ServiceNow has made Arc, a personal, long-running AI agent for everyday enterprise work, generally available. The company announced the launch on October 21 at Nvidia's GTC conference in Berlin, with Arc governed by ServiceNow's Super Intelligence Control Tower and enforced at runtime through Nvidia OpenShell.
Arc ships as part of ServiceNow EmployeeWorks and is included at no additional cost with the Foundation, Advanced and Prime SKUs. Mac support is live today, and ServiceNow expects Windows support within weeks.
The agent works across ServiceNow workflows, email, Microsoft Teams and other common productivity tools. Employees hand Arc a goal rather than step-by-step instructions; the agent plans the work, executes it, checks its own output and keeps going until the task is finished. It also surfaces action items pulled from an employee's backlog, inbox, meetings and conversations.
Key facts
- 01ServiceNow made Arc, a long-running personal AI agent for enterprise work, generally available on October 21.
- 02Arc is included at no extra cost in EmployeeWorks Foundation, Advanced and Prime SKUs.
- 03Mac support is live now; ServiceNow expects Windows support within weeks.
- 04AI Passport revocations propagate across Nvidia software and hardware layers in about one second.
- 05AI Passport enforcement in Nvidia OpenShell and BlueField DPUs is slated for a future release.
Arc can run on its own inside a secure Nvidia OpenShell runtime or directly on an employee's desktop. When an integration it needs does not exist, Arc can build one, and it runs that new integration inside the OpenShell boundary. ServiceNow says the agent builds up memory and skills over time.
ServiceNow is splitting the compute between local and cloud models. Arc uses local models on Nvidia RTX Spark hardware for everyday tasks and calls frontier models for more demanding work, a setup the company says lowers inference costs and keeps more data on the device.
“Arc is the first personal AI agent built for everyday work and governable by IT”— Jon Sigler, EVP and GM of AI Platform at ServiceNow
Governance is the other half of the launch. ServiceNow's Control Tower sets policies that OpenShell enforces at runtime, letting IT departments decide which systems agents can access, which APIs they can call and what data they can read.
ServiceNow and Nvidia are also expanding their partnership around AI Passport, an attested record ServiceNow issues for each AI agent, MCP server and other governed AI asset. A passport carries business rules such as risk tolerance and which external services an agent may use. Controls can block individual actions, such as writes to production systems, while the agent continues with the rest of its work.
Enforcement runs at two layers. Nvidia OpenShell stops an agent's execution in software when its passport is revoked, while Nvidia Sentry on BlueField DPUs provides hardware-level enforcement. ServiceNow says revocations propagate across both layers in about one second without halting other workloads on the same host.
“policy has to be enforced everywhere an agent runs”— Justin Boitano, VP of Enterprise at Nvidia
That Nvidia-side enforcement is not shipping yet. AI Passport enforcement in OpenShell and BlueField DPUs is planned for a future release, while Arc itself is available now.
ServiceNow says more than 100 billion workflows run on its platform each year, the installed base Arc and AI Passport are now aimed at.
Working on something we should cover, or seeing a story we missed? Send leads, documents, or feedback to hello@aichatdaily.com. For sensitive tips, see our secure tips page for Signal and PGP options.
Spotted an error? Email hello@aichatdaily.com with the URL and the issue, or read our full corrections policy.
