OpenClaw

This afternoon, after hearing so much hype lately and listening to an interview with the inventor this morning, I installed OpenClaw, a personal AI Agent that has taken the internet by storm. Read on for experience and impressions!

Installation and first job

I am using a throw away server so when it asked me to run as root, I did not hesitate, even though all the real unix admins out there will stop reading now and curse my name to their dying day:

After that I pointed it to the AI Model to use and it had a long list to choose from:

It then guided me through setting up a channel to communicate with it. I choose WhatsApp as I have it and use it, and because the programmer mentioned that was the first way he implimented to interact with it and how he uses it. Was very easy and even generate a QR code in a terminal session to link my account:

And then gave me an option to install any skills:

And that was it!

I fired up an SSH tunnel so I can connect from my local computer’s browser to the remote throw away server (it gave me the command) and then finalized the config, giving my new personal AI agent its personality:

I named the little doggy “StationDomo” as it will serve as the JupiterStation major domo, helping with what ever is needed.

The first thing I asked of it via WhatsApp, StationDomo had no problem performing even though the prompt had some errors that were introduced via the voice to text transcribing:

And I had it sent me a sample report:

Thanks for reading and feel free to give feedback or comments via email (andrew@jupiterstation.net).