Now Assist vs. the ChatGPT Tab Your Employees Already Have Open

I'm going to say the quiet part loud: most of your employees are already using AI at work, and it isn't yours.

It's a browser tab. They paste the ticket, the error message, the half-written email into a consumer chatbot, copy the answer back, and move on. It's fast, it's free, and from a security standpoint it's a slow-motion data breach. This is the real competitive landscape Now Assist lives in, not other ITSM vendors, but the tab everyone already has open.

So here's the question that actually matters: why would anyone use the embedded AI when the external one writes just as well?

The answer is not "ours writes better." It probably doesn't. The answer is action and context. A consumer chatbot can write a beautiful paragraph about how to reset a VPN. It cannot see that this specific user is in the EMEA org, on a managed device, with an open incident from yesterday, and then actually reset the thing while respecting the permissions that say it's allowed to. Now Assist can, because it sits on top of what ServiceNow now calls the "system of action", the governed layer where work actually executes.

And here's the genuinely interesting move: ServiceNow opened that system of action to any agent through a generally available MCP Server. So even if your team prefers Claude or Copilot, those agents can now reach into ServiceNow and take governed, logged, permission-aware actions. The differentiator stopped being the model. It became the safe, auditable hands.

Let me give you a clean decision rule, because that's what you actually came for.

Use the external tab when the task is open-ended reasoning with no consequences, brainstorming, explaining a concept, drafting something you'll heavily edit. Use Now Assist (or an external agent wired in through MCP) the moment the task touches real data, real permissions, or a real state change in a system of record. The dividing line isn't "how good is the writing." It's "does this action need to be governed and remembered."

The shadow-AI tab wins on convenience and loses on accountability. Your job isn't to ban it, you'll lose. Your job is to make the governed path the path of least resistance. That's the whole game.