ITOM MCP Server is now in ChatGPT. Your ops context just went global.

[YuvalMamka]

NOW AVAILABLE • OCTOBER 28, 2026

ServiceNow is now in ChatGPT.
Your ops context just went global.

ChatGPT orchestrates. ServiceNow stays the governed system of action where the work gets done.

Your teams already think out loud in ChatGPT. Now they can act there too.

The ServiceNow connector for ChatGPT is live in the OpenAI marketplace. Connect ChatGPT to your own ServiceNow instance, choose which capabilities to turn on, and your people can investigate issues and take the next step in the same conversation. For operations teams, that includes the ITOM MCP Server and its AIOps, Service Reliability Management, and LEAP tools.

SOUND FAMILIAR?

It's 2 a.m. in Austin. A payment service starts throwing alerts. Tel Aviv is just logging on. Bangalore chased this exact pattern yesterday, but the notes are buried in someone's chat history.

Your people aren't the problem. The handoffs are. Every shift change, context leaks out.

SEE IT IN ACTION

Meet Priya. Customers can't check out.

Watch her go from page to cause to fix to automation in minutes, all from one conversation in ChatGPT.

00:04: What if your whole company could get work done just by asking?

00:08: ServiceNow is coming to ChatGPT.

00:11: With one plugin, companies can trigger, automate, and orchestrate enterprise

00:15: workflows across IT, CRM,

00:18: employee experiences, security, and more.

00:21: It only sees what you allow it to see, and every action is tracked.

00:26: Meet Priya, the engineer on call.

00:28: She's just been paged for a high-priority incident: customers can't check out.

00:32: ServiceNow has already linked the alerts, so she only has to ask what happened. In

00:39: seconds the problem is right in front of her, surfacing live data from ServiceNow.

00:44: What

00:45: else is affected? A

00:48: live map shows everything connected to the failing service. But

00:53: what's causing it?

00:57: ServiceNow reads the alerts and points to the likely cause.

01:04: Who should apply the fix?

01:07: It suggests the right team with the notes already written.

01:10: Priya approves and it's done.

01:12: Now for the big ask.

01:14: Priya has a hunch this alert will happen again, so she asks for this to be automated

01:19: next time.

01:20: ServiceNow builds the workflow, ready for her team to review.

01:23: From alert to answer to automation in minutes.

01:27: With ServiceNow Action Fabric and ChatGPT,

01:30: automating your business is one conversation away.

ONE INCIDENT. THREE TIME ZONES.

Zero lost context.

Austin asks ChatGPT, "What's going on with the payment service?" and gets the alert story, the impact, and the likely cause.

Tel Aviv picks up the thread hours later. The acknowledgment and work notes already live on the alert in ServiceNow.

Bangalore notices the pattern keeps coming back and turns it into an automation candidate.

Nobody starts from scratch. Everybody starts from the same page.

Not another API wrapper

Our MCP tools are built around outcomes, not raw tables. They cut down the read, query, reason, retry loop that happens when an AI client works directly against generic table access.

Better outcomes

Purpose-built tools return more relevant context and apply domain logic, so the AI client gets to more accurate answers.

Fewer tokens

Each tool returns focused context for the job instead of broad table results, so the model reads less and spends less.

Fewer calls, faster results

Domain logic runs inside the tool, which means less back-and-forth between the AI client, APIs, and data.

WHY THE TOKEN BILL GOES DOWN

~52%

lower AI client cost

72%

faster execution

~6.5x

fewer output tokens

Source: ServiceNow Launches Out-of-the-Box MCP Servers for Claude, Copilot, Gemini, and Other AI Clients

What your ops team can do from ChatGPT

The ITOM MCP Server brings three tool sets into the conversation.

AIOPS

Know what's happening. Then act.

• Alert analysis and investigation
• Impact analysis in business terms
• Hypothesizer for where an alert is headed
• Remediation, acknowledge, and update alerts

SERVICE RELIABILITY MANAGEMENT

Protect what customers feel.

• CI reliability status and topology
• Incident-linked reliability impact
• One-request SLO creation for any CI

LEAP

Stop fixing the same thing twice.

• Automation opportunities from repetitive work
• Setup and configuration checks

Governance stays on the ServiceNow AI Platform

ChatGPT decides which tool to call. ServiceNow controls what that tool can see and do. Every request runs in the user's context, through the same platform controls you've already built and trust.

ACL and role enforcement

Access follows the signed-in user's permissions. MCP doesn't bypass ServiceNow ACLs.

Server-side orchestration

Domain logic, workflow steps, and validations run on ServiceNow, not reconstructed by the model.

Governed action

Actions stay in the ServiceNow workflow layer, with full platform visibility and controls.

Your team stops stitching screens together and starts fixing what matters. AI does the gathering. Your people make the calls. That's what it looks like to put AI to work for people, in every time zone.

Choose your AI client. Keep the work on ServiceNow.

Find the ServiceNow connector in the OpenAI marketplace, connect it to your instance, and tell us how it goes:

• What would your team ask ChatGPT first?
• Which handoff hurts most today?
• What tool should we build next?

Drop your answers in the comments. We read every one.

Yuval Mamka, Product Manager, ITOM AIOps, ServiceNow

© 2026 ServiceNow, Inc. All rights reserved.