Rpa Bot Generation

RPA Bot Generation

Overview

This skill generates Robotic Process Automation (RPA) bots in ServiceNow RPA Hub from natural language process descriptions. It covers:

• Translating business process descriptions into structured bot definitions
• Creating bot actions for UI automation, data extraction, and system interaction
• Configuring triggers to launch bots from ServiceNow workflows, schedules, or events
• Implementing error handling strategies with retry logic, exception capture, and fallback paths
• Managing credentials securely for bot access to target applications
• Organizing bots into robot groups for workload distribution and high availability
• Monitoring bot execution with performance analytics and failure tracking

When to use:

• When automating repetitive manual processes identified in process mining
• Building bots to bridge legacy systems that lack APIs
• When ServiceNow Integration Hub spokes are not available for a target application
• Automating data entry, extraction, or validation across desktop applications
• Creating attended bots for agent-assisted automation scenarios

Prerequisites

• Roles: sn_rpa.admin, sn_rpa.designer, sn_rpa.operator, or admin
• Plugins: com.sn_rpa (RPA Hub), com.sn_rpa.designer (RPA Desktop Design Studio)
• Access: Read/write access to sn_rpa_bot, sn_rpa_action, sn_rpa_process tables
• Knowledge: Understanding of RPA concepts (attended vs. unattended bots, selectors, recording), target application UI structure, and your organization's automation governance framework
• Infrastructure: RPA robot machines provisioned with ServiceNow RPA agent installed

Key RPA Tables

Procedure

Step 1: Analyze the Process Description

Parse the user's process description to identify:

• Process scope: What the bot should automate end-to-end
• Target applications: Which systems the bot will interact with
• Input data: What data the bot needs to start (spreadsheet, record, queue)
• Actions: Sequential steps (click, type, read, navigate, extract, validate)
• Output: What the bot produces (updated records, reports, confirmations)
• Exceptions: Known error scenarios and expected handling

Step 2: Create the Process Definition

Register the business process in RPA Hub.

Using MCP (Claude Code/Desktop):

Tool: SN-Create-Record
Parameters:
  table_name: sn_rpa_process
  fields:
    name: "Invoice Data Entry Automation"
    description: "Automates entry of approved invoice data from ServiceNow into the legacy ERP system. Process handles single invoice entry including vendor lookup, line item entry, and approval submission."
    category: finance
    owner: [process_owner_sys_id]
    state: design
    estimated_savings: 2400
    complexity: medium
    manual_steps: 15
    avg_manual_time_minutes: 12
    volume_per_month: 200

Using REST API:

POST /api/now/table/sn_rpa_process
Content-Type: application/json

{
  "name": "Invoice Data Entry Automation",
  "description": "Automates entry of approved invoice data from ServiceNow into the legacy ERP system.",
  "category": "finance",
  "owner": "[process_owner_sys_id]",
  "state": "design",
  "estimated_savings": "2400",
  "complexity": "medium"
}

Step 3: Create the Bot Definition

Define the bot linked to the process.

Using MCP:

Tool: SN-Create-Record
Parameters:
  table_name: sn_rpa_bot
  fields:
    name: "Invoice Entry Bot"
    description: "Unattended bot that enters approved invoices into the legacy ERP system"
    state: draft
    bot_type: unattended
    process: [process_sys_id]
    version: "1.0"
    owner: [bot_owner_sys_id]
    robot_group: [robot_group_sys_id]
    max_retries: 3
    timeout_minutes: 30

Using REST API:

POST /api/now/table/sn_rpa_bot
Content-Type: application/json

{
  "name": "Invoice Entry Bot",
  "description": "Unattended bot for legacy ERP invoice entry",
  "state": "draft",
  "bot_type": "unattended",
  "process": "[process_sys_id]",
  "version": "1.0",
  "max_retries": "3",
  "timeout_minutes": "30"
}

Step 4: Define Bot Actions

Create sequential actions that form the bot's workflow.

Action 1: Open target application:

Tool: SN-Create-Record
Parameters:
  table_name: sn_rpa_action
  fields:
    bot: [bot_sys_id]
    action_type: open_application
    order: 100
    name: "Launch ERP Application"
    target_application: "SAP GUI"
    input_data: '{"executable_path": "C:\\Program Files\\SAP\\SAPgui\\saplogon.exe", "connection": "PRD"}'
    timeout_seconds: 30
    error_handling: retry
    retry_count: 2

Action 2: Login with managed credentials:

Tool: SN-Create-Record
Parameters:
  table_name: sn_rpa_action
  fields:
    bot: [bot_sys_id]
    action_type: credential_login
    order: 200
    name: "Login to ERP"
    credential: [credential_sys_id]
    selector: '{"username_field": "#txtUser", "password_field": "#txtPassword", "login_button": "#btnLogin"}'
    timeout_seconds: 15
    error_handling: fail
    error_message: "ERP login failed - check credentials in vault"

Action 3: Navigate and enter data:

Tool: SN-Create-Record
Parameters:
  table_name: sn_rpa_action
  fields:
    bot: [bot_sys_id]
    action_type: data_entry
    order: 300
    name: "Enter Invoice Header"
    target_application: "SAP GUI"
    input_data: |
      {
        "transaction": "FB60",
        "fields": {
          "vendor": "{invoice.vendor_number}",
          "invoice_date": "{invoice.invoice_date}",
          "posting_date": "{invoice.posting_date}",
          "amount": "{invoice.total_amount}",
          "currency": "{invoice.currency}",
          "reference": "{invoice.sn_number}"
        }
      }
    timeout_seconds: 20
    error_handling: screenshot_and_fail

Action 4: Validate and submit:

Tool: SN-Create-Record
Parameters:
  table_name: sn_rpa_action
  fields:
    bot: [bot_sys_id]
    action_type: validation
    order: 400
    name: "Validate Entry and Submit"
    input_data: |
      {
        "validation_checks": [
          {"field": "status_bar", "expected": "Document posted", "selector": "#statusBar"},
          {"field": "document_number", "not_empty": true, "selector": "#docNumber"}
        ],
        "on_success": "capture_document_number",
        "on_failure": "screenshot_and_rollback"
      }
    output_variable: erp_document_number
    timeout_seconds: 15

Action 5: Update ServiceNow record:

Tool: SN-Create-Record
Parameters:
  table_name: sn_rpa_action
  fields:
    bot: [bot_sys_id]
    action_type: servicenow_update
    order: 500
    name: "Update Invoice Record in ServiceNow"
    input_data: |
      {
        "table": "ap_invoice",
        "sys_id": "{invoice.sys_id}",
        "fields": {
          "erp_reference": "{erp_document_number}",
          "state": "posted",
          "work_notes": "Bot: Invoice posted to ERP. Document: {erp_document_number}"
        }
      }

Step 5: Configure Credential Management

Set up secure credential storage for bot authentication.

Using MCP:

Tool: SN-Create-Record
Parameters:
  table_name: sn_rpa_credential
  fields:
    name: "ERP Service Account - Invoice Bot"
    application: "SAP ERP"
    credential_type: username_password
    username: svc_rpa_invoice
    vault_id: [credential_vault_id]
    rotation_policy: 90_days
    owner: [security_team_sys_id]

Step 6: Set Up Bot Triggers

Configure when and how the bot executes.

Using MCP:

Tool: SN-Create-Record
Parameters:
  table_name: sn_rpa_trigger
  fields:
    bot: [bot_sys_id]
    trigger_type: flow
    flow: [flow_sys_id]
    condition: "invoice.state == 'approved' && invoice.erp_reference == ''"
    description: "Trigger bot when invoice is approved and not yet posted to ERP"
    active: true
    priority: normal

Schedule-based trigger:

Tool: SN-Create-Record
Parameters:
  table_name: sn_rpa_trigger
  fields:
    bot: [bot_sys_id]
    trigger_type: schedule
    schedule: "0 */2 * * 1-5"
    description: "Run every 2 hours on weekdays to process queued invoices"
    active: true
    max_batch_size: 50

Step 7: Monitor Bot Execution

Query execution logs for performance and error tracking.

Using MCP:

Tool: SN-Query-Table
Parameters:
  table_name: sn_rpa_execution
  query: bot=[bot_sys_id]^ORDERBYDESCstart_time
  fields: sys_id,bot,state,start_time,end_time,duration,status,error_message,robot,transactions_processed
  limit: 50

Generate execution analytics:

Tool: SN-Execute-Background-Script
Parameters:
  script: |
    var analytics = {
      period: 'Last 30 days',
      total_executions: 0,
      successful: 0,
      failed: 0,
      avg_duration_seconds: 0,
      total_transactions: 0,
      success_rate: 0,
      common_errors: {},
      time_saved_hours: 0
    };

    var ga = new GlideAggregate('sn_rpa_execution');
    ga.addQuery('start_time', '>=', gs.daysAgo(30));
    ga.addAggregate('COUNT');
    ga.addAggregate('AVG', 'duration');
    ga.addAggregate('SUM', 'transactions_processed');
    ga.query();
    if (ga.next()) {
      analytics.total_executions = parseInt(ga.getAggregate('COUNT'));
      analytics.avg_duration_seconds = Math.round(parseFloat(ga.getAggregate('AVG', 'duration')) || 0);
      analytics.total_transactions = parseInt(ga.getAggregate('SUM', 'transactions_processed')) || 0;
    }

    var success = new GlideAggregate('sn_rpa_execution');
    success.addQuery('start_time', '>=', gs.daysAgo(30));
    success.addQuery('status', 'success');
    success.addAggregate('COUNT');
    success.query();
    if (success.next()) {
      analytics.successful = parseInt(success.getAggregate('COUNT'));
      if (analytics.total_executions > 0)
        analytics.success_rate = Math.round((analytics.successful / analytics.total_executions) * 100);
    }
    analytics.failed = analytics.total_executions - analytics.successful;

    // Estimate time saved (12 min manual per transaction)
    analytics.time_saved_hours = Math.round((analytics.total_transactions * 12) / 60);

    gs.info('RPA EXECUTION ANALYTICS:\n' + JSON.stringify(analytics, null, 2));
  description: "RPA: Generate bot execution analytics for last 30 days"

Tool Usage

Best Practices

• Process First: Always document the manual process completely before building the bot; poor process understanding is the top cause of bot failure
• Modular Actions: Break bots into small, reusable actions; do not create monolithic bots with 100+ steps
• Error Screenshots: Configure screenshot capture on failure for every critical action to aid debugging
• Credential Rotation: Use the credential vault with automatic rotation; never hardcode passwords in bot definitions
• Idempotency: Design bots to be safely re-runnable; check if work is already done before processing
• Throttling: Add delays between actions to avoid overwhelming target applications; respect rate limits
• Robot Groups: Distribute workload across multiple robots for high-volume processes
• Monitoring Alerts: Set up notifications for bot failures, especially for business-critical processes
• Version Control: Increment bot versions for every change; maintain rollback capability
• Human-in-the-Loop: For high-risk processes (financial, compliance), add human approval steps before final submission

Troubleshooting

Bot Fails to Launch Target Application

Symptom: First action fails with "application not found" or "timeout" Cause: Robot machine does not have the target application installed, or the path is incorrect Solution: Verify the executable path on the robot machine. Check that the RPA agent service is running. Ensure the robot user has permission to launch the application.

Selector Not Found Errors

Symptom: Bot cannot locate UI elements in the target application Cause: Application UI changed (update, patch, or dynamic element IDs) Solution: Re-record selectors using RPA Desktop Design Studio. Use stable selectors (name, automationId) over dynamic ones (runtime IDs). Add wait conditions before element interactions.

Credential Vault Access Denied

Symptom: Bot cannot retrieve credentials from the vault Cause: Robot service account lacks vault access, or credential record is inactive Solution: Verify the robot's service account has sn_rpa.credential_user role. Check that the credential record state is active and assigned to the correct robot group.

Bot Execution Queue Backlog

Symptom: Bot executions are queuing but not starting Cause: All robots in the group are busy, offline, or at max concurrent execution limit Solution: Check robot availability in sn_rpa_robot_group. Increase max_concurrent if resources allow. Add more robots to the group for horizontal scaling.

Examples

Example 1: Employee Onboarding Bot

Scenario: Automate new employee account creation across 3 systems

Process definition:

• Input: Approved HR onboarding request from ServiceNow
• Step 1: Create Active Directory account
• Step 2: Create email mailbox in Exchange
• Step 3: Provision access in legacy HR system
• Step 4: Update ServiceNow onboarding task as complete

Example 2: Report Generation Bot

Scenario: Generate and distribute weekly compliance reports from legacy system

Tool: SN-Create-Record
Parameters:
  table_name: sn_rpa_bot
  fields:
    name: "Weekly Compliance Report Bot"
    description: "Logs into compliance portal, generates report, downloads PDF, attaches to ServiceNow task"
    bot_type: unattended
    state: draft
    version: "1.0"

Trigger: Scheduled every Monday at 6:00 AM

Related Skills

• genai/flow-generation - Flow Designer for orchestrating RPA with other automation
• genai/playbook-generation - Process Automation playbooks that trigger bots
• genai/spoke-generation - Integration Hub spokes as API-first alternative to RPA
• admin/workflow-creation - Workflow integration with RPA triggers
• development/scheduled-jobs - Scheduling bot executions