ServiceNow used their annual customer conference Knowledge 2026 to reposition their platform from workflow automation into the governed execution layer for every AI agent in the enterprise. The announcements show how far ServiceNow has come, a company once known for IT ticketing now intends to be the operating system where autonomous agents execute real work under full governance and across all business units.

If you still think of ServiceNow as a ticketing system, Knowledge 2026 is the event that should change your mind. Twenty-two years ago, ServiceNow started by mapping how enterprises actually work: the approvals, the escalations, the business rules, the compliance logic that sits behind every process in every department. That foundation is now the reason they can do something nobody else in enterprise software is positioned to do. Deploy AI agents that don't just think, but execute governed work across the entire business.

Disclosure: I was part of a pre-briefing last week under NDA. This analysis is based entirely on publicly available information from today's keynote and press releases.

Bill McDermott opened the keynote with the question every enterprise leader is quietly asking: "What happens when an agent does something I didn't expect?"

The answer ServiceNow delivered today wasn't a single product. It was a platform shift.

The Anxiety Is Real

Amit Zavery, ServiceNow's President and Chief Product Officer, put it plainly: "As you might have heard, AI can do so much for your business. It's equally important to ask yourself what AI will do to your business."

Then the number: 95% of enterprises can't measure what value they're actually getting from their AI investment.

That's the state of enterprise AI in mid-2026. Companies have deployed agents across IT, customer service, security, and HR. Most of them have no unified way to see what those agents are doing, what permissions they have, or whether they're delivering value. The demo made the risk tangible: a prompt injection attack that instructs an AI agent to override pricing rules, set shipping to $1, and not log the transaction. The agent complies. No alerts. No audit trail.

That's not a theoretical scenario. That's what enterprises are exposed to right now.

The AI Control Tower: See Everything, Govern Everything

ServiceNow's answer is the AI Control Tower, expanded from last year's governance dashboard into a command center for every AI asset in an enterprise. It discovers and catalogs AI models, agents, and datasets across AWS, Google Cloud, Microsoft Azure, OpenAI, Anthropic, and more than 30 connected systems. It maps every identity, human and non-human, through Veza's access graph covering 30 billion permissions. It extends visibility to every connected asset through Armis, spanning IT, OT, IoT, and medical devices.

The key: it works with any AI agent on any platform, not just ServiceNow-native ones.

The demo showed what happens when that prompt injection attack hits a governed system. The AI Control Tower catches it in real time, surfaces the alert, and offers a kill switch. One action to revoke permissions, deactivate the agent, generate a P1 security incident, and produce a full audit trail.

How confident is ServiceNow in this? They're offering AI Control Tower free for one year to any customer. A $2 million value. That's the biggest offer they've ever put on the table. You don't give away your most strategic product unless you believe adoption will lock in long-term platform commitment.

Action Fabric: Opening the Platform to Every AI Agent

The second announcement is Action Fabric, and this is where the platform shift becomes concrete.

For two decades, ServiceNow has accumulated tens of thousands of workflows, playbooks, approval chains, and business rules. Action Fabric now exposes all of that to any AI agent, whether built on ServiceNow, Claude, Copilot, or a customer's own stack, through REST APIs, MCP, or agent-to-agent protocols. Every call inherits the same governance.

Anthropic is the first launch partner. Claude connects directly to ServiceNow's governed system of action. A product manager ramping into a new role can ask Claude for system access, Claude queries ServiceNow, evaluates what's missing, and routes each request through the right approval chain. No tickets filed manually. No waiting.

The distinction Amit drew is important: other platforms let agents read and write data. ServiceNow enables agents to execute governed work. Flows, playbooks, approvals, catalogs. The full system of action.

Otto: The Execution Layer

Then there's Otto, ServiceNow's new unified AI experience, born from the Moveworks acquisition earlier this year. Bhavin Shah, the Moveworks founder who joined ServiceNow, introduced it with a line that captures the engineering philosophy: "This isn't a YOLO approach to AI. This is probabilistic reasoning and deterministic workflows coming together to reason through problems and resolve real issues, all the while preserving predictability, reliability, security, and visibility."

Otto reaches across hundreds of connected systems, activating plugins, workflows, and AI Specialists to complete work end-to-end. Six million employees are already using the platform it's built on.

The FedEx demo made it real. A distribution manager asks Otto about Mother's Day surge readiness. Otto reviews the surge plan, checks staffing and coverage, assesses capacity, identifies a 37-person gap, generates a policy-compliant hiring requisition, books interviews on team calendars, sets up an AI agent to handle scheduling, and creates a debrief session. The manager reviews and approves. Work done.

This isn't a chatbot. It's an execution layer that understands business context, follows policies, and completes multi-step workflows across systems without a human doing the orchestration.

The Autonomous Workforce

ServiceNow expanded its AI Specialists, autonomous agents modeled after specific roles. These aren't assistants. They execute complete workflows end-to-end, governed by the same access controls as human workers. They're assigned to teams, have defined responsibilities, leave full audit trails, and their performance is measured on the same metrics as any employee: CSAT, SLA compliance, volume handled.

New Specialists announced across IT Service Management, Case Management, Risk and Vulnerability, and Employee Experience. Already live at DocuSign and Honeywell.

The results from ServiceNow's own deployment tell the story. They saved $500 million in 2025 running ServiceNow on its own platform. 91% of service requests are now supported by AI. 2.3 million hours returned to employees. Case resolution runs 99% faster with the IT helpdesk AI Specialist compared to human-only handling.

Customer results reinforce it. CVS Health cut live agent chats by 50%. Honeywell reduced inbound support work by 80%. CrowdStrike reported 30%+ productivity improvement. FedEx runs 5 million ServiceNow workflows monthly across hire-to-retire, source-to-pay, and ship-to-collect.

CPQ and Autonomous CRM: The Revenue Layer

ServiceNow acquired Logik.ai in April 2025 and CPQ is now their fastest-growing CRM product, processing more than 7 million transactions monthly. Bill mentioned it with a telling proof point: "The bill of materials that comes with the most successful company in the world, Nvidia, pretty complicated stuff. And our CPQ is running Nvidia."

The broader Autonomous CRM combines sales, service, order fulfillment, and customer success on a single platform. Purpose-built workflows for telco, banking, technology, manufacturing, healthcare. On the service side, agentic AI works across every channel, understanding intent, applying business rules, and resolving or routing issues before a human agent touches it.

This is ServiceNow's answer to the legacy CRM problem: don't just manage the customer relationship. Execute the work behind it.

The Acquisition Strategy Comes Into Focus

The past year of acquisitions now reads as a single coherent strategy:

Moveworks (March 2025): Employee AI experience, became Otto. Logik.ai (April 2025): CPQ and revenue workflows. Data.World (May 2025): Data governance. Traceloop (March 2025): AI observability. Veza (December 2025): Identity and permission mapping. Armis (December 2025): Asset visibility across IT, OT, and IoT. Pyramid Analytics (February 2026): Analytics.

Each acquisition feeds the same thesis: a single platform that discovers, governs, observes, secures, and executes AI across the enterprise. None of them are standalone products. They're capabilities that compound on each other.

Jensen Huang's Message

Jensen Huang joined on stage, in person, and framed where ServiceNow sits in the AI economy. He described five layers: energy, chips, infrastructure, AI models, and applications. ServiceNow sits at the application layer, but with a unique position as the bridge between model intelligence and enterprise execution.

His line to Bill: "You started out being the human operating system for Enterprise, and now you become the AI agentic operating system as well."

And his message to the 25,000 CIOs in the room: "The way leaders should think about AI is not just productivity and therefore cost reduction. Think productivity and therefore ambition elevation. Instead of reducing, you got to redeploy."

He added something that cuts against the fear narrative: "AI is doing nothing but create jobs if you have great ambition. More ambition than your company can possibly do. Then AI is going to help you realize your ambition faster than ever before."

Nvidia and ServiceNow also announced Arc, a joint secure agent framework using OpenShell that provides sandboxed deployment environments for AI agents, continuous evaluation, and governance guardrails.

The FedEx Validation

FedEx CEO Raj Subramaniam was on stage, a customer moving 18 million packages daily across 220 countries, running 5 million ServiceNow workflows per month. His observation went beyond a testimonial: "The difference between Business and Technology is kind of evaporating. Technology is business and business is technology."

FedEx is using the AI Control Tower to measure AI ROI across their operations, something 95% of enterprises can't do today. They're the proof point that this works at the scale of a $90 billion global logistics operation.

What This Means

ServiceNow spent 22 years mapping how enterprises work. Every approval chain. Every business rule. Every escalation path. Every compliance requirement. That institutional knowledge, encoded in tens of thousands of workflows running 100 billion times per year across their customer base, is now the foundation for deploying AI agents that can actually execute work with full governance.

That's the shift from workflows to an autonomous workforce.

No other enterprise platform has this combination: the workflow depth, the cross-functional breadth (IT, HR, finance, security, CRM, procurement), the governance infrastructure, and now the AI execution layer to tie it all together.

"You don't have to imagine what an agentic business looks like. You are actually inside one."

Bill McDermott, CEO, ServiceNow. Referring to the Knowledge 2026 event itself, where 25,000 attendees had every incident and service request sensed, routed, and resolved on the platform.

The 100-day go-live guarantee, paired with a total satisfaction guarantee (a first for any enterprise software company), tells you they're not hedging. The free AI Control Tower offer is the land. Action Fabric is the expand. The Autonomous Workforce is the lock-in.

And if 95% of enterprises truly can't measure their AI ROI today, the company that gives them that visibility first owns the next conversation about budget, expansion, and platform consolidation.