News
Research Report
The State of the Digital Workplace
ServiceNow made a pitch to enterprises today at its Knowledge 2026 conference in Las Vegas: that whatever AI agent your employees end up using — Claude, Microsoft Copilot, a homegrown bot or one of ServiceNow's own — the work those agents do should run through ServiceNow's platform.
The announcements come at a moment when most enterprises are wrestling with a basic question about generative AI: how do you let agents do things without losing control? ServiceNow's answer involves opening up its platform to outside agents, giving employees a single front door to it, expanding governance tooling and shipping a fleet of role-specific AI workers.
It's also a story about acquisitions. Over the past 18 months, ServiceNow has spent billions assembling the pieces of this strategy: Moveworks ($2.85 billion, closed December 2025), Armis ($7.75 billion, completed April 2026), identity security firm Veza, observability company Traceloop, plus Pyramid Analytics and data.world. Most of what was announced today is built on top of those deals.
Table of Contents
- Action Fabric: An Open Door for Outside Agents
- Otto: A Front Door for Employees
- AI Control Tower Adds Discovery, Cost Tracking and a Kill Switch
- Autonomous Workforce: AI That Finishes the Job
- What This Means for Practitioners
Action Fabric: An Open Door for Outside Agents
The headline announcement was ServiceNow Action Fabric, a new release which exposes the company's workflows, approval chains and business rules to any AI agent through the company's Model Context Protocol (MCP) Server. MCP is the open standard, originally introduced by Anthropic, that's quickly become the connective tissue between AI assistants and enterprise systems.
"Others let agents read and write data. We let agents execute governed work. Our flows, our playbooks, approvals, catalogs, the full system of action," said Nenshad Bardoliwalla, group vice president, product management, AI products, during a press briefing.
Anthropic is the first design partner. ServiceNow shared this example: a product manager who has just changed roles internally is working in Claude Cowork, hitting walls because they don't have access to the right repos, Slack channels and tools. Claude queries ServiceNow, surfaces the missing items and routes the requests through the appropriate approval chains in the background.
"The gap between knowing what needs to happen and making it happen is where productivity dies," said Boris Cherny, Head of Claude Code at Anthropic, in ServiceNow's announcement.
Worth noting: ServiceNow is hedging across model providers. In January 2026, the company announced a three-year agreement with OpenAI that made OpenAI the only natively embedded model on the platform. Action Fabric is the other side of that strategy — a way to stay open to any agent, regardless of who built it.
The MCP Server is available now, included in every Now Assist and AI Native SKU, with additional features expected in the second half of 2026.
Otto: A Front Door for Employees
ServiceNow launched EmployeeWorks, a conversational interface based on its Moveworks buy, in February. Today it introduced Otto, which pulls EmployeeWorks together with the company's existing Now Assist and AI Experience layers into a single AI experience for the enterprise.
The premise is a familiar one for most knowledge workers. Employees don't know which system handles what. They open the IT portal for a laptop request, go to a different portal for HR, search SharePoint for documents, file a ticket somewhere else for facilities and toggle between three or four apps to track the status of any of it.
Otto is meant to sit on top of that mess. Employees can ask it anything in natural language — by chat, voice or search — and it routes the request to the right agent or workflow in the background.
It supports four interaction modes: conversational AI for natural-language requests, enterprise search across documents and SharePoint, AI voice agents in multiple languages, and an AI Data Explorer for plain-language analytics queries.
Otto starts inside EmployeeWorks and AI Control Tower, with rollout across the rest of the product portfolio over the next year.
AI Control Tower Adds Discovery, Cost Tracking and a Kill Switch
ServiceNow introduced AI Control Tower at Knowledge 2025 as a way to keep tabs on AI deployments. Today's update tries to turn it into a single dashboard for every AI system running in an enterprise, not just the ones built on ServiceNow.
Control Tower will now act as the oversight layer for all activity in an AI environment. It breaks that oversight down across five functions: discovering what AI systems and agents exist, observing what they’re doing, governing how they’re allowed to behave, securing access and identities, and measuring performance and cost.
To do that, it’s plugging into the major systems where work already happens: Amazon Web Services, Google Cloud, Microsoft Azure, plus enterprise platforms like SAP, Oracle and Workday. The idea is to pull in not just human users, but also service accounts, bots, IoT devices and operational tech — basically anything that can take action inside a system.
The ServiceNow acquisition of Traceloop adds a deeper layer of visibility. It goes beyond just logging what an AI agent did to show how the agent made decisions in real time: what inputs it considered, what steps it took and where things might have gone wrong.
On the governance side, ServiceNow is packaging prebuilt risk frameworks aligned with standards like National Institute of Standards and Technology and the EU AI Act, so companies don’t have to design policies from scratch.
Finally, through an integration with Veza, the platform focuses on enforcement. If an AI agent tries to act outside its assigned permissions, the system can flag it and stop it immediately. That’s what ServiceNow is branding as a “kill switch” — a way to cut off an agent mid-action if it crosses a line.
Underpinning all of this is ServiceNow's configuration management database, or CMDB, which aims to function as a live system of record that updates in real-time with every ticket, workflow run and policy change.
"We've spent 22 years architecting incremental value on top of this layer. And now we've furnished all of this intelligence and historical context for decision making as part of our context engine," said John Aisien SVP Security, Risks and OT Products, ServiceNow, during a press briefing.
Webinar
May
6
Retention Is a Communication Problem: New Data from 6,000+ Employees
This session gives leaders a clear diagnosis and practical path to treat retention as an infrastructure problem.
Register
Conference
Jun
2
Gartner Application Innovation & Business Solutions Summit Las Vegas 2026
Register
Webinar
On demand
Do Learning Programs Really Work? How to Turn Education Into Engagement In Healthcare
See how leaders are using learning programs to build trust with healthcare professionals and create measurable engagement.
Watch Now
Webinar
On demand
Fuel Your Front Line: Workforce Trends Shaping 2026
Get expert insight into the tools and tactics top frontline teams will need in 2026.
Watch Now
Webinar
On demand
Fix the Disconnect Between Projects and Profit
Your margins are hiding in disconnected data. Let's find them.
Watch Now
Webinar
On demand
AI in Customer Service: Faster Resolutions, Happier Customers
Don’t let rising demand burn out your team. See how to build a smarter, more resilient support org.
Watch Now
The company claims that structured operational context is what lets agents resolve issues against current reality rather than stale documentation — and what makes governance enforceable rather than theoretical.
There's also a new cost-tracking layer aimed at one of the more under-discussed pains of the agentic era: runaway model spend. Anyone who has watched a token bill creep up will recognize the problem.
The expanded Control Tower enters Innovation Lab in May with general availability expected in August 2026.
Autonomous Workforce: AI That Finishes the Job
Finally, ServiceNow also announced a major expansion of its Autonomous Workforce — a catalog of "AI specialists" purpose-built for IT, customer relationship management, employee services and security and risk.
The pitch is familiar one in the agentic AI space. ServiceNow argues that most enterprise AI to date has been advisory: they summarize, draft and recommend, but humans still do the work. Its specialists, the company states, complete end-to-end processes.
Terence Chesire, Group VP ServiceNow CRM explained this differentiator as this: "AI without workflows is just expensive advice."
Some of what's being announced:
- IT: New specialists for AIOps, site reliability engineering, asset lifecycle and portfolio planning.
- CRM: Specialists across sales qualification, quoting, order fulfillment, invoice disputes and renewals.
- Employee Services: Digital workers for HR, finance, legal, procurement and workplace services.
- Security and Risk: Available for preview in June 2026 and generally available in September, these specialists are aimed at vulnerability triage, SOC incident investigation and third-party vendor risk.
ServiceNow also announced that its AI specialists will be available in the Microsoft Agent 365 Marketplace later this year, where they'll appear as "digital employees," able to draft Word documents, respond to Outlook messages or act on PowerPoint comments, subject to Microsoft's permissions and admin controls.
What This Means for Practitioners
ServiceNow is betting that two things will be true over the next few years. First, that enterprises will run a sprawling mix of AI agents from many different vendors and will need a single layer to govern all of them. Second, that the value of AI agents will come from their ability to safely execute work against existing systems of record.
That's a reasonable read of where the market is heading. Whether ServiceNow becomes the default operating system for it is another question. Microsoft, Salesforce, AWS and Google are all making variations of the same pitch, and most large enterprises will likely end up with overlapping governance layers.
For digital workplace and employee experience leaders, we return to the original question: "What work are we comfortable letting an agent finish on its own, and how do we know it did it correctly?"
The addition of the kill switch is telling. The fact that vendors are now selling one suggests the industry is, at minimum, taking seriously the possibility that they'll be needed.
Have a tip to share with our editorial team? Drop us a line: