Dell launches Deskside Agentic AI for on-premises testing

On the opening day of Dell Technologies World 2026, Dell introduced Deskside Agentic AI, a local sandbox designed to build, test, run and fine-tune always-on AI agents on Dell high-performance workstations. The tool is intended to run on systems such as the GB10, GB300 and Precision 9 Tower and to keep agent workloads on enterprise hardware rather than public cloud APIs.

Deskside Agentic AI is built on Nvidia NemoClaw, an open-source software stack that combines OpenClaw, the Nvidia Agent Toolkit, Nvidia OpenShell and Nemotron-3. Dell said the stack lets developers create continuous agents and gives IT teams control over operational policies and security guardrails enforced at the workstation or inside the company network.

Dell estimated the offering can cut API-related cloud spend by up to 87% over two years and said organizations can reach break-even on deployment costs in roughly three months. Jon Seigal, senior vice president of Dell’s client solutions group and online marketing, told reporters that some developers have seen large, unexpected cloud bills; he gave an example of a developer who incurred about $3,400 after using roughly one billion tokens in 24 hours.

On security, Dell emphasized that running agents locally reduces the risk of exposing sensitive company data to external environments. Seigal noted that enterprise IT leaders have expressed concern about data privacy risks and a lack of guardrails in public stacks; Deskside Agentic AI places those guardrails inside NemoClaw so policies can be applied directly on the workstation or within the corporate network.

Alongside Deskside Agentic AI, Dell and Nvidia announced expanded features for the Dell AI Data Platform. The platform now integrates Nvidia Omniverse to support digital twin and robotics workflows, allowing customers to use simulation data and curated, labeled datasets for training and validation pipelines that combine virtual and physical testing.

Dell presented the announcements as part of updates to its AI Factory and workstation portfolios at the event. Company executives said the tools address customer demand for agent autonomy while aiming to provide predictable costs and controls for regulated or sensitive workloads. The Deskside offering is aimed at organizations that prefer to keep model inference, agent orchestration and related data processing on local hardware rather than in external clouds.