happy-technologies-llc/build-agent

Build Custom AI Agents

Overview

This skill provides a comprehensive guide to building custom AI agents in ServiceNow:

• Defining agent capabilities and scope of responsibility
• Configuring knowledge sources for agent grounding (KB articles, catalog items, documentation)
• Setting up tool access so agents can perform actions (create records, query data, trigger flows)
• Designing conversational flows with intent recognition and entity extraction
• Implementing guardrails for safety, accuracy, and compliance
• Testing and iterating on agent behavior before production deployment

When to use: When building a new virtual agent topic, extending Now Assist skill capabilities, creating a domain-specific AI assistant, or configuring agent tool access for automated task execution.

Prerequisites

• Roles: admin, va_admin, or now_assist_admin
• Plugins: com.glide.cs.chatbot (Virtual Agent), com.snc.now_assist (Now Assist) recommended
• Access: Write access to sys_cs_topic, sys_cs_intent, sys_cs_entity, sys_cb_topic, kb_knowledge tables
• Knowledge: Understanding of conversational AI concepts (intents, entities, slots, fulfillment)
• Related Skills: genai/agent-miner for discovering what to automate, genai/skill-kit-custom for Now Assist skills

Procedure

Step 1: Define Agent Scope and Capabilities

Document the agent's purpose, target users, and boundaries.

=== AGENT DEFINITION ===
Name: [Agent Name]
Domain: [IT Support / HR / Facilities / Custom]
Target Users: [Employee / Customer / Technician]

CAPABILITIES (What the agent CAN do):
- [ ] Answer questions about [topic]
- [ ] Create [record type] on behalf of users
- [ ] Look up [information type]
- [ ] Trigger [workflow/process]
- [ ] Escalate to [team] when needed

BOUNDARIES (What the agent CANNOT do):
- [ ] Access [restricted data/system]
- [ ] Approve [request type]
- [ ] Modify [critical records]
- [ ] Share [sensitive information]

SUCCESS METRICS:
- Deflection Rate Target: [%]
- User Satisfaction Target: [score]
- Resolution Rate Target: [%]
- Average Handle Time Target: [minutes]

Step 2: Configure Knowledge Sources

Set up the knowledge base articles and data sources the agent will use for grounding.

MCP Approach:

Tool: SN-Query-Table
Parameters:
  table_name: kb_knowledge
  query: kb_knowledge_base=[KB_SYS_ID]^workflow_state=published^active=true
  fields: sys_id,number,short_description,kb_category,article_type
  limit: 100

REST Approach:

GET /api/now/table/kb_knowledge
  ?sysparm_query=kb_knowledge_base=[KB_SYS_ID]^workflow_state=published^active=true
  &sysparm_fields=sys_id,number,short_description,kb_category,article_type
  &sysparm_limit=100

Verify knowledge coverage for each agent capability:

| Capability | KB Articles | Gaps Identified | |-----------|------------|-----------------| | Password reset | KB001, KB002 | None | | VPN setup | KB010 | Missing macOS instructions | | Software request | None | Need catalog mapping article |

Step 3: Create Agent Topic

Build the primary topic that defines the agent's conversational scope.

MCP Approach:

Tool: SN-Create-Record
Parameters:
  table_name: sys_cs_topic
  data:
    name: "IT Support Assistant"
    description: "Handles common IT support requests including password resets, VPN access, and software installations"
    state: draft
    category: IT
    greeting_message: "Hi! I'm your IT Support Assistant. I can help with password resets, VPN access, software requests, and general IT questions. What do you need help with?"
    no_match_message: "I'm not sure I understand. Could you rephrase your request, or would you like me to connect you with a live agent?"
    max_retries: 3

REST Approach:

POST /api/now/table/sys_cs_topic
Body: {
  "name": "IT Support Assistant",
  "description": "Handles common IT support requests",
  "state": "draft",
  "category": "IT",
  "greeting_message": "Hi! I'm your IT Support Assistant...",
  "no_match_message": "I'm not sure I understand...",
  "max_retries": "3"
}

Step 4: Define Intents and Training Utterances

Create intents that the agent should recognize and respond to.

MCP Approach:

Tool: SN-Create-Record
Parameters:
  table_name: sys_cs_intent
  data:
    name: "Reset Password"
    topic: [TOPIC_SYS_ID]
    description: "User wants to reset their password for any system"
    training_utterances: |
      I need to reset my password
      My password expired
      I can't log in
      Password not working
      Forgot my password
      Need a new password
      Locked out of my account
      Account locked
      Can't access my account
      Reset credentials

Repeat for each capability:

| Intent | Sample Utterances | Entity Slots | |--------|------------------|--------------| | Reset Password | "reset my password", "locked out" | system_name, username | | VPN Access | "connect to VPN", "VPN not working" | vpn_type, device_type | | Software Install | "install software", "need [app]" | software_name, justification | | General Question | "how do I...", "what is..." | topic_keyword |

Step 5: Define Entities for Slot Filling

Create entities to extract structured data from user messages.

MCP Approach:

Tool: SN-Create-Record
Parameters:
  table_name: sys_cs_entity
  data:
    name: "system_name"
    topic: [TOPIC_SYS_ID]
    type: "list"
    prompt: "Which system do you need the password reset for?"
    required: true
    values: |
      Active Directory|AD|Windows login
      Email|Outlook|Exchange
      VPN|Cisco AnyConnect
      SAP|ERP
      Salesforce|CRM

REST Approach:

POST /api/now/table/sys_cs_entity
Body: {
  "name": "system_name",
  "topic": "[TOPIC_SYS_ID]",
  "type": "list",
  "prompt": "Which system do you need the password reset for?",
  "required": "true"
}

Step 6: Configure Tool Access and Actions

Define what actions the agent can perform in ServiceNow.

MCP Approach:

Tool: SN-Create-Record
Parameters:
  table_name: sys_cb_action
  data:
    topic: [TOPIC_SYS_ID]
    name: "Create Incident for Password Reset"
    action_type: "create_record"
    table: "incident"
    field_values: |
      category=inquiry
      subcategory=password
      short_description=Password reset request: ${system_name}
      assignment_group=IT Help Desk
      caller_id=${sys_user}
      urgency=3
      impact=3

REST Approach:

POST /api/now/table/sys_cb_action
Body: {
  "topic": "[TOPIC_SYS_ID]",
  "name": "Create Incident for Password Reset",
  "action_type": "create_record",
  "table": "incident",
  "field_values": "category=inquiry^subcategory=password^short_description=Password reset request"
}

Common Action Types:

| Action Type | Purpose | Configuration | |------------|---------|---------------| | create_record | Create a new record | table, field_values | | update_record | Update existing record | table, sys_id, field_values | | query_record | Look up records | table, query, return_fields | | run_flow | Trigger a Flow Designer flow | flow_sys_id, inputs | | run_script | Execute server-side script | script body | | kb_search | Search knowledge base | query, kb_base | | escalate | Transfer to live agent | queue, message |

Step 7: Implement Guardrails

Configure safety boundaries and compliance controls.

Input Guardrails:

Tool: SN-Create-Record
Parameters:
  table_name: sys_cs_topic_goal
  data:
    topic: [TOPIC_SYS_ID]
    name: "Input Validation"
    type: "guardrail"
    configuration: |
      BLOCKED_INTENTS:
      - Requests for other users' personal data
      - Attempts to bypass approval processes
      - Requests for admin/root access

      INPUT_SANITIZATION:
      - Strip HTML/script tags from user input
      - Validate employee ID format before lookup
      - Limit free-text input to 500 characters

      AUTHENTICATION:
      - Require verified session for record creation
      - Verify caller identity before sharing account details
      - MFA challenge for sensitive operations

Output Guardrails:

RESPONSE RULES:
- Never expose sys_id values or internal table names to users
- Never share other employees' personal information
- Always confirm destructive actions before executing
- Include disclaimer for estimated timeframes
- Redirect compliance/legal questions to appropriate teams
- Log all agent actions for audit trail

ESCALATION TRIGGERS:
- User expresses frustration (sentiment < -0.5)
- 3 consecutive failed intent matches
- Request involves VIP or executive user
- Security-related request (account compromise)
- Agent confidence score below threshold (< 0.7)

Step 8: Design Conversation Flows

Map out the full conversational flow for each intent.

CONVERSATION FLOW: Password Reset

[User Message] -> Intent Recognition -> "Reset Password" (confidence > 0.8)
  |
  v
[Slot: system_name] -> "Which system?" -> User provides system
  |
  v
[Slot: username] -> "What is your username?" -> User provides username
  |
  v
[Validation] -> Verify user exists in system
  |-- User found -> Proceed
  |-- User not found -> "I couldn't find that username. Please verify and try again."
  |
  v
[Action] -> Trigger password reset flow
  |-- Success -> "Your password has been reset. Check your email for the temporary password."
  |-- Failure -> "I wasn't able to reset your password. Let me connect you with a live agent."
  |
  v
[Follow-up] -> "Is there anything else I can help you with?"

Step 9: Test the Agent

Validate agent behavior with test scenarios.

MCP Approach:

Tool: SN-Query-Table
Parameters:
  table_name: sys_cs_topic
  query: sys_id=[TOPIC_SYS_ID]
  fields: sys_id,name,state,test_status,last_tested

Test Matrix:

| Test Case | Input | Expected Behavior | Result | |-----------|-------|-------------------|--------| | Happy path | "Reset my AD password" | Collect username, trigger reset | | | Missing entity | "Reset my password" | Ask which system | | | Unknown intent | "What's the weather?" | No-match response | | | Guardrail trigger | "Reset John's password" | Refuse, explain why | | | Escalation | "This isn't working, get me a human" | Transfer to live agent | | | Edge case | Empty message | Prompt for input | |

Step 10: Deploy and Monitor

Publish the agent and set up monitoring.

MCP Approach:

Tool: SN-Update-Record
Parameters:
  table_name: sys_cs_topic
  sys_id: [TOPIC_SYS_ID]
  data:
    state: published

Post-Deployment Monitoring Queries:

Tool: SN-Query-Table
Parameters:
  table_name: sys_cs_conversation
  query: topic=[TOPIC_SYS_ID]^sys_created_on>=javascript:gs.daysAgo(7)
  fields: sys_id,state,live_agent,transfer_reason,user_satisfaction,sys_created_on
  limit: 100

Tool Usage

| Tool | Purpose | When to Use | |------|---------|-------------| | SN-Query-Table | Find existing topics, intents, and knowledge articles | Discovery and validation | | SN-Create-Record | Create topics, intents, entities, and actions | Building agent components | | SN-Update-Record | Modify agent settings, publish topics | Configuration and deployment | | SN-Get-Table-Schema | Discover topic and intent table structures | Initial setup | | SN-Read-Record | Get detailed record for specific topic or intent | Reviewing existing configurations |

Best Practices

1. Start narrow, expand gradually -- launch with 2-3 well-defined intents before adding more
2. Use at least 10 training utterances per intent for reliable recognition
3. Test with real user language -- avoid technical jargon in training data
4. Implement graceful fallbacks -- always provide a path to live agent assistance
5. Monitor and retrain weekly during the first month after deployment
6. Version your topics -- maintain draft versions while published version serves users
7. Separate concerns -- one topic per domain area (IT, HR, Facilities)
8. Log all actions -- maintain audit trail for compliance and debugging
9. Set confidence thresholds -- do not act on low-confidence intent matches
10. Gather user feedback -- use post-conversation surveys to identify improvement areas

Troubleshooting

| Issue | Cause | Resolution | |-------|-------|------------| | Agent not recognizing intents | Insufficient or poor training utterances | Add more diverse utterances; include typos and abbreviations | | Wrong intent matched | Overlapping utterances between intents | Review and deduplicate training data; increase confidence threshold | | Entity not extracted | Entity values do not match user input | Add synonyms and variations to entity value list | | Action fails silently | Missing permissions or invalid field mapping | Check agent user roles; validate field names against table schema | | Agent loops on same question | Required slot not being filled by user response | Add skip logic or alternative slot-filling prompts | | High escalation rate | Agent scope too narrow or responses not helpful | Expand knowledge sources; review escalated conversations for patterns |

Examples

Example 1: IT Support Bot

Scenario: Build a virtual agent for common IT support requests.

AGENT: IT Support Assistant
INTENTS:
1. Password Reset -> Flow: Reset Password (AD, Email, VPN)
2. Software Request -> Action: Create RITM from catalog
3. VPN Troubleshooting -> KB Search: VPN articles
4. Hardware Issue -> Action: Create incident, escalate if P1
5. General IT Question -> KB Search: IT knowledge base

GUARDRAILS:
- Cannot reset passwords for other users
- Software requests require manager approval flow
- Hardware replacements require asset verification

Example 2: HR Benefits Bot

Scenario: Build an agent for benefits enrollment questions.

AGENT: Benefits Assistant
INTENTS:
1. Benefits Overview -> KB: Benefits summary article
2. Enrollment Period -> KB: Open enrollment dates and process
3. Plan Comparison -> KB: Plan comparison matrix
4. Life Event Change -> Action: Create HR case for life event
5. Dependent Add/Remove -> Action: Create HR case with dependent info

GUARDRAILS:
- Cannot share other employees' benefits information
- Directs medical questions to benefits provider
- Logs all interactions for compliance

Example 3: Facilities Service Bot

Scenario: Build an agent for workplace service requests.

AGENT: Facilities Assistant
INTENTS:
1. Room Booking -> Action: Query room availability, create reservation
2. Maintenance Request -> Action: Create facilities incident
3. Badge Access -> Action: Create access request RITM
4. Parking -> KB: Parking policy and waitlist info
5. Catering -> Action: Create catering request with details

GUARDRAILS:
- Room bookings limited to 4 hours maximum
- Maintenance requests auto-prioritized by type
- Badge access requires security team approval

Related Skills

• genai/agent-miner - Discover automation opportunities from interaction data
• genai/flow-generation - Build flows that agents can trigger
• genai/playbook-generation - Create playbooks for agent-assisted workflows
• genai/skill-kit-custom - Build custom Now Assist skills
• knowledge/article-generation - Generate KB articles for agent knowledge sources