ServiceNow cut hundreds of employees in June and said something most companies are still too scared to say out loud: the AI is working.
"Our platform is generating real AI efficiencies inside our own business," the company stated while announcing the layoffs. They were also clear they'd keep hiring — for AI-focused skills specifically. Not a hiring freeze. A hiring redirect.
CEO Bill McDermott had already spelled out the logic back in April, when ServiceNow reported earnings:
"As you have attrition in the company, you don't have to backfill it. So we can still have a great culture. We can still have enormous, enormous, high-performance standards, and at the same time, we can capture massive efficiencies to expand the free cash flow margin of the corporation."
— Bill McDermott, CEO, ServiceNow, CNBC, April 22, 2026
That's the confession. And it changes the conversation.
The "Efficiency" Euphemism Just Got Retired
For the past two years, every company with a chatbot pilot and a board asking about AI ROI learned to say the same thing: "We're using AI to augment our people, not replace them." It was true enough, for long enough, that most executives believed it.
ServiceNow just retired that talking point. They're not augmenting. They're replacing — in specific, targeted roles — and they're saying so.
This is a company that builds the workflow automation platforms that most enterprises run their IT operations on. They sold you the tool that was supposed to make your people more efficient. Now they're using that tool to cut the people who manage it.
If that sentence makes you uncomfortable, it should.
Because the logical next question is: what does that mean for your team?
Meanwhile, Codex Is Learning to Code While You Sleep
The same week ServiceNow dropped its layoff announcement, OpenAI closed its acquisition of Ona — formerly Gitpod — a German startup that runs AI agents inside isolated cloud sandboxes. The purpose is specific: it lets OpenAI's Codex agent keep working on a task for hours, or even days, after the engineer closes their laptop.
Codex now has more than 5 million weekly active users. That's up from 3 million in April. The growth rate alone is a signal.
But the Ona acquisition isn't a Codex feature upgrade. It's an architecture decision. OpenAI is building a system where the AI doesn't wait for you to be at the keyboard. It has the environment, the execution context, and now the persistent infrastructure to operate autonomously. The engineer defines the task. The agent executes it. The engineer reviews the output.
That's not augmentation. That's a different division of labor.
What You're Actually Hiring For Now
Here's what these two stories, read together, tell you about the talent market:
The roles that survive AI automation are not the ones adjacent to it — they're the ones above it.
ServiceNow hasn't announced cuts to its technical AI roles. The reported reductions hit solution consulting, sales, product marketing, and learning and development — the roles AI now handles. The people who set policy, define requirements, audit outputs, and make judgment calls under ambiguity — those are the roles growing.
OpenAI isn't replacing senior engineers with Codex. It's giving senior engineers a force multiplier that makes junior execution-layer work less necessary.
The talent question for every CTO right now isn't "how do I protect my headcount from AI?" It's "what does the person who manages, governs, and directs AI systems actually look like — and do I have any of them?"
Most companies don't. Not yet.
The Hiring Redirect Is Real
ServiceNow said they're "managing headcount with discipline to end the year where we started" — while simultaneously investing in AI-focused skills. That's a net-zero headcount story with a massive composition shift inside it.
This isn't new for them, either. CFO Gina Mastantuono put a number on it last summer:
"We talked about at Knowledge $100 million in savings in headcount alone in 2025. And we're seeing that come to fruition as planned."
— Gina Mastantuono, President & CFO, ServiceNow, The Register, July 25, 2025
This is what the next 18 months of enterprise hiring looks like across every major technology company: flat or slightly shrinking overall headcount, with substantial churn underneath. The people who built and maintained the old workflows are out. The people who can design, govern, evaluate, and course-correct AI agents are in. The ratio of those two categories is shifting.
If you're planning next year's hiring budget against your current role mix, you're planning against a world that's already gone.
What To Do About It
Three things I'd push any engineering leader to do before Q3 planning:
1. Audit your current roles against what AI can execute today. Not in theory — in practice, with the tools available now. Codex, Devin, Claude, ServiceNow's own AI workflows. Which tasks in each role are now automatable? What's left?
2. Define what "governing AI" means on your team. Who reviews agent outputs? Who sets the constraints? Who is accountable when an agent ships a bad call at 2 a.m. with nobody watching?
If the answer is "nobody" or "we haven't figured that out yet," you have a gap before you have a hiring plan.
3. Don't let "we're still hiring" become a cover story. ServiceNow is still hiring. So is everyone else laying people off. The composition of who you're hiring for matters more than the number. Hiring 10 AI governance roles while cutting 30 workflow administrators is not a stable equilibrium — but it's the right direction.
The companies that get this transition right won't be the ones who moved fastest to cut. They'll be the ones who knew exactly what they were building toward when they did.
ServiceNow said the quiet part loud. The question now is whether you're ready to hear it.