Why Use NLU?
Virtual Agent becomes more powerful when it understands what users mean, not just what they type. Natural Language Understanding (NLU) enables intent-based topic recognition, allowing Virtual Agent to identify user goals and respond intelligently instead of relying on rigid keyword matching.
Introduction
Natural Language Understanding (NLU) enables ServiceNow to interpret free-text user input by identifying the user’s intent (what they want to do) and extracting entities (key details required to complete the action). This allows Virtual Agent to respond accurately, reduce unnecessary follow-up questions, and provide a more natural conversational experience.
NLU Terminologies
- Intent: An intent represents the user’s goal or action they want to perform, such as opening an incident, reporting an email issue, or resetting a password.
- Entity: An entity is a value or object in a user’s sentence that provides context and data needed to complete the intent, such as device, application, urgency, date, or location.
- Utterance: Utterances are the different ways users express the same intent using natural language. They help train the NLU model to accurately recognize user intent.
Plugins Required
- NLU Workbench – Advanced Features (Licensed)
- Glide Virtual Agent
Configuration Process
1. Enable NLU in Conversational Interfaces Settings.
Navigate to Conversational Interfaces → Settings and enable Natural Language Understanding.
2. Navigate to NLU Workbench → Models and create a new NLU model tailored to your use case, such as handling IT-related issues like email access problems.
3. Open the created NLU model and select View Phase under the Build and Train section to configure intents, add utterances and entities, and train the model.
4. Create new intents based on your business requirements, such as Report Email Issue.
5.Add multiple utterances for each intent and define entities where required. Entities help improve intent prediction accuracy and enable automatic data population.
Example Utterances:
Email is not working
Outlook is down
Not receiving emails
6. Attach the NLU Model to a Virtual Agent Topic
Navigate to Virtual Agent → Topics, open the required topic (for example, Email Issue), and select the created NLU model in the Properties tab under the NLU section.
7. Test the Configuration
Test the configuration using the portal or Virtual Agent Preview. The appropriate topic will now be triggered based on the user’s intent and meaning, rather than relying on keyword-based matching.
Conclusion:
By implementing NLU in Virtual Agent, organizations can move from keyword-driven interactions to intent-based conversations. Understanding user intent and extracting relevant entities enables Virtual Agent to select the right topic, capture context automatically, and reduce unnecessary user prompts. This results in faster resolutions, more natural and efficient user experience.
