Customizing a Virtual Agent chat experience
Summarize
Summary of Customizing a Virtual Agent chat experience
ServiceNow's Virtual Agent enables you to create tailored chat experiences for end users based on the context in which they start a conversation. A chat experience controls the conversation flow, featured topics, and AI Search settings to optimize user interactions. You can use the default chat experience or create custom ones specific to different service portals or channels, enhancing relevance and usability.
Show less
Key Features
- Setup Topics: Pre-built conversational elements like greetings, feedback requests, live agent support, error handling, and conversation closing. These run automatically at appropriate points based on user input, context, or AI models (NLU or LLM). Setup topics come with default intents and keywords, and can be previewed, duplicated, and customized if needed.
- Promoted Topics: Up to six user-relevant topics displayed prominently in the chat window to provide quick access. These are customizable per chat experience but may not be supported on all clients.
- Search Profile: Defines how AI Search results are generated and displayed within Virtual Agent conversations, including fallback results when no matching intents are found. The profile controls data sources and UI elements like answer cards and multi-link outputs.
- Customization and Context Awareness: Custom chat experiences inherit default settings but allow overrides for setup topics, promoted topics, and search profiles based on where and how users engage with Virtual Agent.
- Virtual Agent Lite Limitation: Only allows configuration of certain setup topics in the default chat experience and does not support creating custom chat experiences.
Practical Use and Configuration
Admins and topic authors can manage chat experiences via the Virtual Agent settings interface. Key actions include:
- Configuring the default chat experience to activate or modify setup topics and promoted assets globally.
- Creating custom chat experiences tailored to specific user contexts or channels, overriding default settings as needed.
- Testing chat experiences to simulate conversations and verify behavior before deployment.
Setup topics form the backbone of a consistent conversation structure, ensuring users receive standard greetings, assistance options, and fallback messages automatically. Promoted topics and search profiles enhance discoverability and response relevance, improving user satisfaction and engagement.
Key Outcomes
- Deliver personalized and context-aware chat interactions that improve user experience across different access points.
- Maintain control over conversation flow and topic availability to align with organizational goals and user needs.
- Leverage AI Search integration to provide relevant fallback options and enriched conversational responses.
- Efficiently test and iterate chat experiences to ensure they meet user expectations before going live.
Create different chat experiences for your end users based on the context in which they initiate a conversation with Virtual Agent.
How chat experiences work
A chat experience defines the structure of a bot conversation (setup topics), an initial set of topics displayed to users (promoted assets), and the AI Search configurations for displaying search results in conversations. Virtual Agent provides a preconfigured, default chat experience that you (admins and topic authors) can use to control the user experience with Virtual Agent.
- Setup topics
-
Virtual Agent provides common conversational elements, such as a welcome greeting to begin a conversation and a conversation closing, and automatically includes them in virtual agent conversations. These pre-built elements, called setup topics, are part of a basic conversation structure that is applied to all your conversations. They are installed with Virtual Agent and Virtual Agent Lite.
These setup topics run automatically at appropriate points in a conversation, based on the context, utterances (if NLU is enabled), topic description (if LLM is enabled) or keywords that users enter during the conversation.
Figure 1. Example setup topics in a conversation - Greeting
- LLM topic (conversational catalog request)
- Closing
Virtual Agent provides the following types of pre-built setup topics:
Setup topic Description Greeting Presents a welcome greeting to users, and asks users to enter a request or see what items it can assist with. Provide Virtual Agent Feedback Displays a survey to get user feedback on the conversational experience, before the conversation ends. Live Agent Support Enables users to request a live agent transfer and view items that the live agent can assist with. Virtual Agent Capabilities Presents a list of what the virtual agent can help with. Error Handling Topic Displays a standard error message when an unrecoverable system error occurs and transfers the user to a live agent. Closing Conversation Displays a closing message that ends the conversation. AI Search Fallback Generates AI Search results for a user when Virtual Agent can't find a matching intent and topic or keyword. Fallback Topic Presents standard messages that ask the user to enter another request or select a different topic when the virtual agent does not understand a user entry or selection. Anything Else Topic After completing a task or request, virtual agent asks if the user needs further assistance with another request or task. In NLU conversations, setup topics have corresponding intents (set in the pre-built NLU model for setup topics) and also keywords. Virtual Agent recognizes the utterances or keywords that users can enter to run certain setup topics, such as the Virtual Agent Capabilities (help) topic or the Live Agent support topic. Setup topics for LLM conversations rely on their own descriptions as the intents, which the LLM uses for topic discovery.
You can preview (test) the pre-built setup topics to see how they work. On the Virtual Agent Designer home page, use the filter icon
to sort only for Setup Topics, and then select the setup topics to test. Pre-built topics are read-only, but you can duplicate, customize (modify) if needed, and publish them.
- Promoted topics
-
You can display a collection of up to six topics that are relevant to your users, displayed before the Show me everything button in the chat window. You can use these promoted assets in a chat experience to give your users quick and convenient access to the topics used for a given context.Note:Promoted topics may not be supported on all clients.
For more information, see Promote or demote LLM conversational subflows, actions, and topics in Virtual Agent Designer
- Search profile
-
A search profile defines how Virtual Agent search results are generated for an AI Search application. The search application's configuration specifies a search profile to use, and also determines how the generated results are displayed to users. Virtual Agent uses a default search application configuration and search profile that controls the data sources used for searches. Its UI configuration (set in the Entity View Action Mapping application) controls the search results displayed in Genius Result answer cards and multi-link output. These configurations also apply to search results generated as a fallback when Virtual Agent can't determine the appropriate topic. For more information, see Improving the user experience with AI Search.
Get started with chat experiences
Navigate to , and then select Virtual Agent. Use the Custom Greetings and Setup tile to manage the default and custom chat experiences. You can do the following:
- Configure the default chat experience, including the setup topics, promoted assets, and search criteria used. For example, you can activate the Virtual Agent Feedback setup topic so that it is applied to all your bot conversations.
- Create different custom chat experiences based on the context in which your end users run Virtual Agent. The settings that you define in a custom chat experience override the default chat experience.