AI Gives You the Map. ServiceNow Claims It Drives the Car.
Bill McDermott walked onto the floor of the New York Stock Exchange this week to ring the opening bell while his company's stock sat down 25% year-to-date. That takes a particular kind of composure. And the conversation that followed on CNBC showed exactly where that composure comes from: a clear-eyed view of what his company does that AI does not.
His CNBC interview with the Squawk on the Street team was worth watching carefully. McDermott was composed and media-ready, but what stood out was the business clarity underneath it. This was a leader who had clearly thought through every dimension of the challenge his company faces and was not reaching for reassuring sound bites. He made an argument that cuts through the current AI hysteria in a way most enterprise software leaders have failed to do. And for those of us advising organisations across APAC on how to actually deploy AI, the argument he made is exactly the one your board should be hearing.
The Market Is Asking the Wrong Question
The "AI disrupts software" trade that has hammered ServiceNow, Workday, Salesforce, and others this year is built on a fairly simple thesis: if AI can do what enterprise software does, why pay for enterprise software?
McDermott's rebuttal is direct. "AI is great advice. Workflow is where the action layer takes over. And we are the workflow company."
That line deserves to sit with you for a moment.
The fear trade assumes that because AI can generate an answer, it can also execute the outcome. It cannot. Not yet. And frankly, in most large enterprise environments, not for a long time. Not because the models are not smart enough. Because the systems those models would need to touch are extraordinarily complex, decades old, and deeply interconnected.
McDermott cited 80 billion workflows in flight globally, touching seven trillion transactions, built across six decades of legacy infrastructure. That is not a number you automate away with a prompt.
He used two simple examples to make the point. A payroll error on aisle five. A broken VPN in a hotel room. AI can tell you exactly how to fix both. But it cannot actually fix either. Not without the orchestration layer that connects intent to execution across disparate enterprise systems.
That orchestration layer is the product that ServiceNow is selling. And right now, it is the product most enterprises do not have.
Why This Matters in APAC Specifically
I spend a lot of time working with organisations across markets like Singapore, India, Japan, and Australia. The workflow integration problem McDermott described is not just a Western enterprise challenge. It is arguably more acute in APAC for two reasons.
First, enterprise technology stacks in Asia are often more heterogeneous. You have a mix of regional ERP vendors, hyper-local compliance requirements, and decades of custom-built systems that were never designed to talk to each other. The integration debt is significant.
Second, the pace of AI adoption in the region is accelerating faster than governance frameworks can keep up. BCG's AI Radar 2025 report showed adoption rates climbing steeply across financial services, manufacturing, and government sectors in APAC. But execution infrastructure, the workflow layer McDermott keeps pointing to, is lagging.
That gap is not just a technology problem. It is a governance risk. When AI advice cannot be connected to auditable, controlled execution, you get shadow AI. Decisions are made by tools that no one in the organisation can fully explain or govern. That is not a competitive advantage. That is a liability waiting to surface.
The Pricing Model Question Is the Right One
One of the CNBC anchors pushed McDermott on seat-based pricing, and this is where I think the interview got genuinely interesting.
The concern is real. If AI agents replace human users, the seat-based SaaS model starts to look fragile. McDermott acknowledged it directly, which is unusual for a CEO in media mode.
His answer has three parts.
Active users are up 25% year-over-year. Not because companies are hiring more people. ServiceNow is covering more enterprise territory per account, expanding horizontally across functions that previously ran on separate systems.
Hybrid revenue is already shifting the mix. The company is preparing for a world where a significant portion of its user base is non-human. McDermott cited a projection that 3 billion digital agents will be added to enterprise environments by 2030. ServiceNow claims it is the only platform that can natively manage both its own agents and those of every other software vendor.
The Control Tower is the new category. This is the most strategically important claim he made. As enterprises deploy AI agents across functions, someone needs to govern them. Who authorised this agent? What data can it access? What can it do without human approval? How do you audit what it did?
That is not a capability most AI vendors are building. They are building capability, not governance. McDermott is betting that the governance layer is where the durable enterprise value will sit. I think he is right.
The Jobs Conversation Nobody Wanted to Have
Credit to the CNBC team for pushing on employment. McDermott did not dodge it.
He said youth unemployment could move from 9% today into the mid-30s within a few years. That is a striking figure to say on live television. His reasoning: agents are replacing non-differentiating roles faster than most policymakers or HR leaders anticipate.
He pointed first to ServiceNow’s own internal operations as proof: the platform took out 90% of the use cases their own customer service team used to handle manually. He then cited customer deployments where agents are now handling configuring, pricing, and quoting functions that humans used to do.
I want to be direct here because I think the comfortable response, which is to say "AI creates as many jobs as it destroys," is inadequate to the reality unfolding. The transition will be real; the displacement will be concentrated in specific roles and age groups; and the APAC region will feel this acutely, given the proportion of younger workers in service and administrative functions across markets such as India, the Philippines, and Vietnam.
The question of responsible leadership is not whether to deploy agents. It is how to plan for workforce transitions while doing so. That is a governance conversation, not just a technology one.
What McDermott Got Right About the CEO's Real Problem
One of the most honest moments in the interview was when McDermott described a conversation with a CEO the day before. The CEO told him he genuinely did not understand what was happening in his own enterprise.
That is the problem statement that does not appear in vendor decks. The gap is not between AI capability and enterprise need. The gap lies between AI capability and executives' understanding of how to connect the two.
McDermott invoked Lou Gerstner's IBM turnaround and the line that business process reinvention is like putting your hair on fire and trying to put it out with a hammer. The point is that rethinking how work actually gets done in a large organisation is brutally hard. AI does not make that easier. It adds urgency to a problem that was already difficult.
ServiceNow's pitch is that it provides the platform on which AI gets a "clear shot on goal," to use McDermott's term. A single orchestration layer that does not require ripping out legacy infrastructure but instead sits on top of it, connecting everything.
Whether that is achievable at the scale they are describing is a fair question. But the problem they are naming is the right one.
A Practical Playbook for Senior Leaders
If you are a CXO or board member watching the enterprise AI market right now, here is how I would frame what McDermott said in terms you can act on.
1. Audit your execution layer before your AI layer. Before asking what AI can do for your organisation, map where decisions currently stall in your workflows. The bottlenecks are not usually a lack of intelligence. They are integration failures, approval loops, and system incompatibilities. That is your real problem.
2. Separate the advice layer from the action layer in your AI strategy. Most organisations are investing heavily in AI tools that generate recommendations. Very few are investing in the infrastructure that allows those recommendations to be acted upon at scale. The ratio should be closer to equal.
3. Build agent governance before agent deployment accelerates. If you are already running AI agents in any business function, you need a governance framework for non-human identities, access controls, auditability, and escalation paths. The security and compliance risk of unmanaged agents is not theoretical. It is arriving now.
4. Plan workforce transitions explicitly, not implicitly. If your AI strategy does not include a workforce transition plan, it is incomplete. This is not just an ethical point. It is a risk management point. The social and regulatory pressure on companies that automate without proper planning is mounting, particularly in markets like Singapore, Australia, and Japan, where government attention to workforce impacts is increasing.
5. Evaluate your platform strategy, not just your tool strategy. The market is full of AI tools that solve point problems. The durable competitive advantage will belong to organisations that build or adopt a coherent platform layer that connects tools, data, agents, and governance. That is what McDermott is selling. Whether it is ServiceNow or another vendor, the strategic logic is sound.
The Takeaway
For Builders and Practitioners: The workflow layer is the moat. If your organisation is deploying AI without a clear answer to the question "and then how does it actually execute?", you have a gap. Closing that gap is not a technology project. It is an enterprise architecture decision that belongs on the CEO’s agenda.
For Decision-Makers and Boards: The software sell-off is pricing in a risk that is partially real. Enterprise software that cannot adapt to agent-based consumption models will struggle. But the companies that own the governance and orchestration layer, the ones that sit between AI capability and enterprise execution, have a stronger position than the market currently reflects. Watch that category carefully.
McDermott ended the interview with “I’m here to win.” But the line that actually matters for every enterprise leader came earlier.
“So we are the do it company.”
That is the strategic question for every enterprise in 2026. Not whether to adopt AI. But who owns the layer that turns AI's advice into actual outcomes? In your organisation, do you know the answer to that question?
You can watch the full interview on the ServiceNow LinkedIn post.
Adopt AI with Confidence and Clarity. Are you struggling to build a business case for AI or unsure about governance and compliance? AIdeate Solutions guides organisations through practical, responsible AI adoption. We help you move beyond the hype to implement workflows that create real value.
