ServiceNow Launches AI-Powered Agent for Enterprise Use with Nvidia Partnership

A new AI-powered agent designed for enterprise use has been launched by software company ServiceNow in partnership with Nvidia. The agent, called Arc, is intended to tackle everyday work tasks such as email management and Microsoft Teams interactions.

One of the major concerns holding back companies from adopting more autonomous AI systems is ensuring adequate security controls, particularly after a series of high-profile breaches attributed to agent malfunctions over the summer. ServiceNow and Nvidia are addressing this issue by combining AI capabilities with robust security measures.

Arc is governed and secured through the ServiceNow Super Intelligence Control Tower, which serves as a centralized framework for monitoring and managing AI agents within an organization. This system also enforces Nvidia's OpenShell tool, allowing Arc to operate independently while maintaining control over enterprise security.

The agent has been designed to run efficiently on Nvidia's RTX Spark devices and can be deployed locally on an employee's desktop or within Nvidia's OpenShell on a separate device. This flexibility makes it possible for Arc to build memory and skills over time, leading to improved performance in handling daily tasks.

The new AI agent from ServiceNow and Nvidia is designed to be efficient in handling complex tasks while minimizing data transfer. This approach allows Arc to keep data local whenever possible, only handing off more complex tasks to frontier models when necessary.

Arc's efficiency is also due to its ability to proactively surface work and action items by scanning an employee's inbox and meeting notes. This feature enables employees to stay on top of their tasks without having to manually sift through large amounts of information.

One key aspect that sets Arc apart from other AI agents is its multi-level security framework. According to ServiceNow and Nvidia, this approach provides more reliable security by moving enforcement beyond the software layer into the runtime and hardware layers.

This enhanced security framework includes ServiceNow's Super Intelligence Control Tower, which enforces company policies while the agent is working on tasks. Additionally, Nvidia OpenShell restricts the agent's behavior where necessary, giving IT departments granular control over system access and data usage.

The companies also introduced AI Passport, a layer of agent security that holds an enterprise's set of rules about risk tolerance and service access. This allows Nvidia's features like OpenShell to intervene efficiently if an AI agent takes unauthorized action.

The integration of ServiceNow's AI Passport and Nvidia's security features aims to create a robust system for monitoring and controlling AI agent actions within an enterprise environment. This is achieved through continuous monitoring of an agent's permissions and actions, preventing potential breaches or unauthorized activities.

If an AI agent's passport is revoked due to suspicious behavior, OpenShell will isolate the task being executed by containing it in a sandbox for further review. Importantly, this interruption is task-specific, allowing other approved agent actions to proceed without disruption. This ensures that only the compromised action is impacted, minimizing downtime and maintaining overall system efficiency.

The result of this collaboration is an enhanced security mechanism that combines software-based controls with hardware-enforced enforcement, providing a powerful deterrent against potential breaches. By integrating Nvidia's BlueField technology with ServiceNow's Control Tower, the companies have created a robust defense against unauthorized AI activity.