Enhance Discovery With Artifical Intelligence?

[Tom Rausch]

What are some ways I can enhance ServiceNow Discovery with features provided by artifical intelligence?

[Tom Rausch]

GitHub Copilot indicates it can analyze ServiceNow Discovery Logs. I have not tested this capability, but apparently the capability is present.

[Tom Rausch]

I asked GitHub Copilot, "How can GitHub Copilot help me automte operations for ServiceNow Discovery?"

Here is the response from GitHub Copilot.

GitHub Copilot

GitHub Copilot can help automate operations for ServiceNow Discovery in several ways, primarily by assisting you in writing scripts, integrations, or workflows that interact with ServiceNow’s Discovery APIs and tools.

Here’s how Copilot can be beneficial.

Script Automation

ServiceNow Discovery often requires writing custom scripts (in PowerShell, Python, or JavaScript) to automate tasks such as triggering discoveries, collecting configuration items (CIs), processing discovery results, or handling errors.

Copilot can assist by:

• Suggesting code snippets for ServiceNow REST API calls.
• Completing Python or PowerShell scripts to automate scheduled discoveries.
• Providing boilerplate code for processing and parsing discovery results.

Workflow Integration

If you need to integrate ServiceNow Discovery with DevOps tools (like GitHub Actions or Jenkins):

• Copilot can generate workflow files for orchestrating ServiceNow discovery via REST or MID Servers.
• Copilot can create scripts to fetch discovery statuses, schedule new scans, or update CMDB from discovery results.

Error Handling & Logging

Automating error handling and logging is crucial for reliable operations:

• Copilot can help write robust try-catch logic to handle various discovery API errors and log them in a suitable format.
• It can suggest logging strategies compatible with ServiceNow and external monitoring tools.

API Integration

Most ServiceNow automation requires interacting with ServiceNow REST APIs (e.g., ServiceNow Discovery API documentation)

• Copilot can help generate code to authenticate, query, and manipulate discovery jobs and results.
• Copilot can prompt best practices for pagination, token management, and result parsing.

Custom Discovery Patterns

If you’re developing custom Discovery Patterns:

• Copilot can assist with creating XML or JSON definitions and mapping logic, streamlining configuration and troubleshooting.

Infrastructure as Code

For organizations using IaC:

• Copilot can help automate CMDB updates based on infrastructure changes, ensuring they are discovered and recorded in ServiceNow.

Example

Suppose you want to automate discovery job creation via ServiceNow’s REST API in Python:

import requests

url = "https://yourinstance.service-now.com/api/now/table/discovery_job"
headers = {
"Content-Type": "application/json",
"Authorization": "Bearer YOUR_TOKEN"
}
payload = {
"name": "Weekly Network Discovery",
"type": "network",
"schedule": "weekly"
}

response = requests.post(url, json=payload, headers=headers)
print(response.json())

create_discovery_job.py

Copilot can autocomplete, suggest improvements, and even generate documentation or README files for these operations.

Summary

GitHub Copilot helps automate ServiceNow Discovery operations by expediting script development, integration with external tools, error handling, API usage, and pattern creation—making IT operations more efficient and less error-prone.

[Christopher Hub]

I have done pretty heavy testing of GPT, CP, Claude, and Grok for parsing and triage of automation_error_message table, and it's a rich mess of hallucination despite quite a bit of training effort.  I similarly find that AI hallucinates heavily in the design of discovery solutions because it can't span knowledge of platform, knowledge of target devices, and discovery strategy.  I'd love to hear that someone else has found traction, but so far I'm not seeing it.  You can get very specific help with tactical issues from AI, but it falls on its face with complex work.