# Section 07: Skill Kit Implementation - Incident Summarization with OpenAI

**Estimated time: 30 minutes**

## Overview

The Now Assist Skill Kit enables you to create custom AI-powered skills that extend ServiceNow's native AI capabilities. In this section, we'll build a custom skill that uses OpenAI to automatically summarize incident records, providing agents with quick, actionable insights.

### What We'll Build

A custom skill that:

* **Analyzes incident records** automatically
* **Generates concise summaries** using OpenAI
* **Extracts key information** like impact, urgency, and resolution steps
* **Integrates seamlessly** with ServiceNow workflows
* **Provides consistent formatting** for agent consumption

### Business Value

**For IT Agents:**

* Faster incident understanding and resolution
* Consistent summary format across all incidents
* Reduced time spent reading lengthy descriptions

**For Organizations:**

* Improved MTTR (Mean Time to Resolution)
* Better knowledge capture and sharing
* Enhanced agent productivity and satisfaction

## Prerequisites

* Completed previous sections (8.1-8.3)
* OpenAI integration configured and tested
* Access to Now Assist Skill Kit
* Administrative privileges for skill creation
* Sample incident records for testing

## Step 1: Access Skill Kit Interface

### Navigate to Skill Kit

1. Navigate to **All > Now Assist > Skill Kit**

![Skill Kit Access](/files/A8DI6JERV9PjIa4cwMej)

### Explore Available Templates

2. **Review available skill templates** and select the appropriate base template for incident summarization

The Skill Kit provides pre-built templates for common use cases, which can be customized for your specific needs.

## Step 2: Create New Incident Summarization Skill

### Initialize Skill Creation

![Skill Creation Interface](/files/QfdPI9AuXF29CwxkzyAs)

3. **Create a new skill** with the following configuration:

**Basic Skill Information:**

* **Name:** "OpenAI Incident Summarizer"
* **Description:** "Automatically generates concise summaries of incident records using OpenAI"
* **Category:** IT Service Management
* **Version:** 1.0.0

### Configure Skill Parameters

![Skill Configuration](/files/6sgbPfsURo4ciXP5hAnz)

**Input Parameters:**

* **Incident Number:** Required field for incident identification
* **Summary Type:** Option to choose summary style (brief, detailed, technical)
* **Include History:** Boolean to include/exclude work notes history

**Output Parameters:**

* **Summary Text:** Generated incident summary
* **Key Points:** Bulleted list of critical information
* **Recommended Actions:** AI-suggested next steps
* **Confidence Score:** AI confidence level in the summary

## Step 3: Design the Skill Logic

### Configure AI Processing Logic

![Skill Logic Design](/files/TPEasiHY9X2J5qBAeOpt)

**Core Processing Steps:**

1. **Data Extraction:**

   ```javascript
   // Extract incident data
   var incident = new GlideRecord('incident');
   if (incident.get(input.incident_number)) {
       var incidentData = {
           number: incident.number.toString(),
           short_description: incident.short_description.toString(),
           description: incident.description.toString(),
           category: incident.category.toString(),
           priority: incident.priority.getDisplayValue(),
           state: incident.state.getDisplayValue(),
           assigned_to: incident.assigned_to.getDisplayValue(),
           work_notes: getWorkNotes(incident.sys_id)
       };
   }
   ```
2. **OpenAI Prompt Construction:**

   ```javascript
   // Build comprehensive prompt for OpenAI
   var prompt = `
   Analyze the following ServiceNow incident and provide a concise summary:

   Incident: ${incidentData.number}
   Title: ${incidentData.short_description}
   Description: ${incidentData.description}
   Category: ${incidentData.category}
   Priority: ${incidentData.priority}
   Current State: ${incidentData.state}
   Assigned To: ${incidentData.assigned_to}

   Work Notes History:
   ${incidentData.work_notes}

   Please provide:
   1. A 2-3 sentence executive summary
   2. Key technical details (bullet points)
   3. Current status and progress
   4. Recommended next actions
   5. Potential risks or escalation triggers

   Format the response in a clear, structured manner for IT support agents.
   `;
   ```
3. **Response Processing:**

   ```javascript
   // Process OpenAI response
   var aiResponse = callOpenAIAPI(prompt);
   var summary = {
       executive_summary: extractSection(aiResponse, 'executive_summary'),
       key_details: extractBulletPoints(aiResponse),
       current_status: extractSection(aiResponse, 'status'),
       recommendations: extractBulletPoints(aiResponse, 'recommendations'),
       confidence: calculateConfidence(aiResponse)
   };
   ```

### Error Handling and Fallbacks

![Error Handling Configuration](/files/tocgWzJu4uuArk85JXdm)

**Robust Error Handling:**

```javascript
// Error handling for AI service issues
try {
    var aiSummary = generateAISummary(incidentData);
    return formatResponse(aiSummary);
} catch (error) {
    // Fallback to template-based summary
    gs.warn('AI summarization failed: ' + error.message);
    return generateTemplateSummary(incidentData);
}

function generateTemplateSummary(data) {
    return {
        summary: `Incident ${data.number}: ${data.short_description}`,
        details: `Priority: ${data.priority}, Category: ${data.category}`,
        status: `Current state: ${data.state}`,
        recommendations: ["Review incident details", "Check for similar incidents", "Follow standard procedures"]
    };
}
```

## Step 4: Configure Integration Points

### Workflow Integration

Configure the skill to integrate with existing ServiceNow workflows:

**Trigger Points:**

* **Incident Creation:** Auto-summarize new incidents
* **Incident Update:** Re-summarize when significant updates occur
* **On-Demand:** Manual trigger by agents
* **Scheduled:** Batch processing for reporting

### UI Integration

**Agent Interface Enhancements:**

* **Summary Panel:** Display AI summary in incident form
* **Quick Actions:** One-click summarization button
* **History Tracking:** Maintain summary version history
* **Feedback Collection:** Allow agents to rate summary quality

## Step 5: Testing and Validation

### Unit Testing

Test the skill with various incident scenarios:

**Test Case 1: Simple Incident**

```
Incident: INC0000123
Description: "User cannot access email"
Expected: Brief, focused summary with standard troubleshooting steps
```

**Test Case 2: Complex Incident**

```
Incident: INC0000456
Description: Multiple system failures with extensive work notes
Expected: Comprehensive summary highlighting key events and current status
```

**Test Case 3: Escalated Incident**

```
Incident: INC0000789
Description: Critical system outage with multiple stakeholders
Expected: Executive-level summary with clear escalation status
```

### Integration Testing

Verify the skill works properly within ServiceNow:

**Workflow Testing:**

* Trigger mechanisms work correctly
* Output integrates properly with incident forms
* Error handling gracefully manages failures
* Performance meets acceptable thresholds

## Step 6: Deployment and Configuration

### Production Deployment

![Deployment Configuration](/files/bZD8heduiAUNHjH9STr4)

**Deployment Checklist:**

✅ **Security Review:** Validate data handling and API security ✅ **Performance Testing:** Confirm acceptable response times ✅ **User Training:** Prepare documentation and training materials ✅ **Monitoring Setup:** Configure logging and alerting ✅ **Rollback Plan:** Prepare contingency procedures

### Configuration Management

**Environment-Specific Settings:**

* **Development:** Full logging, test API endpoints
* **Staging:** Production-like configuration with test data
* **Production:** Optimized performance, minimal logging

**Access Controls:**

* **Skill Execution:** Limit to authorized roles
* **API Access:** Secure OpenAI credentials
* **Data Access:** Respect incident visibility rules

## Advanced Features and Enhancements

### Multi-Language Support

```javascript
// Detect incident language and adjust prompt
var userLanguage = detectLanguage(incidentData.description);
var localizedPrompt = getPromptTemplate(userLanguage);
```

### Context-Aware Summarization

```javascript
// Adjust summary based on incident context
var summaryStyle = determineSummaryStyle(incidentData);
// Executive summary for P1 incidents
// Technical summary for development teams
// User-friendly summary for end-user communications
```

### Learning and Improvement

```javascript
// Collect feedback for continuous improvement
function collectSummaryFeedback(incidentId, summaryId, rating, comments) {
    var feedback = new GlideRecord('x_custom_ai_feedback');
    feedback.initialize();
    feedback.incident = incidentId;
    feedback.summary_id = summaryId;
    feedback.rating = rating;
    feedback.comments = comments;
    feedback.insert();
}
```

## Monitoring and Analytics

### Performance Metrics

Track key performance indicators:

**Operational Metrics:**

* **Response Time:** Average time to generate summary
* **Success Rate:** Percentage of successful summarizations
* **Error Rate:** Frequency and types of failures
* **API Usage:** OpenAI token consumption and costs

**Quality Metrics:**

* **Agent Satisfaction:** User ratings of summary quality
* **Usage Adoption:** Frequency of skill utilization
* **Time Savings:** Reduction in incident review time
* **Accuracy:** Correlation between summaries and actual resolutions

### Continuous Improvement

**Monthly Reviews:**

* Analyze user feedback and ratings
* Review error logs and failure patterns
* Optimize prompts based on performance data
* Update skill logic for better accuracy

**Quarterly Enhancements:**

* Expand to additional record types
* Integrate new AI capabilities
* Enhance integration points
* Scale to additional use cases

## Security and Compliance

### Data Privacy

**Sensitive Information Handling:**

* **PII Detection:** Automatically detect and mask personal information
* **Data Minimization:** Only process necessary incident data
* **Retention Policies:** Implement appropriate data retention
* **Audit Logging:** Track all AI processing activities

### API Security

**OpenAI Integration Security:**

* **Key Rotation:** Regular API key updates
* **Rate Limiting:** Prevent abuse and manage costs
* **Error Handling:** Avoid exposing sensitive error information
* **Network Security:** Secure API communication channels

## 🎉 Skill Kit Implementation Complete!

You have successfully:

* ✅ **Created a custom AI skill** using the Now Assist Skill Kit
* ✅ **Integrated OpenAI** for intelligent incident summarization
* ✅ **Implemented robust error handling** and fallback mechanisms
* ✅ **Configured workflow integration** for seamless operation
* ✅ **Established monitoring** and continuous improvement processes
* ✅ **Applied security best practices** for data protection

### Key Benefits Achieved

🤖 **Intelligent Automation:** AI-powered incident analysis and summarization ⚡ **Improved Efficiency:** Faster incident understanding and resolution 📊 **Consistent Quality:** Standardized summary format across all incidents 🔄 **Scalable Architecture:** Framework for expanding to additional use cases 📈 **Measurable Impact:** Clear metrics for ROI demonstration

### Skill Kit Capabilities Unlocked

**Current Implementation:**

* Incident summarization with OpenAI
* Workflow integration and automation
* Error handling and fallback procedures
* Performance monitoring and analytics

**Future Expansion Opportunities:**

* **Change Request Summarization:** Apply to change management
* **Knowledge Article Generation:** Auto-create KB articles from incidents
* **Predictive Analysis:** Identify potential issues before they escalate
* **Multi-Record Analysis:** Summarize related incident patterns

***

**Next Section:** [Section 8 - Virtual Agent Designer with LLM](https://github.com/rajatrakesh/DataAILab/blob/main/section8-virtual-agent-designer.md) **Previous Section:** [Section 6 - Gen AI Controller Config](https://github.com/rajatrakesh/DataAILab/blob/main/section6-genai-controller-config.md) **Back to:** [Main README](/dataailab/readme.md)


---

# Agent Instructions: Querying This Documentation

If you need additional information that is not directly available in this page, you can query the documentation dynamically by asking a question.

Perform an HTTP GET request on the current page URL with the `ask` query parameter:

```
GET https://servicenow-data-and-ai.gitbook.io/dataailab/section07-skill-kit-with-open-ai.md?ask=<question>
```

The question should be specific, self-contained, and written in natural language.
The response will contain a direct answer to the question and relevant excerpts and sources from the documentation.

Use this mechanism when the answer is not explicitly present in the current page, you need clarification or additional context, or you want to retrieve related documentation sections.