I disagree with practically every take here, and likely will get hate for this message.
Personally it has been a huge unlock and here are some specific use cases.
1.Context engineering for “any task”: my claw agents debate with me, and write me files to be picked up by Claude code or Claude cowork to write code/build presentations.
2. Discovering this page: my personal newsfeed that took me 10 mins to setup. Never had any newsfeed before this which could be personalised like this.
3. Strategy: similar to point 1 above. I set up a “tough-coach” skill. Now all my 1am ideas go through the wringer of why they suck, and what, if anything could be differentiated about them.
4. Many other small-ish things that are important enough for me to spend on them, but not important enough to invest my time.
I run a conversational AI startup in India working mostly in Financial inclusion space - so most of time is spent in meetings (:-/).
With openclaw, the 2-3 hours before I sleep (low energy, high on ideas), convert into pretty productive sessions. No more “this is a great idea, I will do it tomorrow”. Instead “oh this sounded great but my agent just called it shit - what if I modify it a bit, and then let’s see”.
> my claw agents debate with me
I don't really want an imaginary friend. (Or intrusive thoughts either.)
Honestly, the best use of AI I've found so far is also getting them to call me out on bullshit assumptions and logical fallacies. Helpful for the kind of thought exploration + planning that another human won't want to discuss at 6am.
Seems like all of this can be done directly with chatgpt and claude? Why add complexity of running claw?