I wanted to try OpenClaw, but I've seen close friends use it and am not convinced it would be helpful/useful for me, and am also scared by the number of concerning stories of security vulnerabilities, etc.
This resolves YES if
1) There's an AI personal assistant, with functionality similar to OpenClaw, released in 2026 that I feel comfortable using / installing / trying out
AND
2) It is "good" in the sense that it works, doesn't require extreme technical know-how to configure, and does basic personal assistant functions even remotely competently
For example, if OpenAI / Anthropic / Google release a mainstream version of OpenClaw, I expect it would likely although not necessarily be sufficient to resolve this YES.
I don't expect that other open source vibecoded projects like OpenClaw would be quite enough to resolve this YES, judging from what I've seen, although it's possible.
I don't actually have to personally install it for this to resolve YES. It's possible it costs $50/month or something and I just don't really want to get it, but if I can see many other people using it with good results, I would resolve YES.
Resolves per my judgment, obviously, so I will not bet on this market.
1,000@bens How strict is the definition of "AI personal assistant, with functionality similar to OpenClaw" here? Ie, does that mean it has to be fully autonomous and be interactable and active outside of a single dedicated chat interface? Which reason(s) is something current like Claude Cowork excluded under?
@Dssc Ya, basically. Claude Cowork is just not a "personal assistant", per se, nor is it intended to be. It's just Claude Code for ppl with less coding expertise who don't want to work in the terminal.
I'd be looking for some combination of:
-autonomous email mgmt, messaging, calendar mgmt, things like that
-being able to interface with it through conventional communications means (ie, not going to my computer, opening up the software, and prompting it)
-curating notifications, bringing things to my attention, filtering out spam and bs
-going and agentically performing tasks on my behalf without explicit prompting (although hopefully with confirmation for anything actually important)