Skip to content
HomeNewsNew NVIDIA RTX Spark laptops offer OpenClaw and Hermes Agent at first boot: what owners get on day one

Subscribe to our newsletter

One short email when we publish. No spam, unsubscribe anytime.

Hermes opens a new winged golden chest on Olympus while the OpenClaw crab leans out over a brand new laptop, under the headline New RTX Spark laptops at first boot

New NVIDIA RTX Spark laptops offer OpenClaw and Hermes Agent at first boot: what owners get on day one

By OneClickClaw EditorialOctober 17, 2026

Buyers of the new NVIDIA RTX Spark laptops can install OpenClaw or Hermes Agent from the first screen Windows shows them, without typing a single command. NVIDIA said in a press release on the 16th of October 2026 that the laptops went on sale that day from ASUS, Dell, HP, Lenovo, Microsoft and MSI, starting at $2,599 (€2,310).

The euro figures in this article use the European Central Bank reference rate of the 16th of October 2026, when one euro bought 1.1253 US dollars.

An AI agent is a program that does tasks for you, such as answering messages, sorting files or writing text, after you give it instructions in plain language. OpenClaw and Hermes Agent are open source agents, which means anyone can read and reuse their code for free.

Both agents sit in the Windows setup

According to the same press release, guided setup and simple installers let owners start with six agents without a command line: GitHub Copilot, Hermes, OpenClaw, Perplexity, Pi and WorkBuddy. A command line is the black text window where programmers type instructions.

NVIDIA wrote on its blog on the 13th of October that the laptop greets its owner with a "Getting Started" screen at first boot, where the owner picks an agent app and installs it. The blog said the apps then suggest a model and settings sized for the machine and finish the setup in a few clicks.

The two projects confirmed their part on the 16th of October. Nous Research, the company that makes Hermes Agent, wrote on X that Hermes Agent is a launch partner for RTX Spark, that owners can pick it in the built-in Windows agent setup, and that a new guided onboarding gets them working on the laptop or in the cloud.

The OpenClaw project wrote on X the same day that owners can set up OpenClaw straight out of the box, choose a model on the laptop or connect a cloud provider, and follow a guided setup without opening a terminal. The project thanked the NVIDIA and Microsoft teams that worked with it.

Dillon Rolnick, chief executive of Nous Research, said in the NVIDIA press release: "Personal AI becomes more useful when it can work with your tools and help get things done."

What the owner does on day one

The OpenClaw crab sits beside a new laptop and steps onto a path of glowing stepping stones

NVIDIA's own step by step guide for RTX Spark lists five steps.

  1. Choose and install an agent.
  2. Choose a model, the "brain" that writes the agent's answers.
  3. Connect apps, such as email or a calendar.
  4. Talk to the agent.
  5. Optionally, teach it more so it helps in more ways.

The guide, updated on the 8th of October, describes OpenClaw as an agent for life and work on the owner's own terms, and Hermes Agent as a personal agent that learns and improves with every interaction.

Local model or cloud provider, in plain words

Hermes holds a small oil lamp in one hand and points to a distant temple on the mountain with the other

Every agent needs a model to think with. The setup offers two kinds.

  • A local model runs on the laptop's own chip. The guide says local models suit everyday tasks, keep the owner's data on the device and cost nothing per use.
  • A cloud provider runs the model on a company's computers over the internet. The guide says cloud models suit heavier work such as coding and need a paid subscription.

Memory decides how large a local model the laptop can hold. NVIDIA says RTX Spark machines come with up to 128 gigabytes of memory shared by the whole system. Its guide suggests a model of around 30 billion parameters for agent work, which typically needs a machine with 48 gigabytes or more. A parameter is one of the many small settings a model learned during training, so a larger count means a larger model.

NVIDIA also released its own model for agents on these laptops, Nemotron 3.5 Bolt, on the 15th of October. The company's blog said it is packed to fit in 24 gigabytes of memory.

Security comes built in

NVIDIA said in its press release that its OpenShell runtime and Microsoft Execution Containers let owners control what an agent is allowed to touch. Both work like a fenced yard: the agent does its job inside, and the owner decides which doors open.

The NVIDIA blog of the 13th of October said OpenShell adds credential protection, which keeps passwords and keys out of the agent's reach, plus security monitoring. It said OpenClaw supports Microsoft's containers, and that Hermes Agent adds the OpenShell controls and the containers by the end of October.

When the laptop closes

An agent installed on a laptop answers while the laptop is switched on and awake. People who want their agent replying to messages around the clock can also run OpenClaw or Hermes Agent on a server that stays on day and night. OneClickClaw hosts both, with the Starter plan at €14.99 a month and a 7-day free trial.

Helpful documentation

Want your own AI agent running without the setup? We host it on a dedicated EU server. Free for 7 days, no credit card.

Start your free trial

How managed OpenClaw hosting works

Get notified when we publish new articles

No spam, unsubscribe anytime.

By subscribing, you agree to our Privacy Policy.