TABLE OF CONTENTS

Conversation behavior settings let you control the Chat AI Agent’s interactions with customers throughout a conversation. These settings help create a consistent customer experience, from the initial greeting to the collection of information before a handover. For details on the complete Chat AI agent setup, see Create Chat AI agents.


Note: For details on setting up the conversation behaviour for Email AI Agents, see Configure Email AI Agent conversation behavior.


Prerequisites

  • Ensure that AI Agent Studio is enabled for your account.
  • Ensure that you have administrator access to AI Agent Studio.
  • Create a Chat AI Agent before configuring its conversation behavior.

Configure conversation behavior

You can customize introductory and fallback messages, collect customer information, gather feedback, and configure the actions the agent takes when it cannot resolve a query.

To configure the conversation behavior:

  1. Go to AI Agent Studio from the left navigation.
  2. Select the Chat AI agent you want to configure.
  3. Go to Build > Configurations > Conversation behavior.
  4. Configure the following settings:
    • Send introductory message
    • Send fallback message
    • Collect feedback
    • Collect user details

You can use the default messages or customize them to match your brand voice.

Send an introductory message

The introductory message is the first message customers receive when they start a conversation with the AI Agent. Customize this message to welcome customers and set the context for the interaction.


The message editor supports the following formats:

  • Rich text formatting
  • Hyperlinks
  • Emojis
  • Placeholders
  • Multiple languages

You can also configure an action that the Chat AI agent should perform before proceeding with the conversation. For example, you can prompt customers to provide the configured contact or conversation properties after the introductory message is sent or trigger a configured workflow.


To configure the introductory message:

  1. Under Conversation behavior, select Send introductory message.
  2. Enter or edit the message.
  3. Click Add action after message and select Collect user details or Trigger Workflow.


  4. If you select Trigger Workflow, choose the workflow to run.
  5. If you select Collect user details, define the properties to collect from customers. For details, see Collect user details.
  6. Click Review translations to verify and edit the translated messages. Click Manage languages to add or remove supported languages.
  7. Click Save.

Send a fallback message

A fallback message is displayed when the Chat AI agent cannot answer a customer's query. It lets customers know that the agent cannot provide an answer and directs the conversation to the next appropriate action. You can choose to send a pre-configured message or route it to a workflow.

To configure the fallback action:

  1. Under Conversation behavior, select Send fallback message.
  2.  Choose one of the following actions:
    • Send configured message: Select this option to display a custom message when the Chat AI agent cannot answer a query. Use the rich text editor to format the message and add links, emojis, and translations.

      You can also enable Transfer to a human agent to hand over the conversation after sending the fallback message. 


      If you enable this option, click Manage to collect the customer details (contact or ticket properties) before the handover.

      Note: The customer details required before an agent handover differ between Freshdesk standalone and Freshdesk Omni. For product-specific requirements and configuration steps, see Collect user details.

    • Execute a workflow: Choose a workflow to start a configured workflow when the Chat AI agent cannot answer a query. This option lets you route unanswered queries through a guided process instead of displaying a fallback message and initiating a handover.


Collect feedback

Collect customer feedback to understand whether customers find the Chat AI agent's responses helpful. You can personalize the feedback prompt using rich text formatting, links, and emojis to match your brand voice.

You can collect the feedback in the following scenarios:

  • After the AI Agent responds using a knowledge source.
  • After a workflow completes.
Note: The customer details required before an agent handover differ between Freshdesk standalone and Freshdesk Omni. For product-specific requirements and configuration steps, see Collect user details.


To configure feedback:

  1. Under Conversation behavior > Collect feedback, enable Always ask for feedback.
  2. Enter the feedback message. For example, "Was this helpful?"
  3. Configure the response for positive (Yes) and negative (No) feedback.


  4. Enter the messages customers see for each response.
  5. Click Save.

Note: For workflows, feedback is collected after the entire workflow is completed.


Collect user details

Use Collect user details to gather customer information at specific stages of a conversation. You can collect contact fields or ticket properties to capture customer details at two stages:

  • During introduction: Collect customer information at the start of a conversation, such as the customer's name, email address, phone number, or customer ID.
  • Transfer to a human agent: Collect customer information before handing over a conversation to a human agent. You can configure this action in the following scenarios:
    • Fallback message: When the AI Agent cannot resolve a customer's query and needs to transfer the conversation for further assistance.
    • Collect feedback: When a customer selects No in response to a feedback prompt.
    • Handover settings: When a conversation is configured to be transferred to a human agent. This can be configured under Configurations > Handover settings.

Note: Click Manage in any of these settings to configure the customer details to collect before the handover.


What happens when a conversation is transferred to a human agent?

The handover experience differs between Freshdesk Omni and Freshdesk standalone. This also determines the customer information you must collect before enabling the handover.


ScenarioFreshdesk OmniFreshdesk standalone
Conversation handlingThe conversation is transferred to a human agent or group.Live transfer to a human agent is not supported. 

The chat ends when the handover is initiated, and further communication happens via email.

Customer information to collect
Configure the contact fields or ticket properties the Chat AI Agent must collect before transferring the conversation.The customer's email address is mandatory to continue the conversation over email after the chat ends. 


Note: The Chat AI Agent on Freshdesk standalone is currently available only for new signups.


To configure the properties to collect:

  1. Under Conversation behavior, select Collect user details.
  2. In the Collect user details pane, enable the required stage:
    • During introduction
    • Transfer to a human agent
  3. Add the properties that the AI Agent should collect. Choose from the available Text, Number, or Paragraph fields.
  4. Add a description to provide additional context for a property, if required.
  5. Mark the properties that customers must provide as Required. Customers can skip properties that are not marked as required.
    Note: In Freshdesk standalone, ensure that you add Email as a mandatory property before transferring the conversation to a human agent.
  6. Click Sync fields to refresh the available contact fields and ticket properties.
  7. Click Save.


Notes
  • Ensure that Collect user details is enabled in the relevant sections, such as Send introductory message, send fallback message, collect feedback, or handover settings, depending on where you configure the action.
  • Collecting unique identifiers, such as email addresses and phone numbers, from visitors may have security implications. Ensure that you handle this information appropriately.


Manage or remove properties

You can modify or stop collecting properties at any time.

  • Turn off During introduction or Transfer to a human agent to disable property collection at that stage.
  • Click the minus (-) icon next to a property to remove it.
  • Click Sync fields to refresh contact fields and ticket properties after making changes to your account.
  • Click Save to apply your changes.

If you enable a section without adding any properties, the page displays a validation message. Add at least one property before saving.

If a property has been removed from your contact fields or ticket properties, select another supported property or remove the invalid entry before saving.



Important: When workflows retrieve sensitive information, such as order details, refunds, or account information, ensure that the workflow collects and validates the required identifiers before providing the information.
If verified user properties are passed through the widget script, the AI Agent can use them without prompting the customer again. If these properties are not passed, configure the workflow to collect and validate the required details, such as the order ID and email address.
For best results, configure the widget to pass verified user properties.


Supported languages

By default, the AI Agent understands queries in all languages and responds in the following manner:

  1. If the user's query is in one of the added languages in the supported languages section: The AI Agent will reply in the language of the query.
  2. If the user query is outside the added languages in the supported languages section: The AI Agent will respond in the primary language with a language disclaimer: “Apologies! I am trained to answer in <added languages>”


The supported languages include the following:

AmharicAssameseBengaliBosnianBulgarian
CatalanChineseChinese (Traditional)CroatianCzech
DanishDutchEnglishEstonianFilipino
FinnishFrenchGermanGreekGujarati
HebrewHindiHungarianIcelandicIndonesian
ItalianJapaneseKannadaKazakhKhmer
KoreanKurdishLatvianLithuanianMalay
MalayalamMarathiNepaliNorwegianOdia
PolishPortuguese (BR)Portuguese (Portugal)PunjabiRomanian
RussianSerbianSinhaleseSlovakSlovenian
SomaliSpanishSpanish (Latin America)SwahiliSwedish
TamilTeluguThaiTurkishUkrainian
UrduVietnameseWelsh