The Now Platform® Washington DC release is live. Watch now!

Help
cancel
Showing results for 
Show  only  | Search instead for 
Did you mean: 
Nabil Asbi
ServiceNow Employee
ServiceNow Employee

Before getting started with Virtual Agent, it is important to understand a few concepts related to its implementation in the ServiceNow platform.

The Virtual Agent consists of a few different components that create the experience for the end user. As an admin, you must make sure each of these components are properly setup and configured. This article will take you through the basics of getting everything activated to drive out-of-the-box conversations that provided with the solution. The following diagram shows the high-level building blocks of the Virtual Agent.

 

find_real_file.png

 

Virtual Agent Environments

Runtime Environment

The Virtual Agent runtime environment is the running system responsible for user interaction. The behavior of the Virtual Agent is based on the published topic, and the dialog interactions flow within that topic. On the NLU side, the topic returned to the user is based on the published and trained NLU model and the user's utterance.

 

Design Environment The Virtual Agent design environment is configured in separate interfaces. This allows the administrator to make modifications to either the conversation flow and entity/intent processing (in the Virtual Agent Designer) or the trained NLU model (in the NLU Workbench). The changes made in the design environment only show up in the runtime environment after publication.

NOTE: If you make changes to both the Virtual Agent and the NLU model, you have to publish each change separately.

 

Virtual Agent Components

Now that you know about the two modes, let's take a look at the individual components in the Virtual Agent runtime environment:

find_real_file.png

Chat Interface: The Chat Interface, i.e. chat client, is the component that the user directly uses. In other words, the user types directly into the Chat Interface to ask questions or get services. The chat client on web can be set up as part of the Service Portal or Employee Center. When the user inputs a statement into the chat client, it is passed to the VA Runtime Engine.

It is also possible to use a mobile client, or a 3rd party chat interface, like Microsoft Teams or Slack, for the Virtual Agent. For more information, refer to the Virtual Agent integration with messaging apps documentation.

 

find_real_file.png

Virtual Agent Runtime Engine: When a user submits a message through the Chat Interface, the VA Runtime Engine processes the message via topic discovery. The VA Runtime Engine interprets the user utterance and determines the intent (what the user really wants) and entities (attributes/values describing the intent), and then passes it to the NLU engine as a Prediction Request for further processing. The VA Runtime Engine gets an extracted intent from the NLU Engine and uses that to continue the dialog with the user.

The entire conversation flow (and what gets presented to the user) is controlled by the Dialog component, which is what actually delivers the response to the user in the Chat Interface.

 

If NLU is unable to discern the intent, the user receives search results (either via AI Search or Contextual Search) as a fallback. 

 

find_real_file.png NLU: ServiceNow® Natural Language Understanding (NLU) provides an NLU Workbench and NLU engine that learns and responds to human-expressed intents. When entering natural language inputs (utterances) into the Virtual Agent, it understands word meanings and contexts to provide answers or system actions. For more information, refer to the NLU documentation

 

Comments
PaulSylo
Mega Sage
Mega Sage

@Nabil Asbi  @Victor Chen  - What level of data or prerequisite ( I would say) to have a virtual agent implemented successfully? I would like to know the level of tied up components like catalogs, tuning the knowledge article 

Version history
Last update:
‎12-06-2023 04:41 PM
Updated by:
Contributors