Lilly Tech Systems
Intermediate
Virtual Agent
Design and deploy conversational AI experiences that enable self-service, deflect tickets, and resolve common issues automatically across IT, HR, and customer service.
What is Virtual Agent?
Virtual Agent is ServiceNow's conversational AI platform for building chatbot experiences. It uses NLU (Natural Language Understanding) to interpret user intent and guides users through structured conversations called Topics to resolve issues or fulfill requests.
Now Assist + Virtual Agent: With Now Assist integration, Virtual Agent can provide generative AI responses grounded in knowledge articles, going beyond scripted conversations to handle novel questions naturally.
Topic Design
Topics are the building blocks of Virtual Agent conversations:
| Component | Purpose |
|---|---|
| NLU Model | Classify user intent from natural language input |
| Topic Blocks | Flow-like visual designer for conversation logic |
| Entities | Extract specific data (dates, names, ticket numbers) from messages |
| Script actions | Execute backend logic, API calls, and record operations |
| Live agent handoff | Seamlessly transfer to human agent with full context |
Pre-Built Conversations
- ITSM topics: Password reset, VPN issues, software requests, incident status check
- HR topics: PTO balance, benefits enrollment, payroll questions, onboarding tasks
- CSM topics: Order status, return requests, account inquiries, appointment scheduling
- Catalog fulfillment: Browse and order from service catalog directly in chat
Channel Deployment
- Service Portal: Embedded chat widget in ServiceNow portals
- Microsoft Teams: Native Teams integration for employee self-service
- Slack: ServiceNow bot for Slack workspace integration
- Mobile: ServiceNow mobile app with Virtual Agent chat
- Custom channels: API integration for any messaging platform
Measuring Success
- Deflection rate: Percentage of conversations resolved without human agent
- Containment rate: Percentage of users who stay in the virtual agent (don't abandon)
- Topic accuracy: How often the correct topic is matched to user intent
- CSAT: Post-conversation satisfaction surveys for quality measurement
Key takeaway: Virtual Agent drives ticket deflection and faster resolution. Start with high-volume, well-documented use cases (password resets, status checks), measure deflection rates, and continuously improve NLU accuracy.