04:15 UTC

6 weeks of using openclaw and 2 weeks of hermes.

Ended up killing my openclaw instance after setting up a fleet of 5 bots.

I built clawed-club and tried to automate an openclaw ai newsletter. It was fun, but im killing the newletter too.

My learnings. The speed to boost to get an idea built from nothing to something is incredible. Giving an AI access to a computer to run commands, works crazy well. Until it doesn’t.

Openclaw and hermes can brick itself. Perhaps cos im using a cheap model…

I broke my openclaw instance numerous times. It forgot who it was at one point.

Openclaw became a bit of a maintenance nightmare.

Hermes seems better for now.

It was fun to create different personas for different projects I was working on. I set up cron jobs related to different tasks. E.g newsletter curation, day round up. Web crawling.

I became the ceo of a fleet of agents. And it got very noisy.

I turned them all off.

I have one hermes bot now. Its calmer and im more deliberate with its usage.

Was openclaw a failed experiment? Sort of. But I learned a lot from it.