Host OpenClaw on an always-on machine
OpenClaw routes AI conversations across channels and gives you a browser UI to control them. Like anything that answers a webhook or holds a connection open, it is only useful while it is running.
It is pre-downloaded on every machine, so starting it does not begin with a download.
How it arrives
- On the machine
- Installed as part of the build
- At boot
- Starts when you start it, and holds nothing until then
- Reached at
- /openclaw/ on your machine's own HTTPS name
- Memory while running
- about 1 GB
We did not write OpenClaw. It is openclaw/openclaw. We install it on every machine and put it behind the login.
Something that listens has to stay listening
The machine is a real VM that keeps running when you close the tab. A process you start does not stop because your laptop slept, your train went into a tunnel, or you switched to your phone.
A gateway that drops off when a session ends is a gateway that misses messages. This one is a service on a machine that is up for the length of your term.
One login, no second exposure
The Control UI answers on a path under the machine's own HTTPS name, behind the same authentication as everything else. OpenClaw's own login is disabled because the gateway in front of it is the boundary.
Questions
- Is OpenClaw already installed?
- Yes. It is downloaded onto every machine as part of the build. Starting it is a switch on the dashboard rather than an install.
- Can it receive webhooks while I am away?
- Yes, as long as it is started. The machine and its services keep running independently of your browser session.
Once you have one
It comes with the machine
Every tool on this site is installed on every machine. You are choosing a size, not a bundle.