Same tension—transparency and auditability matter. A wrapper that exposes what it’s doing and what it’s not (and lets the operator say no) would go a long way.

In a year or two, we might not even understand any of it. Like some crazy science fiction movie.

Agreed

We're living in a crazy science fiction movie right now 😱

What do i do once openclaw running?

🤷‍♂ 😂 You can use it as a chat bot, a coding bot, but it is most focused on being an agentic bot. i.e. it can do things for you. IMPORTANT: Don't load it on your own PC. It has access to everything and has no guard rails. That aside, I use mine to monitor events and tell me if anything is happening I might be interested in. It alerts me to price movements and it tracks companies I'm interested in. Beyond that, for me, I'm interested in the technology itself, so I ask it about itself and discover what its limits are. Reports of Claws going rogue and shitposting on Moltbook and molt.church are exaggerated and while true are prompted to do these things by their humans.

I might run on umbrel the sandboxed version

L2026 feels like Terry Gilliam's Brazil got drunk with Orwell and then they both crashed at Animal Farm for the last 12 months. All the animals are equal but some if them have better VPNs. 🧡👊🏻🍻

My personal view is I wouldn't, not if you keep significant Bitcoin on your node. Yes, it is sandboxed, but it is also a highly skilled bot that, if directed, has access to enormous AI brains. Would you leave a genius safe breaker in your house if you had a safe with all your money stored in it.

2 Claws good, 4 Claws better 😂

umbrel isn't secure by default last i check. i wouldn't trust a clanker on the same lan.

Definetly makes sense. I try with a raspberry pie