Your own AI assistant, running 24/7 on hardware you own.
ClawBox ships pre-configured and ready to talk to — message it from Telegram or the web. Two editions of the same hardware are available: ClawBox (Connect) ships running OpenClaw, and ClawBox Hermes Edition ships running Hermes Agent by Nous Research. You choose which one your box ships with when you order.

📺 Hands-on walkthrough (English voice-over) — “ClawBox: A Ferramenta que Muda Como Você Trabalha,” including a manual ClawBox install via installer.sh → watch on YouTube.📺 Watch an independent review — maker Kevin McAleer breaks down “What is ClawBox, and who is it for?” → watch on YouTube.

Quickstart

Unbox, power on, and talk to your ClawBox in minutes.

First Boot

Walk through the guided on-device setup wizard.

Choose Your AI

ClawBox AI by default — or bring Claude, GPT, Gemini, or local models.

Hardware

Specs for ClawBox Connect and ClawBox Workstation.

​
What is ClawBox?

A small, always-on computer running a self-hosted agent that connects your chat apps to an AI. It’s for people who want a private, dedicated AI assistant without renting cloud infrastructure or babysitting a server install.

​
Two editions, same hardware

You choose which agent your box ships with when you order, and you can switch between them on the device with a single click afterward. Both editions are the same NVIDIA Jetson Orin Nano Super hardware at the same price. These are separate products at checkout, so your order is identifiable as one or the other through fulfilment, invoicing and support. The edition decides which agent arrives running, not which agent the box can run. Most of this documentation applies to both editions; pages that are specific to one say so. What makes it different?
  • Owned hardware — it lives on your desk, drawing as little as 7–15 W.
  • Multi-channel foundation — Telegram and web are the managed ClawBox paths; advanced users can add more OpenClaw channels where supported.
  • Flexible AI — ClawBox AI out of the box, or connect Claude / GPT / Gemini / OpenRouter / local Ollama.
  • Pre-configured — no Linux setup, no Docker wrangling. Plug in and go.

​
Two models

ClawBox Connect

NVIDIA Jetson Orin Nano · 67 TOPS · always-on assistant. €549

ClawBox Workstation

NVIDIA DGX Spark · 1 PFLOP · run 200B-param models locally. €5,499

​
Going deeper

Technical Reference

Architecture, networking, auth, updates, and the agent interface — for power users and AI agents.

Troubleshooting & Recovery

Symptom-first fixes with exact commands, plus ordered recovery options.

​
Need a hand?

Talk to support

Email us or join the community Discord — we’ll get you unstuck.