TABLE OF CONTENTS

• Prerequisites
• Overview of configurations in AI Agent
• Manage multilingual support 
• Manage conversation behaviour
• Manage handover settings
• Manage agent identity
• Best practices for configurations

Fine-tune your agent’s conversation behavior, multilingual responses, identity, and escalation flow using the Configurations settings in AI Agent. They allow administrators to control how the agent interacts with users, which languages it supports, how conversations are handed over or closed, and how the agent appears to end users. This article explains each configuration option and provides steps to manage them effectively.

Prerequisites

Before configuring your AI Agent, ensure,

• Human agents, groups, and assignment rules are configured for successful handovers.
• Business hours are defined under Admin settings if you plan to use Transfer outside business hours.
• Knowledge sources are published (recommended) to ensure accurate multilingual and fallback responses.

Overview of configurations in AI Agent

The following configuration options are available for your AI Agent:

• Multilingual support: Manage the languages the agent responds in.
• Conversation behaviour: Define how the agent interacts, including tone, response length, and fallback handling.
• Handover settings: Configure when and how conversations are closed or transferred to a human agent.
• Agent identity: Personalize the agent’s name and avatar visible to users.

These settings collectively determine how your AI Agent behaves across channels and how smoothly it escalates conversations when needed.

Manage multilingual support 

Multilingual support enables the AI Agent to reply in various languages depending on user input. Choose from the pre-configured list of supported languages—click the 'See supported languages' link to see them. This feature guarantees a consistent user experience across different regions and customer groups.

How multilingual support works

• The agent automatically detects the user’s language.
• Responses are generated in the detected language.
• If the language is unsupported, fallback handling is applied.
• Handover rules still apply regardless of language.

Note: Ensure your knowledge files, solution articles, and QnA support the selected languages for accurate responses.

Manage conversation behaviour

The Conversation behaviour settings help you standardize greetings, handle unanswered queries, collect structured information, and manage conversation closure or escalation.
You can,

• Send introductory message
• Send fallback messages
• Collect feedback
• Collect user details

For detailed capability explanations and advanced configuration guidance, see Configure conversation behaviour for AI Agents.

Manage handover settings

The Handover settings determine when conversations are transferred to a human agent or automatically closed, ensuring a smooth escalation without disrupting the customer experience.

You can,

• Transfer to a human agent
• Auto-resolve conversations
• Transfer outside business hours

For detailed capability explanations and advanced configuration guidance, see Configure handover settings for AI Agents.

Note: When you enable an AI Agent for a Web Chat topic, the default topic-level Away Experience will be disabled. Instead, the AI Agent–level handoff experience will be applied to that specific Web Chat topic.

Manage agent identity

The Agent identity setting controls how your AI Agent appears to end users. A well-defined identity increases user trust and improves engagement.

You can customize,

• Name: Enter a name that reflects your brand identity. This name will be visible to end users in conversations.
• Avatar: Choose from a library of avatars or upload a custom image using the upload icon. The avatar is visible in chat interfaces, adding a personal touch to the AI. 

Best practices for configurations

Follow these best practices to ensure your AI Agent configurations deliver a consistent and scalable support experience:

• Align with strategy: Ensure tone, fallback handling, and handover rules reflect your brand voice and overall support model (self-serve or assisted).
• Match SLA timelines: Configure auto-resolve durations in line with your SLA commitments to avoid premature closures or stale conversations.
• Validate multilingual flows: Review greetings, fallback messages, and feedback prompts across supported languages before enabling in production.
• Monitor escalation rates: Track fallback and transfer trends after making configuration changes to identify knowledge gaps or overly aggressive handovers.
• Check agent availability: Ensure agent groups, queues, and business hours are properly configured to handle transferred conversations efficiently.
• Review collected properties: Collect only necessary user details and ensure compliance with internal data governance and security policies.

Periodically audit these settings to maintain optimal automation performance and a balanced human–AI support workflow.