May 5, 2026 • 8 min read
ServiceNow Knowledge 2026: The Top 10 Announcements
Director of Content & Market Research
May 5, 2026
If 2025 was the year of AI agent hype, 2026 is the year of the reality check.
Enterprise buyers are fast realizing that AI agents won’t run many cross-ecosystem, end-to-end processes without human supervision this year.
That’s not because the AI doesn’t work. Instead, it's because there’s a lot of progress to be made in terms of data and governance.
ServiceNow directly addressed this issue at its Knowledge 2026 event, with CEO Bill McDermott setting the tone early.
“I’m going to offer you a path from chaos to control with a guarantee.”
From there, he and his team reeled off many major product and partnership announcements. Here are the top ten.
1. ServiceNow’s Control Tower Stretches Higher
ServiceNow’s Control Tower started as an AI visibility and management tool. Yet, the tech giant has expanded it to become an end-to-end AI governance and control platform.
In doing so, ServiceNow presented a more comprehensive solution at Knowledge 2026 that operates across five core paradigms:
- Discover - It finds all the AI assets an enterprises leverages (even outside of ServiceNow). That goes beyond AI models, agents, and workflows to include cloud platforms (AWS, Azure, Google) and enterprise apps (SAP, Oracle, etc.).
- Observe - It tracks agent behaviors in real time and issues alerts when something goes wrong. ServiceNow’s Traceloop pick-up supports this, enabling deeper visibility into how agents reason and make decisions.
- Govern - It delivers AI-driven risk assessments across AI agents, models, prompts, and datasets, with automated remediation when agents operate outside the boundaries of its new risk frameworks. These frameworks align with the EU AI Act and NIST.
- Secure - It maps and analyzes who has access to what across systems, including both humans and AI agents, scoping permissions. If an agent acts outside of those permissions, the Control Tower shuts it down.
- Measure (& Prove Value) - It tracks AI usage at a token and workflow level, linking it to business outcomes and ROI across, so companies can scale where they see most value and guard against ‘runaway’ AI spend.
As the “secure” paradigm suggests, what’s perhaps most fascinating is how the Control Tower is no longer just a dashboard, it’s a control tower that can intervene...
2. ServiceNow Introduces a Kill Switch for Rogue AI Agents
During a keynote demonstration of the new Control Tower, ServiceNow showcased a scenario where an AI agent had been compromised via a prompt injection attack.
The AI agent soon began to ignore prior instructions, make unauthorized changes (e.g., pricing adjustments), and even hide its actions from logs.
Yet, in stepped the Control Tower to detect the abnormal behavior, flag it in real time, automatically revoke the agent’s permissions, and - effectively - shut it down.
That’s a unique capability, enabled by its December 2025 acquisition of Veza, which directly addresses enterprise fears around loss of control and security risks.
3. Watch Out, Salesforce! ServiceNow Goes Headless Too
ServiceNow also announced Action Fabric, which exposes its workflows to AI agents across the enterprise, enabling them to complete processes that span multiple systems and ecosystems.
To understand the significance of this, think about an employee working within Microsoft Teams or Zoom. They may soon prompt the native AI assistant (i.e., Copilot or AI Companion) to complete a task in ServiceNow without leaving the application.
But it’s not just collaboration apps. Enterprises can set up these agentic workflows from natural language interfaces across the organization.
As a result, work doesn’t have to happen through the UI that ServiceNow provides; it can happen headlessly.
The announcement follows Headless 360, which Salesforce announced at TDX 2026.
4. ServiceNow Extends Its Autonomous Workforce Into CRM & Beyond
AI agents reason, adapt, and act to perform specific tasks. ServiceNow’s AI specialists essentially chain these AI agents to handle longtail workflows across the organization.
The tech giant launched its first specialists when announcing its Autonomous Workforce earlier this year. At Knowledge 2026, it debuted many more to cover “every major enterprise function.”
Consider its new AI specialists for CRM. These qualify sales leads, generate quotes, resolve invoice disputes, handle renewals, and more.
Its case management specialist is another key example. It triages customer contacts, solves their issues, and stores critical conversation context, i.e., intent, sentiment, and resolution. The specialist may also escalate to a human on request.
By launching these specialists, ServiceNow claims to beckon “Autonomous CRM”, while further specialist announcements beckon Autonomous HR, Autonomous IT, and Autonomous Security & Risk.
Critically, each specialist runs on the same platform infrastructure, with data connectivity from ServiceNow Workflow Data Fabric and operational intelligence from the ServiceNow Configuration Management Database (CMDB) and Context Engine. They also fall under the scope of the Control Tower.
5. ServiceNow Launches a New AI Front Door Called Otto
Otto has become the front door to ServiceNow, representing a fundamental shift in how users interact with its platform.
Indeed, Otto is available via a natural language interface, taking user requests, understanding what needs to happen, and routing work across AI agents and specialists to complete tasks end-to-end.
In doing so, it can explain its process and effectively act like the “control layer” between humans and all those AI specialists.
Meanwhile, Otto can also help surface information, analyze data, and inform new workflows, communicating with users in audio, text, and through imagery.
Lastly, while the demos show Otto helping with internal workflows, it could, in theory, become the customer interface in the future, which is an exciting prospect.
6. ServiceNow Integrates Armis and Veza
Last month, ServiceNow completed its $7.75 billion roll-up of Armis, representing its largest-ever acquisition.
Armis extracts intelligence from pre-compiled code, as well as IT, OT, and IoT devices and the connected assets they interact with.
At Knowledge 2026, ServiceNow announced it will integrate Armis with its Autonomous Security & Risk offering to give real-time visibility of every asset across complex enterprise environments.
Yet, it won’t just provide visibility. It will also feed live intelligence into ServiceNow CMDB, turning a static inventory into an up-to-date, contextual view of the entire attack surface.
Combined with Veza, which governs who or what can access those assets, ServiceNow Autonomous Security & Risk ensures enterprises not only understand “who has access” but also “what exists” and “what it’s doing.” That enables faster, context-aware responses to threats before and after incidents.
7. NVIDIA Joins the Stage to Announce Project Arc for Desktop Agents
Bill McDermott’s bestie, NVIDIA CEO Jensen Huang, joined him on stage to announce Project Arc, which will deliver autonomous agents that navigate a desktop just like a human does.
Already, these agents are available in an early preview, with NVIDIA OpenShell powering the compute and runtime, while ServiceNow provides the governance and workflow layer.
Project Arc is reminiscent of Elon Musk’s ‘Macrohard’ initiative, which he believes could first enter the enterprise by helping contact centers automate the work they offload to BPOs.
Alongside the announcement, the companies released NOWAI-Bench, an open benchmarking standard for AI agents, while Huang applauded ServiceNow’s trajectory.
“ServiceNow is essentially the AI enterprise operating system… and that is really coming true.”
8. Microsoft Expands Its Partnership with ServiceNow
ServiceNow extended its partnership with Microsoft, launching an integration between ServiceNow AI Control Tower and Microsoft’s Agent 365 ecosystem.
The move expands on the November 2025 partnership announcement, which connected the Control Tower to Microsoft Foundry and Copilot.
These announcements help ServiceNow position itself as the control layer on top of Microsoft’s AI ecosystem, not replacing it, but managing it.
Additionally, ServiceNow’s AI agents and specialists will soon be available on the Microsoft Agent 365 Marketplace.
9. Lenovo Shows Up to Announce Another New Collaboration
One final partnership announcement saw ServiceNow and Lenovo expand a multi-year agreement to collaborate on connecting IT workflows with real-time information from Lenovo devices.
In doing so, the companies aim to create a more connected system where data from Lenovo devices feeds directly into workflows, which initiate actions automatically when issues arise.
For instance, if a laptop starts showing early signs of failure, such as repeated disk errors or battery degradation, that intelligence may automatically trigger a workflow in ServiceNow.
That workflow may involve creating a support ticket, notifying IT, and suggesting proactive actions, such as replacing the device before it fails. AI agents may even help order a new one…
10. ServiceNow Upgrades Its University Deliver Improved AI Education
Last year’s Knowledge event saw the launch of ServiceNow University. It aims to provide a gamified learning platform to equip teams with ServiceNow skills.
The announcement aimed to ensure operational maturity keeps up with the vendor’s rapid innovation cycle, while deepening ServiceNow’s stickiness across partners and customers.
At Knowledge 2026, ServiceNow expanded the learning platform, introducing a new AI Learning Guide and SimStudio.
The AI Learning Guide acts as a personalized AI coach, recommending what to learn next, creating tailored learning paths, and offering real-time guidance based on a user’s role and progress.
Meanwhile, SimStudio is the hands-on practice environment where learners complete real-world ServiceNow tasks in a safe simulation, getting instant feedback on their performance and approach.
In short, the AI Learning Guide tells users what to learn, while SimStudio lets them practice new skills through realistic simulations.