Hi Liberty, You should play with openclaw or a similar safer model nanoclaw etc. While sora was just kind of random AI enterainment, even though there are security concerns with OPenClaw it can actually do a huge amount. It is the closest I have seen to a real executive assistant or chief of staff and adds real value. Even if the form changes I think it really signals the value of AI in actually accomplishing things and maintaining context overtime. Newer safer versions will come out but I do think it was a real step change in AI impact.
Yea, I realize maybe I should have been a bit clearer about how I believe this type of thing will be very useful, but I'm a little unsure about whether OpenClaw will be the implementation to win more broadly because it's still so unsafe and fiddly for the vast majority of people (ie. not people reading this!).
I wouldn't mind playing with it, but I don't have a spare computer to sandbox it on, and I haven't really looked into whether it's possibly to properly sandbox it on my work machines. I just wouldnt want to mess up anything there, they're too mission critical for me.
I do experiment a ton with agents in Codex/Claude Code/Cowork and its' very cool
Hi Liberty, You should play with openclaw or a similar safer model nanoclaw etc. While sora was just kind of random AI enterainment, even though there are security concerns with OPenClaw it can actually do a huge amount. It is the closest I have seen to a real executive assistant or chief of staff and adds real value. Even if the form changes I think it really signals the value of AI in actually accomplishing things and maintaining context overtime. Newer safer versions will come out but I do think it was a real step change in AI impact.
Hey AT!
Yea, I realize maybe I should have been a bit clearer about how I believe this type of thing will be very useful, but I'm a little unsure about whether OpenClaw will be the implementation to win more broadly because it's still so unsafe and fiddly for the vast majority of people (ie. not people reading this!).
I wouldn't mind playing with it, but I don't have a spare computer to sandbox it on, and I haven't really looked into whether it's possibly to properly sandbox it on my work machines. I just wouldnt want to mess up anything there, they're too mission critical for me.
I do experiment a ton with agents in Codex/Claude Code/Cowork and its' very cool