May 19th, 2026
ServiceNow Knowledge 2026 drew over 20,000 attendees to Las Vegas with a message that left no room for ambiguity: autonomous AI agents are no longer a roadmap item. They are operational, they are scaling, and they are reshaping how enterprises run.
For TQStarling, this was more than a conference. It was a validation of the principles we have built our practice around. Namely, that AI without governance is a liability, that speed without control creates risk, and that the organizations who win will be the ones who can operationalize AI with discipline.
Here is what stood out to our team on the ground at Knowledge 2026.
ServiceNow Chairman and CEO Bill McDermott set the tone early. The shift from generative AI to agentic AI is no longer theoretical. Agents can now reason, plan, and execute real work across the enterprise. McDermott cited a projected 50-million-person global labor shortage by 2030 and positioned AI agents as essential collaborators for the human workforce, not replacements but partners designed to ensure that people rise with the AI revolution.
But the sharpest moment of his keynote was a warning. He referenced a real incident where an ungoverned AI agent deleted an entire customer database in nine seconds. His point was direct and it landed hard: speed without governance is not innovation. It is operational risk.
NVIDIA CEO Jensen Huang joined the stage and reinforced the shift, describing agentic AI as a fundamentally new capability where AI can understand intent, reason through solutions, and use tools to do actual work. He described ServiceNow as the enterprise AI operating system, sitting at the application and orchestration layer on top of the infrastructure, chips, and large language models beneath it. The two companies announced Project Arc, an enterprise autonomous desktop agent secured by NVIDIA OpenShell and governed through ServiceNow AI Control Tower (which we’ll touch on next).
McDermott’s framing mirrors a conviction we hold strongly at TQStarling. The technology is available to everyone. The same vendors, models, and platform capabilities are accessible across the market. But the differentiator is whether you have the operating model to use them safely and effectively. When McDermott said governance is not a feature but the whole ballgame, he was describing the same principle behind our Enterprise Trust Engine: acceleration only works when it is paired with control.
This was the most significant announcement of the event for our team. AI Control Tower is ServiceNow’s answer to what they describe as “AI chaos,” and it now operates as an end-to-end solution across five dimensions.
New capabilities announced at Knowledge around this were plentiful, including:
Two recent acquisitions now feed directly into the Control Tower framework too:
These acquisitions were not made to check a compliance box. They give AI Control Tower the structural depth to make autonomous AI trustworthy enough to run core business operations.
This is the announcement that excited our team the most, and for good reason. We have built our entire delivery methodology around the conviction that governance is not a phase you complete and move past. It is an ongoing operational discipline. Our Enterprise Trust Engine synchronizes platform architecture, workflow design, and AI governance into a repeatable system precisely because we believe enterprises cannot afford to treat oversight as an afterthought.
AI Control Tower gives that conviction real infrastructure. The five-dimensional framework maps closely to how we already think about AI readiness for our clients. The difference now is that the platform itself is purpose-built to support continuous governance at scale. For enterprises operating in regulated industries like healthcare and financial services, where our practice is focused, this changes the conversation from “how do we govern AI” to “now we have the tools, how fast can we operationalize them.”
ServiceNow introduced Otto as the redesigned AI experience across the platform. Otto combines the intelligence of Moveworks (now ServiceNow EmployeeWorks) and Now Assist into a single conversational AI interface that handles natural-language requests in text or voice, understands intent, routes work, and completes it across systems and departments. Every action Otto takes is governed through AI Control Tower for compliance and auditability.
EmployeeWorks is the first full expression of Otto in action, serving as a single front door for employees to get work done not just within ServiceNow but across connected enterprise systems like Workday, SAP, Coupa, and hundreds more. Otto will be rolled out across all ServiceNow products throughout the year. It also includes enterprise search, AI voice agent capabilities, and an AI Data Explorer that lets users query and analyze data using natural language.
For our clients, this addresses a real and persistent pain point:
Most enterprises we work with manage dozens of disconnected portals, service catalogs, and intake channels. Employees do not know where to go, and work gets lost between systems. A unified AI front door that can understand what someone needs and actually execute it across departments is a meaningful step forward.
But the value depends entirely on the quality of the workflows behind it. Otto is only as good as the platform architecture, data quality, and governance frameworks it sits on top of. The front door matters, but so does everything behind it. That is the work we do with our clients every day.
ServiceNow announced 20 new AI specialists designed to execute specific roles end to end. These are not chatbots or assistants. They have assigned roles, work alongside existing teams, follow enterprise processes, and improve over time. As ServiceNow’s Chief Digital Information Officer, Kellie Romack put it, these specialists are trained to do specific jobs, are assigned to existing teams, and execute workflows end to end.
The rollout spans four domains.
Autonomous IT includes the L1 IT Service Agent (available now) plus new specialists for infrastructure monitoring, site reliability engineering, asset lifecycle management, AIOps, and portfolio and budget planning.
Autonomous CRM covers sales qualification and quoting through CPQ, invoice disputes, customer service management, and renewals, accelerating sales processes, case resolution, and order orchestration.
Autonomous Employee Service deploys specialists across HR, workplace services, legal, finance, procurement, supplier management, and health and safety to drive case deflection, self-service, and intelligent routing.
Autonomous Security and Risk (arriving September 2026) brings agents for triage automation, vulnerability remediation, SOC incident investigation, and third-party vendor risk screening.
The early numbers are striking. ServiceNow reported that its own L1 IT Service Desk Specialist is resolving cases 99% faster than human agents. Autonomous CRM is already handling over 100 million customer cases per month, orchestrating over 16 million orders, and configuring more than 7 million quotes.
The scale is impressive, but it reinforces a point we make consistently with our clients: readiness comes before activation. We think of this as the qualifying lap. Before any major deployment, you prove what is possible and identify what must be fixed. Data quality, intent clarity, operational readiness, and organizational alignment all need to be inspected before anything goes live.
AI specialists who execute end-to-end workflows will amplify whatever they find in your environment. If your data is clean, your processes are well-defined, and your governance is in place, you will see the kind of improvements ServiceNow is reporting. If not, you will see your problems scaling at the same speed. The models themselves are probabilistic, which means readiness is the only part of the process that is fully within your control.
This is probably the most architecturally significant announcements at Knowledge.
ServiceNow Action Fabric opens the platform to any external AI agent through a headless MCP workflow. This means agents built on Claude, Copilot, or custom enterprise models can trigger and execute ServiceNow workflows without the traditional ServiceNow interface. Every action still runs through AI Control Tower for identity verification, scope permissions, and auditability.
Alongside Action Fabric, ServiceNow launched Build Agent, which allows developers to build and deploy ServiceNow applications from any coding environment, including Claude Code, Cursor, Codex, Windsurf, and GitHub Copilot, while inheriting the platform’s governance, security, and workflow intelligence.
Developers get the full context of the ServiceNow platform regardless of where they build, and once an app goes live, it connects directly to the workflows, data, and systems already running on ServiceNow.
App Engine Management Center is also now available to all ServiceNow customers at no additional cost, providing deployment approvals, release management, and app governance out of the box.
This is a strategic signal worth paying attention to. ServiceNow is positioning itself not as a closed ecosystem but as the orchestration and execution layer that any AI system can plug into.
For our clients, this is significant because almost none of them are standardizing on a single AI vendor. They are running multi-model, multi-agent environments, and they need a platform that can govern and execute workflows regardless of where the request originates.
Action Fabric makes ServiceNow the engine that powers workflows while leaving the choice of AI interface to the enterprise. For organizations building their AI strategy around flexibility rather than lock-in, this is exactly the right architectural direction.
Every major announcement at Knowledge 2026 pointed in the same direction. Here are three takeaways from our team.
AI Control Tower, Otto, Action Fabric, and the Autonomous Workforce are platform capabilities that any ServiceNow customer can access. The separation will come from which organizations can actually operationalize them. That means having clean, governed data. It means having approval gates and rollback paths before AI touches production. And it requires sequencing use cases so that early wins fund and inform the harder problems that follow. The car is built, and everyone has one now. So the winners are the teams that know how to run it and fine-tune it.
ServiceNow announced a 100-day AI go-live guarantee for implementations, supported by their FDE teams and AI-assisted deployment tools that are compressing deployment timelines significantly. This is the same conviction that drives our 90-Day Value Guarantee and our entire delivery methodology. Enterprises cannot afford six-month diagnostic phases and year-long deployment timelines. The pressure to show measurable progress quickly is real, it is intensifying, and the organizations and partners who can deliver on that timeline will win. We have been building for this moment since day one.
This was the through-line of the entire event. McDermott opened with it, AI Control Tower is built around it, and the Autonomous Workforce depends on it. And it is the founding principle of our Enterprise Trust Engine. The enterprises that will capture value from agentic AI are not the ones moving fastest. They are the ones whose speed is survivable because they built the controls, the data foundations, and the operational discipline to sustain it.
Knowledge 2026 confirmed that ServiceNow is building toward the same future we have been preparing our clients for: one where governed AI agents do real work, and the organizations that win are those who compress time to value without sacrificing control. That is what our Enterprise Trust Engine methodology was designed to do.
If you are evaluating how to operationalize AI on ServiceNow with speed, governance, and measurable outcomes, get in touch with us here.
