Agent Assist Setup
Agent Assist Setup has transformed how ServiceNow environments handle support interactions. ServiceNow® Now Assist using generative AI is purpose designed to enhance user productivity and efficiency through conversation and proactive experiences, improving ticket resolution times by up to 40%.
ServiceNow’s agentic AI agents autonomously manage IT service requests, incidents, and processes, improving efficiency and reducing manual intervention. This innovative feature leverages artificial intelligence to streamline workflows and enhance the overall service experience. Here are comprehensive insider tips and best practices for testing and continuously refining its performance to ensure impactful positive Agent Assist effectiveness. Embrace the future of service management and transformative impact of AI-powered Agent Assisted service experience.
What are the key foundations of AI Enabling Agent Assist?
The leading AI technologies behind empowering a powerful Agent Assist Experience include:
- Natural Language Processing (NLP): This technology helps AI understand and interpret human language, enabling it to analyze customer queries and provide relevant responses.
- Machine Learning (ML): ML algorithms analyze historical data and ongoing interactions to improve the accuracy and relevance of AI-generated suggestions over time.
- Real-Time Data Analytics: This allows AI to process and analyze data in real-time, providing instant recommendations and support to agents during customer interactions.
- Speech Recognition: This technology transcribes customer calls and identifies keywords and phrases, helping agents quickly access relevant information.
- Knowledge Base Integration: AI tools integrate with knowledge bases and customer service systems, providing quick access to articles, FAQs, and troubleshooting steps.
- Sentiment Analysis: AI performs sentiment analysis to gauge customer emotions and suggest appropriate responses, enhancing the quality of customer interactions.
Why Focus on Agent Assist?
Optimizing ServiceNow with intelligent features propels organizations to the forefront of operational efficiency. Specifically, Agent Assist:
- Provides Immediate Recommendations: Agents receive in-context knowledge articles, past cases, and resolution steps at a glance.
- Cuts Down Learning Curves: New agents can handle complex tickets faster, thanks to dynamic suggestions.
- Elevates Customer Satisfaction: Quick answers and resolutions reduce frustration, improving overall service ratings.
These outcomes resonate with an industry-wide push toward AI-empowered solutions, especially as service desks handle more incidents and user demands for faster support intensify.
Configuring Agent Assist
To take full advantage of this feature, follow these best practices:
1. Activate the Required Plugins
- Predictive Intelligence (com.glide.platform_ml) must be active for ML-driven suggestions.
- Agent Assist (com.sn_agent_assist) relies on the trained models from Predictive Intelligence.
- Confirm that your ITSM Pro or equivalent license permits access to these add-ons.
2. Train Your Machine Learning Models
- Data Preparation: Gather well-labeled historical incidents and knowledge articles.
- Solution Definition: Define classification and similarity models tailored to your environment.
- Model Training: Run model training within Predictive Intelligence. Monitor performance metrics such as accuracy and confidence scores.
3. Configure Agent Workspace
- Enable the Agent Assist Component: Ensure it’s visible in the workspace layout.
- Set Filters and Rules: Restrict or enable certain article types, levels of confidence, or business rules for recommended results.
4. Test and Validate
- Sample Incidents: Create test tickets with varied subjects and categories.
- Check Accuracy: Ensure recommendations align with known solutions.
- Refine and Retrain: Adjust solution definitions or reclassify data for improved results.
Continuous Improvement and Best Practices
Fine-Tune Knowledge Base
A curated knowledge base boosts relevancy. Regularly update content, retire obsolete articles, and gather feedback from agents to improve coverage and clarity. AI-driven Knowledge Article search optimization uses Predictive Intelligence to analyze historical incidents and knowledge base content, improving search relevancy, to surface relevant articles and resolved incidents automatically.
Show Similar Resolved Incidents Feature:
Similarity Matching predicts which past resolved incidents can help solve the current issue. The agent selects “Similar Resolved Incidents” from the Agent Assist dropdown for quick solutions. When an agent selects “Similar Resolved Incidents” from the Agent Assist dropdown, Similarity Matching predicts which past resolved incidents can help solve the current issue. This AI-driven approach eliminates guesswork, ensuring agents quickly find the best solutions.
Agent Assist also identifies the most capable Assignment Groups or retrieves past Incident Resolutions, empowering agents to review, copy, or link solutions for faster resolution. This feature significantly boosts first-call resolution rates, helping service teams close tickets up to 40% faster.

Here is how Predictive Intelligence helps agents resolve tickets efficiently while reducing research time. From this query of Similar Resolved Incidents Agent Assist will apply Similarity Matching to predict which past resolved incidents can help solve the current issue. The agent selects “Similar Resolved Incidents” from the Agent Assist dropdown to quickly find relevant solutions.
Agent Assist can identify the best Assignment Groups for resolving similar issues or retrieve past Incident Resolutions or empower first call resolution advantages to review, copy, or link for faster resolution. Predictive Intelligence streamlines ticket resolution by instantly searching relevant past incidents and knowledge articles, reducing research time by up to 35%.

Monitor Metrics
Staying informed through dashboards that track results can help you spot issues, and opportunity to improve high performance Agent Assisted Experience:
- Accuracy Scores
- Resolution Times
- Knowledge Article Views
Analyzing and tuning using a data driven experience give you context to quickly spot issues and opportunities to evolve and maintain high performance.
Out-of-the-box Similarity Matching in ServiceNow’s Agent Assist may not always return relevant results, especially if the word corpus used for predictions lacks proper training. If AI suggests unrelated incidents, agents spend more time searching instead of resolving.
The root cause? The model might be using generic, unfiltered incident data rather than a focused, high-quality word corpus. Below, we outline two tuning techniques—using a focused Paragraph-Vector word corpus and leveraging a pre-trained GloVe word corpus—to improve prediction accuracy and help agents find the most relevant solutions faster.
Comparison Table: Tuning Similar Resolved Incidents
Solution | Navigation | Selection | Retraining Method |
---|---|---|---|
Default (OOB) Similarity Model | Uses Incidents (last 6 months) word corpus | Generic, may not include resolved incidents | Standard model training |
Tuning Tip #1: Focused Paragraph-Vector Corpus | Go to Predictive Intelligence > Word Corpus > New | Select Paragraph-Vector and filter for Resolved Incidents | Retrain model with refined dataset |
Tuning Tip #2: Pre-Trained GloVe Corpus | Go to Predictive Intelligence > Word Corpus > New | Select Pre-Trained (GloVe) for broader AI context | No need to define corpus content; just retrain |
Final Tip: When to Use Each Approach
- Use Paragraph-Vector when your incidents contain industry-specific terms that a general AI model won’t recognize (e.g., technical jargon).
- Use GloVe when you need broad similarity matching across a large, diverse dataset.
- Consider TF-IDF if your incidents involve machine-generated content like log files or error messages.
Test different approaches, retrain your models, and monitor AI predictions to ensure Agent Assist delivers the most accurate, relevant solutions. For further learning, explore the “Accelerate Incident Resolution with Predictive Intelligence” course on Now Learning.
Conclusion
Implementing and fine-tuning Agent Assist Setup propels your ServiceNow platform toward faster resolutions, heightened user satisfaction, and sustainable growth. By activating the right plugins, training robust machine learning models, and continuously refining your knowledge base, you create an intelligent system that supports agents at every step. Seamless integration and diligent monitoring provide a dynamic service environment where each ticket becomes an opportunity to deliver remarkable support experiences.
Invest in these best practices to remain adaptable, efficient, and responsive as organizational demands evolve. Begin your Agent Assist journey today and witness the impact of transformative AI-driven service management.
Other Agent Assist Setup Resources
The next steps, watch for Agent Assist AI transformation using Workflow Data Fabric:
- Advanced settings for Machine Learning solutions
- Advantages of natural language models over keywords
- AI-powered Service Management Operations
- AI Powered Career Resilience
- AI and Improved Knowledgebase-Search
- Build A Knowledge Management Culture To Increase… | Forrester
- Customer Service Knowledge Base– 26 Best Practices
- Configure Sentiment Analysis (servicenow.com)
- Consortium for Service Innovation site.
- Customer Case Study: Multi-Agent AI Collaboration with ServiceNow and Microsoft Semantic Kernel
- Exploring Virtual Agent (Developers Resources for getting started)
- How generative AI affects highly skilled workers | MIT Sloan
- Jobs N Career Success Network
- Now Assist Genius Results
- Now Assist Guardian
- Now Assist in AI Search
- Now Assist in Virtual Agent.
- ServiceNow Agent Assist Setup Documentation
- ServiceNow platform introduces AI Agent Orchestrator – ServiceNow Press
- Set up Agent Assist
- Tuning Predictive Intelligence Models (part 3) – I… – ServiceNow Community
- Using Now Assist Analytics
