ServiceNow and Nvidia launch Arc, a personal, long-running enterprise AI agent for productivity tasks across email and Teams, for RTX Spark and Nvidia OpenShell
Radhika Rajkumar /ZDNET:
Context & Ripple Effects
ServiceNow entered 2026 with a three-year agreement to integrate OpenAI models and agents into its business software. Arc adds a ServiceNow-Nvidia route for enterprise agents, tying a user-facing productivity tool to Nvidia’s RTX Spark and OpenShell environments.
Nvidia had framed agent deployment around trust, security and safety through its 2025 Inference Microservices, then introduced OpenShell’s agent security and monitoring reference architecture. The pairing makes governance part of the product stack around an agent that operates across everyday work surfaces.
First-order effects
- ServiceNow customers using RTX Spark and Nvidia OpenShell gain Arc as a persistent productivity agent spanning email and Teams, putting the two companies’ software and infrastructure in the same deployment path.
- Nvidia extends OpenShell from a reference architecture into a platform attached to a named enterprise productivity workload, while ServiceNow gains another model-and-infrastructure partner alongside its OpenAI integration agreement.
Second-order effects
- Enterprise buyers evaluating Arc must treat productivity-agent selection and agent governance as connected procurement decisions, because the deployment is designed around Nvidia’s security and monitoring layer.
- Microsoft Teams and email become higher-value integration points for enterprise-agent vendors, shifting competition toward durable agents that can operate across work surfaces rather than single-task assistants.
Third-order effects
- If vendors continue bundling agents with the infrastructure and controls that govern them, enterprise AI stacks will consolidate around integrated workflow, compute and policy layers rather than standalone agent tools.
- The pattern points to governance moving below the application layer: policy enforcement becomes a differentiator for vendors seeking to run long-lived agents in business systems.
The trend: Enterprise AI is moving toward long-running workflow agents sold alongside the infrastructure and governance controls required to deploy them.