ServiceNow's AI Control Tower is renewed, becoming a true command center
AI is evolving faster than businesses can understand, manage, or govern it. The number of AI models, applications, and agents is increasing, and the focus is shifting towards making AI agents more autonomous and independent while setting boundaries and leaving decisions to humans.
ServiceNow aims to bring order to this chaos, and at Knowledge 26, where Edge9 is also present, it revealed all the new developments it is working on. There are indeed many, and the company's focus is quite clear: to enhance the AI Control Tower, launched last year, to make it increasingly connected with corporate AI data and systems, whatever they may be. At the same time, it aims to offer new agents specialized in specific business functions.
The new AI Control Tower from ServiceNow
The new AI Control Tower has been enhanced to work beyond the ServiceNow perimeter, providing integrations for AWS, Google Cloud, Microsoft Azure, and applications like SAP, Oracle, and Workday. Many businesses have begun deploying AI functions in various departments, often within vertical applications, risking loss of visibility on what is used, by whom, with what data, and with what permissions. ServiceNow's control tower aims to bring order to this chaos.
The new version of AI Control Tower introduces over 30 new enterprise integrations, real-time observability of agent behavior, automated risk and compliance controls, identity governance extended to hyperscaler environments, and dashboards to monitor AI spending. This last point is crucial since AI, which was supposed to reduce costs by increasing efficiency, is starting to become extremely expensive: without control, the risk is to waste enormous sums on tokens.
But spending is just one of the problems that companies adopting agent-based AI face: it is not enough to install an agent on their systems. To use it safely, one must understand what it is doing, where it is making decisions, whether it complies with policies, and what economic impact it is generating. This is precisely the main task of the new version of AI Control Tower.
AI specialists: from support to actual process execution
Alongside the novelties of the AI control center, at Knowledge 26, ServiceNow brought updates regarding the Autonomous Workforce, which transforms AI agents from tools that respond or assist into systems that take over structured activities. ServiceNow introduces new agents specialized in CRM, user experience, IT operations, and cyber security. The ambition is to shift AI from the interaction level to the execution level: not just raising a ticket or suggesting an action but managing a flow from intake to resolution, with defined roles, corporate permissions, and built-in governance.
ServiceNow claims that internally, the Autonomous Workforce already manages over 90% of employee IT requests, and the Level 1 Service Desk AI Specialist resolves assigned cases 99% faster than human management.
Action Fabric: third-party AIs under control
In short, AI is proposed not as a separate layer but as an integrated operational component within workflows for sales, support, orders, disputes, renewals, and internal support. This logic is also linked to Action Fabric, through which ServiceNow opens its action system to agents built outside the platform, including those based on Claude, Copilot, or clients’ proprietary developments.
The most important aspect is support for the Model Context Protocol Server, now generally available, which allows external agents to execute enterprise actions in a governed manner. With this approach, ServiceNow aims not only to push its own AI to customers but positions itself as an orchestrator of third-party AIs as well. This is an important aspect considering that no company today limits itself to a single AI but adopts solutions from various vendors.
Otto: the unique interface to search, ask, and let AI work
Another prominent novelty announced at Knowledge 2026 is ServiceNow Otto, the new enterprise AI experience. Otto combines conversational AI, business search, and autonomous workflows into a single interface. The goal is to reduce the fragmentation that currently characterizes AI use in companies: chatbots in one application, assistants in another, document searches elsewhere, and workflows still managed in traditional systems. Otto tries to bring everything into the same access point, allowing people to turn an intention into an operational result across different systems, desktops, and processes.
The difference compared to a generic assistant is in its connection to the underlying platform. Otto is presented not just as a conversational interface but as the front end of a machine that has access to data, operational context, policies, workflows, and AI specialists. Here is where ServiceNow tries to differentiate itself from many enterprise copilots: it is not enough to generate a correct response; it must link it to a verifiable action within the organization. In this perspective, Otto becomes the point where the user asks, the platform interprets the context, agents execute, and AI Control Tower maintains control over security, governance, and outcome measurement.