2.4.4. Virtual Agent

💡 First Principle: Providing an intelligent, conversational interface empowers users with immediate support and enables automated request fulfillment, significantly enhancing the self-service experience and reducing live agent workload.

Scenario: Your IT help desk is overwhelmed with repetitive questions like "How do I reset my password?" or "What's the Wi-Fi password?" You want to implement a chatbot to handle these common inquiries.

The Virtual Agent is ServiceNow's conversational AI chatbot, designed to provide immediate assistance to users by understanding natural language requests. The fundamental 'why' of implementing a Virtual Agent is to enhance the self-service experience and deflect common, repetitive inquiries away from live agents. It provides 24/7 support, speeds up resolution for routine issues, and allows human agents to focus on more complex, high-value interactions that require human judgment.

Key concepts and functionalities of the Virtual Agent:

• Conversational Interface: Users interact with the Virtual Agent through text-based chat, typically within the Service Portal.
• Topics: Pre-built or custom-built conversations that the Virtual Agent can handle. Each topic guides the user through a specific process.
  • Topic Types:
    • Problem Resolution: Guides users through troubleshooting steps for common issues (e.g., "My printer isn't working").
    • Request Fulfillment: Helps users submit requests (e.g., "Request a new software").
    • Information Lookup: Directs users to relevant knowledge articles or FAQs (e.g., "How do I update my address?").
• NLU (Natural Language Understanding): The Virtual Agent uses NLU to interpret user input and match it to the most relevant topic. Training the NLU model with various user utterances is crucial for its effectiveness.
• Integrations: Virtual Agent integrates seamlessly with:
  • Knowledge Management: Directs users to relevant knowledge articles.
  • Service Catalog: Guides users through submitting catalog items.
  • Live Agent Chat: Can seamlessly transfer conversations to a human agent if the Virtual Agent cannot resolve the issue, ensuring no request is left unaddressed.
  • Flow Designer: Virtual Agent topics are often built using Flow Designer to orchestrate the conversation flow and trigger backend actions.
• Pre-built Topics: ServiceNow provides many out-of-the-box topics for common ITSM and HRSD scenarios, accelerating implementation.
• Custom Topic Creation: Administrators or developers can create custom topics using Flow Designer to handle unique organizational requests.
• Metrics: Virtual Agent usage, deflection rates, and successful conversation rates can be tracked to measure its effectiveness.

Implementing and configuring Virtual Agent topics is a key administrative task that directly contributes to reducing the support burden on service desks and improving the overall efficiency of self-service. It's a strategic investment in improving the user experience and optimizing service delivery.

💡 Tip: Start with implementing Virtual Agent topics for the most common, simple, and repetitive questions or requests that your service desk receives. These provide the quickest wins and highest deflection rates. Continuously review user feedback and conversation logs to identify areas for improvement and new topic development.

⚠️ Common Pitfall: Implementing a Virtual Agent without sufficient training data for NLU or without a clear escalation path to a live agent. This leads to user frustration and a poor self-service experience.

Key Trade-Offs:

• Automation (Chatbot) vs. Human Touch (Live Agent): Virtual Agent provides immediate, scalable support for common issues, while live agents handle complex or sensitive inquiries.

Reflection Question: How does the Virtual Agent enhance the self-service experience and contribute to reducing the workload on live support agents?

📋 EXAM QUICK FACTS: Virtual Agent

⚠️ CRITICAL DISTINCTION: Virtual Agent vs. Agent Assist

Exam Trap: Agent Assist is NOT a chat interface. It provides suggestions to human agents.

NLU Concepts:

Key Points:

• After modifying utterances, you must publish the NLU model
• Low NLU confidence triggers fallback behavior (clarification or escalation)
• Best practice fallback: Transfer to Live Agent with context

Topic Types:

• Problem Resolution - Troubleshooting guides
• Request Fulfillment - Submit catalog requests
• Information Lookup - Find KB articles

Virtual Agent Designer: Tool for creating/configuring topics

Deployment Channels:

• Service Portal Web Chat
• Microsoft Teams
• Slack
• Mobile App