Your AI Agents Are Starving. How to Feed Them-Value unlocked ITSM, CSM, HRSD, ITOM, SecOps, ITAM

[KelliA080304624]

Your AI Agents Are Starving. Here's How to Feed Them.

The three critical configurations per workflow that transform ServiceNow from work tracker to intelligence multiplier - Value unlocked ITSM, CSM, HRSD, ITOM, SecOps, ITAM, & SAM

Welcome to the AI Center of Excellence (CoE) team at ServiceNow. I'm Kelli Abbott, a former data scientist now focused on AI strategy and helping customers unlock the full potential of their AI investments.

Here's the irony of my career shift: I spent years as a data scientist finding and organizing good data to build mathematical models. Now? We have incredible models—but organizations are starving them of the data they need to deliver value.

The models are ready. Your data practices need to catch up. Let me show you how.

Your organization has invested in Now Assist and AI Agents. The technology works brilliantly. But adoption remains low because AI needs rich data to learn from, and most organizations are still capturing data like it's 2015—sparse descriptions, missing resolution notes, minimal context.

This guide provides the essential configurations that transform sparse work tracking into intelligent AI training across every ServiceNow workflow. These aren't complex customizations. They're strategic decisions to capture the data your out-of-the-box AI agents need to deliver value. Each configuration directly enables specific Now Assist skills or AI Agent capabilities.

The value to your team is immediate: faster resolutions, better knowledge sharing, less repetitive work. The value to your organization compounds over time: institutional knowledge that persists, junior agents performing like veterans, AI that gets smarter with every interaction.

Here's what I recommend: audit your current configurations against this guide. Identify the gaps. Build a plan to close them within 90 days. Even if you're not focused on demonstrating AI value today, you will be soon—and these foundational configurations determine whether AI becomes your competitive advantage or an expensive disappointment.

Love this?  Comment or thank.  Needs more? Comment. Missing something?  Share your secrets in the comments.

ITSM - IT Service Management

These three configurations enable Incident Summarization, Resolution Notes Generation, Knowledge Article Generation, and Suggested Steps Generation—the core AI capabilities that multiply your service desk's expertise.

1. Minimum Description Length

What: UI Policy requiring 200+ characters in incident description field

Where: System Policy > UI Policies > Create New

Why: Incident Summarization and Knowledge Article Generation need substantial context to generate useful content. Brief descriptions like 'printer broken' produce generic AI output that agents ignore. This configuration ensures AI has enough detail to create summaries and knowledge that actually help.

Value: Agents spend less time reading lengthy incidents because AI summaries capture key information accurately. Organizations see 300-400% increase in usable AI-generated knowledge articles. Teams get better incident visibility without reading every detail.

OCM: Frame positively: 'Provide detailed context to enable AI-assisted resolution for similar future incidents.' Show agents examples where their detailed descriptions became helpful summaries.

2. Mandatory Resolution Notes

What: Business Rule or UI Policy blocking incident closure without resolution notes

Where: Business Rules or UI Policy on incident form

Why: Resolution Notes Generation uses documented solutions as the foundation for Suggested Steps. Without resolution notes, AI cannot learn how your team solves problems, and future agents get no AI guidance on similar incidents. This is the critical link in the AI learning chain.

Value: Agents resolving similar incidents get step-by-step guidance based on your team's proven solutions. Resolution time for common issues drops 25-40%. Junior agents benefit from senior agents' documented expertise without constant interruptions.

OCM: Message as 'You're training AI to help your teammates solve this faster next time—document what worked!' Celebrate agents whose resolution notes become AI suggestions that help colleagues. Show concrete examples of agents saying 'This AI suggestion saved me 30 minutes.'

3. Structured Categorization

What: Mandatory Category, Subcategory, and Assignment Group fields before closure

Where: Data Policy or UI Policy on incident form

Why: AI-powered routing and triage agents need structured classification to identify patterns and route intelligently. The Triage Cases and Classify Tasks AI Agents depend on consistent categorization to learn which teams handle which issues and spot recurring problems.

Value: Incidents route to the right teams faster, reducing reassignments that frustrate agents and delay resolution. AI identifies recurring issues and recommends knowledge article creation. Assignment accuracy improves significantly, and teams spend less time on misdirected work.

OCM: Show agents how proper categorization helps them avoid the frustration of receiving irrelevant tickets. Highlight examples where AI routing saved hours by getting incidents to the right expert immediately. Frame as 'helping everyone work on what they're actually good at.'

CSM - Customer Service Management

These enable Case Summarization, Resolution Notes Generation, Knowledge Article Generation, and the Triage Cases AI Agent—capabilities that reduce agent workload while improving customer experience.

1. Email Parsing Enhancement

What: Configure inbound email actions to populate standard fields by extracting customer name, issue type, priority indicators, and key details from email content

Where: System Mailboxes > Inbound Actions > Configure email parsing rules

Why: Most customer cases arrive via email as unstructured text. Without parsing, AI sees only 'Case created from email' with no context. Email parsing transforms customer communications into structured data that feeds Case Summarization, enabling AI to provide useful summaries immediately instead of forcing agents to read entire email threads.

Value: Cases auto-created from email contain proper context from day one. Agents spend less time gathering basic information from customers. AI generates accurate case summaries that help supervisors understand issues quickly. First-contact resolution improves because agents have complete context immediately.

OCM: Show agents before-and-after examples of email-created cases. Highlight how parsed data means they don't have to ask customers to repeat information already in the email. Frame as 'AI does the tedious data extraction so you can focus on solving the customer's problem.'

2. Case Description Minimum

What: UI Policy requiring 250+ characters for case descriptions when cases are manually created

Where: System Policy > UI Policies > Create New for case table

Why: Case Summarization requires sufficient detail to generate meaningful summaries that supervisors and managers can actually use. Brief descriptions produce useless AI output that erodes trust in AI capabilities.

Value: Supervisors and stakeholders get accurate case summaries without reading entire case histories. Escalation decisions become faster and better informed. Customer handoffs between agents are smoother because AI provides clear case context.

OCM: Position as helping management understand cases quickly, freeing managers to support agents rather than chase down case details. Show managers how AI summaries let them spot problems and provide guidance faster.

3. Resolution Documentation Standards

What: Mandatory resolution code AND detailed resolution notes at case closure

Where: Business Rule or Data Policy on case table

Why: Resolution codes alone categorize outcomes but don't provide the how-to details AI needs. Detailed resolution notes combined with codes enable Knowledge Article Generation and power Suggested Steps. This combination creates the knowledge base that helps agents resolve future cases without searching external resources or interrupting colleagues.

Value: Knowledge base builds automatically from real customer resolutions. Future cases get AI-powered suggested solutions that agents can trust. First-contact resolution improves 15-20%. New agents ramp faster because they have access to proven resolution approaches.

OCM: Celebrate agents whose case resolutions become knowledge articles that help the team. Track 'knowledge created from your work' as a positive team metric. Show agents the knowledge articles generated from their cases and how colleagues used them successfully.

HRSD - HR Service Delivery

These enable HR Case Summarization, Knowledge Article Generation, and context-aware AI assistance that provides appropriate guidance based on employee situations.

1. Comprehensive Categorization

What: Detailed category and subcategory taxonomy covering Benefits, Payroll, Leave, Onboarding, Performance, Offboarding, and Policy Questions

Where: HR Service > Categories configuration

Why: HR AI Agents need precise categorization to provide relevant policy guidance and route cases correctly. Generic categories like 'HR Question' produce generic AI responses that don't help employees or agents. Detailed taxonomy enables AI to understand the specific HR domain and provide accurate answers.

Value: Employees get accurate answers to policy questions from AI self-service. HR agents spend less time on routine inquiries, focusing instead on complex employee situations requiring empathy and judgment. Case routing improves dramatically, getting sensitive issues to the right specialist immediately.

OCM: Show HR team how proper categorization enables employees to get instant answers for routine questions, freeing HR to focus on meaningful employee support—career guidance, difficult conversations, complex benefits counseling. Frame as 'AI handles the repetitive policy questions so you can do the work only humans can do.'

2. Guided Case Creation Forms

What: Role-specific forms with conditional fields that dynamically adjust based on case type, asking relevant questions for benefits, payroll, leave, etc.

Where: Service Catalog > Catalog Items configuration

Why: Guided forms ensure employees provide complete information upfront, reducing the back-and-forth that frustrates both employees and HR agents. This feeds Case Summarization with structured data and enables AI to provide better initial guidance.

Value: HR cases contain all necessary information from day one, eliminating rounds of 'can you provide your employee ID / start date / manager name' emails. Faster case resolution. Better employee experience. AI performs better with complete data.

OCM: Frame as improving employee experience—getting them answers faster by asking the right questions upfront. Show HR team the time saved when employees provide complete information initially instead of multiple follow-up exchanges.

3. Employee Journey Mapping

What: Link HR cases to employee lifecycle stages: onboarding, probation, role change, promotion, offboarding

Where: HR Case form configuration with lifecycle stage field and automation to populate based on employee record

Why: Context-aware AI provides different guidance to new hires versus tenured employees versus departing employees. Journey mapping enables AI to understand the employee's situation and provide relevant, timely support. A benefits question from someone in their first 30 days needs different handling than the same question from a 10-year employee.

Value: New employees get onboarding-specific guidance automatically. AI recognizes lifecycle transitions and provides proactive support. HR agents see employee context immediately without looking up tenure and history. Employee satisfaction improves because support feels personalized and relevant.

OCM: Highlight how this enables HR to provide personalized support at scale—something impossible without AI. Show examples where AI recognized an employee's lifecycle stage and provided perfectly timed guidance. Frame as 'AI helps you give every employee the attention they deserve.'

ITOM - IT Operations Management

These enable Investigate IT Problems, Analyze Incident Trends, and Generate Change Request Plans AI Agents—capabilities that transform reactive operations into proactive problem prevention.

1. Alert Enrichment with CMDB Context

What: Configure alert rules to automatically populate CI information, upstream/downstream dependencies, and business services impacted

Where: Event Management > Alert Rules configuration

Why: AI Agents cannot prioritize alerts without business context. A CPU spike on a test server differs vastly from one on a production payroll server processing time-sensitive transactions. CMDB enrichment provides the critical context AI needs to understand impact and urgency.

Value: AI prioritizes alerts by business impact automatically, so operations teams focus on critical issues first. Mean time to detect critical issues drops 25-30%. Alerts include dependency information, helping teams understand cascading impacts immediately. Operations staff spend less time researching 'what does this server do?' and more time fixing problems.

OCM: Show operations team examples where AI correctly prioritized a business-critical alert over noisy test environment alerts. Highlight saved time when alerts arrive with complete context instead of requiring manual CMDB lookups. Frame as 'AI does the research so you can focus on the fix.'

2. Event Correlation Rules

What: Configure correlation rules to group related events based on time windows, CI relationships, and event patterns

Where: Event Management > Correlation Rules

Why: AI Agents analyzing incident trends need correlated events to identify root causes. Individual events don't reveal patterns—correlation does. When five servers throw errors within 30 seconds, that's one problem, not five. AI needs correlation to understand this.

Value: AI identifies cascading failures and common root causes instead of treating symptoms individually. Operations teams see patterns instead of individual alerts, understanding 'network switch failure affecting 12 servers' not '12 separate server issues.' Problem prevention becomes possible through pattern recognition.

OCM: Share examples where correlated events revealed root causes that would have been missed analyzing individual alerts. Show the time saved investigating one correlated problem instead of five separate symptoms. Frame as 'AI connects the dots so you can solve root causes, not symptoms.'

3. Problem Documentation Standards

What: Mandatory problem statement, root cause analysis, and fix action fields with minimum 200 characters each

Where: Problem Management form configuration with required fields and character minimums

Why: Investigate Problems AI Agent uses historical problem-solution data to guide current investigations. Without detailed problem documentation capturing what was tried, what worked, and why issues occurred, AI cannot learn from past investigations. This robs your team of cumulative operational intelligence.

Value: AI suggests investigation approaches based on similar past problems, dramatically reducing diagnostic time. New problems get resolved faster using institutional knowledge instead of starting from scratch. Repeat problems decrease because root causes are documented and addressable. Operations builds a problem-solving knowledge base that persists beyond individual team members.

OCM: Position as 'building our operations playbook'—show how documented problems become AI guidance for future investigations. Celebrate operations engineers whose problem analysis helped teammates solve similar issues quickly. Frame as 'your troubleshooting expertise helping the whole team.'

SecOps - Security Operations

These enable intelligent triage, investigation automation, and threat prioritization—reducing analyst burnout while improving security posture.

1. Threat Intelligence Integration

What: Enrich security incidents with threat intelligence feeds providing IP reputation, known malware signatures, and threat actor information

Where: Security Incident Response > Threat Intelligence configuration and integration

Why: AI triage agents need context to distinguish genuine threats from benign activity. Without threat intelligence, every failed login looks the same. Threat intelligence integration provides the context AI needs to prioritize real threats over noise.

Value: Security analysts spend dramatically less time on false positives. AI correctly prioritizes genuine threats, enabling analysts to focus investigation effort where it matters. Investigation time drops 35-40%. Alert fatigue decreases as AI filters out known-benign activity.

OCM: Demonstrate false positive reduction—show security team the alerts AI correctly deprioritized, saving hours of investigation time. Highlight examples where threat intel helped AI identify serious threats faster. Frame as 'AI handles the noise so you focus on real threats.'

2. MITRE ATT&CK Framework Tagging

What: Tag security incidents with MITRE ATT&CK framework tactics and techniques

Where: Security Incident form configuration with MITRE ATT&CK taxonomy fields

Why: Standardized classification enables AI to identify attack patterns across seemingly unrelated incidents. Without common taxonomy, AI cannot learn effectively or connect the dots between reconnaissance activity on Monday and data exfiltration attempts on Friday.

Value: AI recognizes attack techniques and suggests proven countermeasures from your team's past responses. Security team benefits from industry-standard threat knowledge applied to your specific environment. Pattern detection improves dramatically—AI spots multi-stage attacks that individual analysts might miss.

OCM: Show how MITRE tagging enables AI to connect seemingly unrelated incidents into coordinated attack campaigns, helping analysts see the bigger picture. Highlight cases where AI pattern recognition revealed attacks that individual incident review wouldn't catch. Frame as 'AI sees patterns across all your incidents—you see the complete attack story.'

3. User Behavior Context

What: Link security incidents to user profiles including role, location, travel status, and normal behavior patterns

Where: Security Incident form with user profile integration and behavioral baseline data

Why: AI needs to understand normal behavior to identify genuine anomalies. User context dramatically reduces false positives by helping AI recognize traveling employees, legitimate admin activities, timezone differences, and expected unusual access patterns.

Value: Failed login alerts from traveling executives don't trigger unnecessary investigations. AI distinguishes legitimate unusual behavior from genuine security threats. Alert fatigue decreases significantly. Security team focuses on real threats instead of investigating normal business activity.

OCM: Share examples of legitimate activities AI correctly identified as non-threatening, avoiding wasted investigation time and 3am pages. Show analysts how user context prevents false alarm burnout. Frame as 'AI understands your users' normal patterns so you investigate real threats, not business travel.'

ITAM - IT Asset Management (I love ITAM, let me help you love it too!)

These enable accurate asset recommendations, procurement optimization, and lifecycle management—preventing costly duplicate purchases and ensuring efficient asset utilization.

1. Accurate Asset Lifecycle States

What: Implement clear, detailed asset states including In Stock, Deployed, In Repair, Reserved, Retired, and Disposed with required status updates

Where: Asset Management > Asset States configuration and workflow rules

Why: AI procurement recommendations depend on understanding what assets are actually available. Without accurate lifecycle states, AI recommends purchasing new laptops while 47 sit unused in a closet. Proper state tracking prevents costly ghost asset scenarios.

Value: AI prevents duplicate procurement by identifying available assets before recommending new purchases. Asset redeployment increases, reducing waste. Procurement decisions become data-driven. Finance sees dramatic reduction in unnecessary spending—often hundreds of thousands annually.

OCM: Show asset managers examples where accurate status prevented duplicate purchases. Quantify the waste avoided. Frame as 'AI helps you be heroes who prevent unnecessary spending and get CFO recognition for cost savings.'

2. Asset Location and Ownership Tracking

What: Mandatory accurate location (building, floor, room) and assigned user/department for all assets with regular validation

Where: Asset Management > Asset form required fields and Discovery integration

Why: AI asset recommendations and refresh planning require knowing where assets are and who's using them. Unknown locations and unassigned assets create blind spots where AI cannot provide useful guidance. Accurate tracking enables AI to optimize distribution and identify underutilized assets.

Value: IT can locate assets instantly when moves or issues occur. AI identifies underutilized assets for redeployment before buying new. Asset audits become dramatically faster and more accurate. Compliance improves because you can actually account for your assets.

OCM: Show IT team the time saved when asset location is current—no more building searches for 'lost' equipment. Highlight redeployment successes where AI identified available assets, saving procurement costs. Frame as 'accurate data means you find and redeploy instead of searching and buying.'

3. Asset-to-Contract Relationship Mapping

What: Link every asset to its purchase contract, warranty information, and support agreements with automated tracking of contract end dates

Where: Asset Management > Contract Management integration

Why: AI needs contract data to provide warranty coverage guidance and optimal refresh timing recommendations. Without these links, agents manually search contracts every time, and AI cannot suggest 'this asset is under warranty—use vendor support' or 'refresh cycle due—plan replacement.'

Value: AI automatically routes hardware issues to vendor support when under warranty, saving internal IT effort. Refresh planning becomes proactive based on warranty expiration dates. Contract renewals happen strategically instead of reactively. Agents get instant warranty status without manual contract lookups.

OCM: Show IT team examples where AI directed them to vendor support for warranty-covered issues, saving hours of internal troubleshooting. Highlight avoided costs from missed warranty coverage. Frame as 'AI knows your contracts so you leverage every dollar of warranty value.'

SAM - Software Asset Management

These enable license optimization, compliance management, and renewal planning—often saving organizations hundreds of thousands in unnecessary software costs.

1. Actual Usage Data Collection

What: Implement automated tracking of actual software usage—logins, active users, feature utilization—not just installations

Where: Software Asset Management > Usage tracking and Discovery integration

Why: AI optimization recommendations require knowing actual usage, not just license counts. Organizations often pay for 1,000 licenses while only 340 users actively use the software. Without usage data, AI cannot identify these optimization opportunities that represent significant wasted spend.

Value: AI identifies unused licenses for reclamation or non-renewal, often saving hundreds of thousands annually. True-up costs decrease because you right-size before vendor audits. Procurement decisions become data-driven instead of guesswork. CFO sees immediate ROI from software spend optimization.

OCM: Show SAM team the specific dollar amounts saved through AI-identified optimization opportunities. Quantify unused license costs. Frame as 'AI finds the money hiding in your software spend—you get credit for massive cost savings that make finance love you.'

2. Complete Contract Data

What: Capture licenses purchased, renewal dates, auto-renewal terms, price per license, and vendor contact information for every software contract

Where: Software Asset Management > Contract Management

Why: AI renewal optimization requires both usage data AND contract terms. Usage data shows who's using what. Contract data shows what you're paying and when renewals occur. Together, these enable AI to recommend 'reduce from 1,000 to 400 licenses at renewal, saving $660K annually.'

Value: AI provides renewal recommendations months in advance with specific cost impact. Auto-renewal contracts get flagged before costly unwanted renewals occur. Negotiation power improves because you have usage data supporting right-sizing requests. Compliance violations decrease because AI monitors license counts against actual usage.

OCM: Show SAM managers examples of AI-flagged auto-renewals that would have wasted budget. Quantify negotiation wins enabled by usage data. Frame as 'AI arms you with facts for vendor negotiations and prevents surprise renewals that drain budget.'

3. License-to-User Entitlement Mapping

What: Map licenses to actual users, departments, and cost centers with automated tracking of license assignments and reclamations

Where: Software Asset Management > Entitlement configuration

Why: AI cannot optimize license allocation without knowing who has licenses, who's actually using them, and which departments are over or under-licensed. Entitlement mapping enables AI to recommend specific reclamation and reallocation actions.

Value: AI identifies specific licenses to reclaim from inactive users and reassign to those needing access, avoiding new purchases. Chargeback accuracy improves because costs map to actual usage by department. License requests get fulfilled from existing pool before purchasing new. ROI from license optimization becomes directly attributable.

OCM: Show SAM team concrete examples of AI-recommended reclamations that fulfilled license requests without purchasing. Highlight department cost allocation accuracy improvements. Frame as 'AI optimizes your license pool so you fulfill requests instantly from existing licenses instead of waiting on procurement.'

The Value These Configurations Unlock

ServiceNow's AI products are on pace to exceed $500M in ACV by end of 2025. Enterprises using properly configured systems report AI handling 70% of inquiries. These aren't future possibilities—they're current realities at organizations that feed their AI quality data.

The pattern across successful implementations is consistent: configure for data quality, frame as team enablement, demonstrate value through examples, measure impact, scale based on results.

Your AI agents are ready. The technology works. Now give them the data they need to multiply your team's expertise across your entire organization.

Start with three configurations in one workflow. Prove the value to your team. Show them AI helping with real work. Then scale. Your agents are waiting to learn—give them the data they need to make your entire organization smarter, faster, and more efficient.

The configuration work is straightforward. The organizational change requires intention. The results transform how your teams work. Your AI investment is ready to deliver—start feeding the brain today.

Note: These are only my personal views and ideas. 

For implementation support, contact your ServiceNow account team or visit the AI Center of Excellence resources in the ServiceNow Community.