There is none. It's just a way for coders to feel or be able to say they "work with AI" imo. Same with doing light wrapper coding to do agents stuff. The real AI work is on actual math and ML with the internet scale data, but only four big companies does that and this is the closest regular coders can get.

Could be a psyops by Anthropic to make people waste Claude tokens and rack up a massive bill.

Tokens generated is a nice metric that they really care about.

I guess we are just boring and/or unimaginative. I don't get that many communications per day to require an abstraction level between me and the messages. The daily automations I need are more efficiently carried out by home assistant / n8n. I'm not in a position where I need automated briefs on every new company started in my area. I genuinely don't see how it could benefit me.

Most humans are unimaginative because to be imaginative is actually really, really hard. People are also incredibly overly optimistic about their own ideas etc... until they go through the craftsmanship of producing something great.

Most peoples thought process is "oh great idea, just gotta do this and that and out pops something that'll improve peoples lives". Erm no... its nothing like that in reality.

n8n can't handle abiguity, unless you program it to.

Like if you want bad weather alerts, you need to specifically define every situation of "bad weather" by givin exact weather / wind / rain limits.

What happens if it's cold, windy and rainy, but every value is juust 0.1 over the threshold of notifying you? It's still objectively shit weather.

Or can n8n go through your emails and check for promo codes or sales you're looking for - in ad mails that are 90% images with text?

I'd counter that n8n (or most other workflow tools) can handle as much ambiguity as OpenClaw - it has a LLM call node. Stuff the ambiguous parts in there, but don't burn a rainforests worth of compute figuring out how to call the weather API each and every time.

Also, in the olden days of pre-AI, if our weather workflow did not notify us because conditions juuust failed to be met, we adjusted the thresholds. Uphill, both ways.

Don't get me wrong, I use a bunch of LLMs for automations. By prompting the model "here is what I want to achieve, here are the tools I have, figure out how to stitch them together". Actual workflows run (mostly) deterministically, with a sprinkling of "classify this image" or "summarize this text" nodes thrown in for a good measure.

So what you have here is n8n-claw :)

It is not lack of imagination, it is the lack of the social need for the performative aspect of an assistant.

There are all these douchebags that use to have to hire a human assistant in order to tell other douchebags that they have an assistant.

Now every wanna be douchebag can just get a mac mini. Of course, that is going to be super popular.

It's useful for clearing out Mac inventory before the launch this week.

I don't doubt that there are people using it for legitimate stuff, but I'd wager the vast majority just set it up for the hype and to feel in the "in crowd".

I set it up, and had it do a few things, then decided its too risky after seeing some of the drastic failures it had caused some people.

Sure I understand you can sandbox it and all, but even then I couldn't think of much stuff I wouldn't want to do myself just nor justify the cost to run it.

Wannabe Tony Stark love these gadgets, and there are a lot of them out there. Just look at what tech content is trending on youtube &co these days, we got gangrened by influencers like most other hobbies/lucrative industries

It's useful for producing content about how you're using it.

Here are some of the things I did with it while running locally: - Ask it to perform a scan of your local network and give you advice on output - Tell it to login to various computers and re-boot them (I have a few servers I host and setup openclaw to have a user on them) - Replace web search by asking openclaw

It's neat but the token use is pretty inefficient and security of course is a mess but it's been fun to play with.

I am messing with NanoClaw now and it's pretty much the same but only support Claude (uses code to do everything)

I don't see how any of those require a constant-heartbeat loop. Those all work just fine in claude code / cowork.

And in reality most of what does need a heartbeat loop can also easily be automated by just asking Claude to set up a cronjob. I think genuinely the most "novel" thing about something like OpenClaw is just that it "feels" more like a "real entity", like a partner rather than a chatbot, and for some reason that resonates with people. Whether that's by itself kind of a huge red flag or kind of a nothingburger, everyone has to decide for themselves.

Do you really need an AI agent to reboot a computer?

This takes maybe 10 minutes to write a script for…

There is a thread from February with more credible use cases from real users. As someone said, it does what everyone expected Siri to do by now.

Which one? None of those that came up when I searched were really containing a lot of real uses. Both top threads[0][1] don't really contain much of substance.

[0]: https://news.ycombinator.com/item?id=46838946 [1]: https://news.ycombinator.com/item?id=47147183

Separating fools from their money and data

you're talking about a free tool you know that right?

Not many people are using local LLMs for their OpenClaw backend, so most are paying money to OpenAI/Anthropic/etc. and getting their data siphoned as a bonus.

You give it your etrade login and retire early.

Retire under a bridge...

Did they stutter?!