn how different layouts can coexist for the same Agent, continue with Manage multiple layouts.

Last updated