There was a noticeable shift in tone during the opening keynote at ServiceNow Knowledge 2026. Not overstatement. Not marketing. Not just excitement around AI, which was (obviously) expected. But a more grounded acknowledgment that enterprise AI has entered a more complex phase.
For the past two years, most of the industry conversation has focused on what AI can generate. The keynote instead centered on what AI can control, govern, orchestrate, and execute safely inside large enterprises operating at massive scale.
That distinction shaped nearly every announcement made on stage. From the repositioning of the AI Control Tower to new autonomous AI specialists, to expanded partnerships with companies like NVIDIA, Microsoft, Anthropic, and FedEx, the message from ServiceNow leadership was that the next phase of enterprise AI will be won by the organizations capable of operationalizing AI safely and visibly, not to mention at scale.
Moving Beyond AI Hype
The ever-quotable CEO Bill McDermott opened the keynote by framing the current moment as an inflection point, not just for software, but for the future of work itself.
He described what he called the “AI Blind Spot,” arguing that much of the market remains overly focused on large language models and generative capabilities while underestimating the operational complexity surrounding enterprise AI deployment. The keynote repeatedly returned to a core idea: intelligence alone is not enough. Enterprises need orchestration, governance, workflow execution, identity management, compliance visibility, and security layered around AI systems.
Maybe this wasn't a headline to those struggling with these very issues, but it held a different impact when hearing it en masse.
As he continued, McDermott pointed to the growing fragmentation many organizations are already experiencing, with AI capabilities scattered across hundreds of disconnected applications, copilots, and agents. In that environment, AI risks becoming another layer of enterprise complexity instead of a force multiplier for the business.
Of course, this led directly into one of the keynote’s central announcements.
(Re)Introducing the AI Control Tower
We've discussed the AI Control Tower at length on these pages, but the concept is now a little broader, as ServiceNow formally positioned itself as the “AI control tower for business reinvention.”
The concept is ambitious, but it directly reflects the underlying problem we keep repeating. Enterprises are rapidly deploying AI tools, agents, copilots, and models across departments without consistent visibility into how they interact, what data they access, or how decisions are being made.
According to the keynote, the AI Control Tower is designed to provide centralized oversight across AI deployments throughout the enterprise. The platform aims to unify visibility into models, agents, permissions, workflows, and connected systems while allowing organizations to monitor performance, governance, compliance posture, and business value in real time.
In turn, AI systems must now be treated much more like members of the workforce themselves. That means identities, access controls, auditability, and operational guardrails become critical.
ServiceNow demonstrated scenarios involving AI prompt injection attacks, rogue workflow behavior, and automated kill-switch capabilities intended to stop AI-driven incidents before they cascade across enterprise systems.
The message resonated because it reflects a growing reality inside enterprise environments. As organizations move from experimentation into scaled AI deployment, governance can no longer be an afterthought.
Meet Otto
As I've mentioned in previous Knowledge blogs, most of the best insights come from hallway conversations. Well, being the pros that we are, we kept our ears firmly to the ground when we arrived, and heard that ServiceNow might be renaming its employee AI front-end.
Well, the rumors were half right, anyway. I won't reveal the name we heard, in case there's another announcement coming this week. But today, we were introduced to Otto, which ServiceNow positioned as the new unified conversational AI experience for the enterprise.
Rather than functioning as another standalone chatbot or assistant layered on top of existing systems, Otto is designed to serve as the conversational front end that connects employees directly to workflows, enterprise search, AI agents, and operational data across the business.
Built on the combined capabilities of Now Assist, Moveworks, and ServiceNow AI Experience, Otto enables users to interact with enterprise systems in natural language while orchestrating work across departments and platforms behind the scenes.
Throughout the keynote, Otto was repeatedly framed as the “agentic front door” to the enterprise, giving users a single conversational layer capable of triggering workflows, surfacing insights, managing requests, and coordinating autonomous AI specialists without requiring employees to navigate disconnected systems manually.
FedEx and the Rise of Operational AI
One of the strongest moments of the keynote came through the appearance of executives from FedEx, including CEO Raj Subramaniam and CDIO Vishal Talwar.
It was a deep discussion of AI in very relatable terms, as the conversation focused on the realities of running one of the world’s largest logistics networks. FedEx moves roughly 18 million packages per day across more than 220 countries and territories, generating enormous volumes of operational data in the process.
The company described its AI initiatives not as isolated experiments, but as part of a broader effort to transform supply chain orchestration, predictive logistics, and operational visibility. What stood out more notably was the emphasis on trust.
The FedEx team repeatedly stressed that AI cannot operate independently of governance, workflow discipline, and strong data foundations. They described AI agents as an “additional workforce” that must be governed with the same rigor as human employees.
In a time when so many are concerned about AI replacing human effort, statements like this are reassuring, since AI is no longer considered an assistive tool. Instead, it's an operational actor within critical business processes.
(And as we learn a little later, the source of many new roles.)
Systems of Action
Another major theme throughout the keynote was ServiceNow’s continued evolution from a workflow platform into what executives repeatedly described as a “system of action.”
President and Chief Product Officer Amit Zavery introduced Action Fabric, a framework intended to connect ServiceNow workflows, governance, approvals, APIs, and orchestration layers directly to AI systems.
The idea is significant because it reframes how enterprise AI operates. Most AI tools today can generate recommendations, summarize information, or answer questions. Far fewer can reliably execute governed enterprise workflows across systems, approvals, compliance frameworks, and operational policies.
Action Fabric attempts to bridge that gap. The keynote positioned ServiceNow as the execution layer capable of turning AI outputs into enterprise actions while maintaining visibility and control across the process lifecycle.
The Expansion of Autonomous Workforce
ServiceNow also expanded its vision for Autonomous Workforce. Earlier releases seemed to focus primarily on IT service management. At Knowledge 2026, the company announced broader "AI Specialists" spanning customer service, security operations, risk management, employee experience, and CRM workflows.
Aiming to surpass typical copilots, these AI Specialists were presented as role-based operational entities capable of independently executing end-to-end workflows within defined business boundaries.
Live demonstrations showed AI Specialists triaging incidents, resolving tickets, escalating complex issues to human employees, and operating against performance metrics similar to human team members.
I should note that throughout the demos, there was a strong emphasis on auditability and explainability. I don't mean to sound redundant, but this focus reflects how organizations increasingly want AI that is accountable, observable, and operationally manageable.
NVIDIA and the Acceleration of Agentic AI
The keynote concluded with an appearance by Jensen Huang, President, CEO, and co-founder of NVIDIA, who discussed the rapid emergence of the “agentic AI era.”
Huang framed the past several years as a progression from generative AI toward systems capable of reasoning, planning, tool usage, and autonomous action. He explained how enterprises are now entering a phase where AI systems actively participate in work execution rather than simply generating information.
Importantly, Huang echoed many of the governance themes raised earlier, including the importance of sandboxing AI systems, enforcing policy frameworks, and securing agent behavior before widespread enterprise deployment.
He also offered a broader perspective on AI’s impact on the workforce, arguing that the technology should be viewed less as a mechanism for reduction and more as a catalyst for expanding organizational ambition and operational velocity.
That perspective aligned closely with the overall tone of the keynote. Despite the repeated warnings around AI chaos, governance failures, and operational risk, the broader message remained optimistic and positioned as a call for more mature AI operationalization.
A Different Kind of AI Conversation
What made this keynote feel different was the shift in emphasis. The conversation has clearly moved beyond AI "ifs" and directly toward harder questions around governance, orchestration, execution, visibility, compliance, identity management, workflow integration, and operational trust.
More notably, the keynote reflected a broader maturation happening within the ecosystem, because organizations are realizing that deploying AI models is relatively basic compared to integrating AI safely into the real operational fabric of a business, which is exactly where ServiceNow is trying to position the platform.
