To create an automated conversational flow, you need a Dialog (to define the conversation path and collect user input) and a Flow (specifically a Subflow or Action step within a Flow) to execute the backend logic, such as calling the Reset Password API or updating the user record. This combination allows the VA to guide the user and perform the action.

A Dialog is designed for multi-turn conversations. By chaining Prompt nodes, you can ask for each piece of information sequentially, storing the answers in variables, which creates a natural and guided user experience. This is the standard and most maintainable pattern for collecting structured data within a VA topic.

If the VA identifies the correct intent (topic) but gives the wrong answer, the issue lies within the topic’s configuration. The most common cause is that the topic’s training data (sample utterances) or its linked response (like a Knowledge Base article) does not accurately match the desired outcome. The VA is doing what it was trained to do, but the training was incorrect.

Virtual Agent Conversation Analytics is the dedicated, pre-built application for this purpose. It provides detailed reports on metrics like topic volume, containment rate, conversation paths, fallback rates, and user sentiment, offering the deepest insights into VA performance and user engagement.

The Quick Reply node presents users with clickable button options (e.g., ‘Approve’ and ‘Reject’) instead of requiring free-text input. This makes the interaction faster, more accurate, and more user-friendly for clear binary choices like an approval.

ServiceNow’s multilingual capability is built on the user’s sys_language preference and Translated Text fields. By setting the user’s language (manually or via location/business rule) and populating the Translated Text fields for topic/dialog prompts and responses, the VA will automatically serve content in the user’s preferred language from a single topic definition.

The Fallback Handler is a centralized configuration point that defines what happens when the VA cannot match the user’s input to any topic with sufficient confidence. This is where you would typically configure an action like creating an incident or offering a handoff to a live agent, ensuring it applies globally.

The ‘Embedded Widget’ deployment method is designed specifically for this purpose. It generates a snippet of JavaScript code that you can insert into the HTML of any external website. This script loads the VA chat interface directly into that page, making it the standard and supported integration method.

A Script Include is the platform standard for reusable server-side logic. By encapsulating the complex audit logging code in a Script Include, you can call it from the Action step (or Subflow) of any topic. This ensures consistency, centralizes maintenance, and follows ServiceNow development best practices.

To create an automated conversational flow, you need a Dialog (to define the conversation path and collect user input) and a Flow (specifically a Subflow or Action step within a Flow) to execute the backend logic, such as calling the Reset Password API or updating the user record. This combination allows the VA to guide the user and perform the action.0