Skip to content

What is SpineOS?

What SpineOS does, how a cloud desktop for an AI agent works, and why an agent needs a real screen, terminal, and file system.

2 min read

Cloud desktops for AI agents #

SpineOS gives every AI agent its own persistent Linux computer in the cloud. Your agent sees the screen, clicks, types, browses the web, manages files, and runs terminal commands, the same way a human would.

Unlike ephemeral sandboxes that disappear after each task, SpineOS desktops persist. Your agent's work, files, installed software, and conversation history survive across sessions.

How it works #

  1. Create a desktop -- pick a tier (Free, Pro, or Max) and we provision a full desktop environment on dedicated infrastructure.
  2. Pick your agent and connect your AI provider -- choose an agent type (OpenClaw, Claude Code, Hermes Agent, Codex, or Grok) and connect the AI account or API key it runs on.
  3. Start working -- use the VNC viewer to watch your agent in real-time, the terminal for shell access, and the chat panel to give instructions.

Every desktop includes:

  • VNC viewer -- live remote desktop in your browser
  • Terminal -- shell access in the browser
  • Chat panel -- talk to your AI agent, review responses
  • File manager -- upload, download, browse files
  • Mission Control -- create tasks, set approval gates, monitor activity
  • Metrics -- CPU, RAM, and disk usage in real-time

Who is it for? #

Developers building autonomous AI agents that need a real operating system. Teams running AI workflows that need persistent state across sessions. Researchers testing agent capabilities in realistic environments. Or anyone who has outgrown ephemeral sandboxes.

What makes it different #

FeatureSpineOSSandboxes (E2B, etc.)
Full GUI desktopYes (streamed to the browser)No
Persistent storageYesNo (ephemeral)
Pre-installed AI agentYes (OpenClaw, Claude Code, Hermes Agent, Codex, or Grok)No
Terminal + VNC + ChatAll threeUsually one
Keep-alive (24/7)YesSession-limited
Mission ControlYesNo