ServiceNow, Nvidia launch Arc enterprise AI agent

Arc was first announced in May. The companies describe it as an agent designed for daily tasks that rely on productivity tools such as email and Microsoft Teams. It is built to run efficiently on Nvidia's new RTX Spark devices and can run locally on an employee's desktop or within Nvidia OpenShell on its own device. ServiceNow and Nvidia said Arc builds memory and skills over time so its performance continually improves.

The companies said Arc is efficient, handing more complex tasks to frontier models only when necessary and keeping data as local as possible otherwise. Arc also scans an employee's inbox and meeting notes to proactively surface work and action items, according to ServiceNow and Nvidia.

The companies are placing particular emphasis on the security framework around Arc. They said their approach offers more reliable security by moving enforcement beyond the software layer into the runtime and hardware layers. While Arc works on tasks, ServiceNow's Super Intelligence Control Tower enforces company policies, and Nvidia OpenShell restricts the agent's behavior where necessary. IT defines policies in the Control Tower, and Nvidia OpenShell can be directed to enforce them at runtime. Users get granular control over which systems AI agents can access, which APIs they can call and which data they can read.

AI Passport, announced by the two companies last week, adds another layer. It holds an enterprise's rules, set by the Control Tower, about risk tolerance and the services Arc and other agents are allowed to access. The companies said AI Passport functions as a bridge between ServiceNow infrastructure and Nvidia's checkpoints, and the system is designed to continuously monitor an agent's actions and permissions to prevent breaches and unsanctioned actions.

If the Control Tower tells OpenShell that an agent's passport has been revoked, OpenShell keeps the task the agent was trying to execute contained within a sandbox where reviewers can audit it later. The companies said the interruption is task-specific, so other approved agent actions can continue without restarting. Nvidia BlueField can enforce the revocation at machine speed before the AI agent can proceed, the companies said, creating enforcement in silicon designed to prevent overrides from compromised software.