ServiceNow Launches Otto and Action Fabric at Knowledge 2026, Betting the Enterprise on Governed AI Agents
ServiceNow unveiled Otto, Action Fabric, and an expanded AI Control Tower at Knowledge 2026, repositioning itself as a unified governance layer for every AI agent across the enterprise.
Overview
ServiceNow used its annual Knowledge 2026 conference in Las Vegas on May 5, 2026 to make the most sweeping platform announcements in the company’s recent history, introducing Otto — a unified AI experience merging Now Assist, Moveworks, and AI Experience — alongside Action Fabric, an expanded AI Control Tower, and a desktop agent project co-developed with NVIDIA. The cluster of announcements amounts to a strategic bet that enterprises will consolidate their AI governance on a single platform rather than manage each tool independently.
What We Know
Otto: A Unified Front Door Across the Enterprise
The centerpiece of Knowledge 2026 was ServiceNow Otto. According to Yahoo Finance, the company describes the product as: “Rather than living inside a single application, ServiceNow Otto sits across the entire enterprise, understanding intent, routing work to the right agent, and executing it to completion.”
Otto draws together three previously separate user experiences — the Now Assist conversational interface, the Moveworks-derived EmployeeWorks portal, and the broader AI Experience layer — into one natural-language entry point that can resolve requests across departments without requiring the employee to know which underlying system owns the task.
Bhavin Shah, SVP and GM of Employee Experience and AI at ServiceNow, stated: “Employees no longer need to know where to go or who to ask. They just ask ServiceNow Otto, and it can handle the rest within the guardrails the enterprise requires,” according to Yahoo Finance.
ServiceNow acquired Moveworks for $2.85 billion in 2025, according to SiliconANGLE. Shah had described EmployeeWorks in February as “one of the first AI front doors that doesn’t just summarize, it completes the work.”
Otto is initially available in ServiceNow EmployeeWorks and the AI Control Tower, with a broader product-wide rollout planned for the year ahead, as reported by SiliconANGLE.
Action Fabric: Opening the Workflow Engine to Any Agent
Alongside Otto, ServiceNow announced Action Fabric, which exposes the company’s governed workflow engine — flows, playbooks, approvals, and service catalog entries — to any external AI agent through a generally available Model Context Protocol (MCP) Server. The practical effect is that an agent built on a third-party platform can execute a ServiceNow workflow without requiring the user to switch applications, as described by SiliconANGLE.
Every action routed through Action Fabric passes through AI Control Tower for identity verification, permission scoping, and audit logging.
AI Control Tower: From Visibility Tool to Governance Platform
ServiceNow expanded AI Control Tower, which it describes as a platform for discovering, observing, governing, securing, and measuring AI across the enterprise. Nenshad Bardoliwalla, group VP of product management, described the problem the platform addresses: “Customers are telling us AI is everywhere, but it isn’t connected, isn’t governed and isn’t finishing the work,” according to SiliconANGLE.
Bardoliwalla also described the company’s growing Autonomous Workforce initiative as “an entirely new class of AI specialists that think, act and work as part of a team.”
On the security side, John Aisien, senior VP of central product management, stated that “AI-powered adversaries are moving faster than teams can human respond to or even detect,” framing the Control Tower’s expanded security capabilities as a direct counter to that threat.
AI-Native Product Portfolio
In April, ahead of the conference, ServiceNow had already signaled the scale of the pivot. Amit Zavery, President and Chief Product Officer, declared: “ServiceNow brings it all together, so customers start with a complete AI-native experience across all products and packages, not a procurement project,” as reported by SiliconANGLE. The company also announced developer tooling for building custom agent skills, with support for third-party environments including Claude Code, OpenAI Codex, and Cursor.
Project Arc: NVIDIA Partnership Brings Agents to the Desktop
NVIDIA CEO Jensen Huang appeared alongside ServiceNow CEO Bill McDermott at the Knowledge 2026 opening keynote to announce Project Arc, described as “a long-running, self-evolving autonomous desktop agent designed for knowledge workers, including developers, IT teams and administrators,” per the NVIDIA Blog. The system runs on NVIDIA OpenShell, described as “an open-source secure runtime for developing and deploying autonomous agents in sandboxed, policy-governed environments.”
Jon Sigler, Executive Vice President and General Manager of AI Platform at ServiceNow, said: “Project Arc represents the next step in our ongoing collaboration with NVIDIA, bringing autonomous execution to the desktop.”
Huang framed the broader market opportunity at the conference, noting that “the service industry is 100x larger than the software industry” and that “compute needed for agentic AI has increased 1,000% compared to generative AI” due to the additional reasoning and tool-use requirements agentic systems place on inference, according to Fortune.
Customer Adoption and Enterprise Governance
McDermott highlighted the fragmentation problem motivating these announcements, noting that enterprises run “367 different applications” with AI “bolted on like a sidecar” to each. He also described a cautionary incident in which “an AI agent gained elevated permissions and deleted an entire production database” in nine seconds, making the case for the AI Control Tower’s intervention capabilities, according to Fortune.
FedEx, one of the announced customers, executes 5 million ServiceNow workflows monthly. Vishal Talwar, FedEx CDIO, articulated the governance approach: “We treat [AI agents] as a digital workforce that needs to be governed with the same rigor and policies as our human teams.”
Earlier deployments of Now Assist at Rolls-Royce Holdings yielded 5,000 hours of IT operations efficiency savings and 300,000 saved shop floor hours, according to SiliconANGLE.
What We Don’t Know
ServiceNow has not disclosed the pricing tiers for Action Fabric access beyond the stated inclusion in Now Assist and AI Native SKUs. The timeline for Otto’s full product-wide rollout beyond EmployeeWorks and AI Control Tower remains described in general terms, with a broader rollout planned across all products over the coming year. The extent to which the Autonomous Workforce specialists can operate without human oversight on high-stakes tasks — approvals, procurement, financial transactions — has not been detailed in public documentation.
Analysis
ServiceNow’s announcements at Knowledge 2026 reflect a strategy that has been building for several years: move from being a workflow platform to being the governance and orchestration layer beneath every AI agent in the enterprise, regardless of which vendor built the agent. The Moveworks acquisition, the Armis and Veza integrations, and now the MCP Server in Action Fabric all extend ServiceNow’s footprint outward into adjacent systems rather than deepening capabilities within its own product boundary.
The governance angle is the key differentiator. Most enterprise AI tools are still deployed in isolation, with organizations left to stitch together audit trails and access controls across dozens of separate systems. By making AI Control Tower the mandatory routing layer for Action Fabric, ServiceNow is proposing to be the answer to that coordination problem — a position that, if it takes hold, would make ServiceNow a required element of any enterprise AI deployment rather than one option among many.
Whether enterprises accept that positioning depends on whether the platform proves more practical than simply extending existing governance tooling from Microsoft, Workday, or SAP — all of which are named as AI Control Tower integration targets rather than rivals.