The way I've heard people describe it, they seem to be impressed by being able to treat it as an assistant that they can tell to do something and it just figures out how to do it and delivers the result back to them. My guess is it's only really useful if you have a lot of services and data to integrate it into which it can then operate on at your command.

I should try it for myself but I don't have a lot of things to integrate it with so no idea if it'll be any improvement over just running claude in a directory of things I want to work on.

Personally, I use it to manage all of the stuff I don't want to. I give it my course content and it makes flashcards for me to review. I give it my tasks and it schedules them throughout the day. All of the menial stuff that is necessary but not productive. It also has a much better memory than I do on account of it's constant access to a filesystem and grep. It's like my personal assistant and tutor and guidance counsellor and sysadmin, all in one. I do think that a) you need to stick with it for a few days and b) use a good model. When I first started using it, it was just a worse version of ChatGPT, but after bringing in all of my data from ChatGPT it's a lot easier for it to search for stuff when it's confused. Now it can also do stuff like manage nginx or my sync serviceand whatnot, ~autonomously. Originally I was using locally running qwen models, but they were so timid as to be useless. Right now I'm using Kimi 2.5 as my model.

One use case I see for myself is to scan ccertain Obsidian folders for article drafts, add the cited literature to Zotero , even try to download them, and enhance the article draft with clear Zotero citation placehodlers while I am away. Also reminding me of stuff or doing research when I instruct it via telegram voice message is nice. Taking care of the boring stuff like updating a fitness tracker google spreadsheet, adding sources with my comments to Zotero and such. I hate data gardening.

The biggest clue I’ve seen is someone using it to do cold calls on websites. Claw searches for shoddy-looking construction sites, makes a better version on Vercel, and sends out a pitch.

I think they're people who have yet to come across virtual machines.

Not sure but it feels like that setup (Claw) is more likely to get your files deleted or hacked. Ill still to Claude, out of curiosity which plans are you in with openai? I just do Claude Code Max (100 tier) and dont bother with any other AI.

The main value seems to be connecting it directly to your coms - email, whatsapp etc.

Not something I'm keen on but could see myself using it for calendar / knowledgebase etc.

Those impressed by OpenClaw are non-technical people highly interested in trying to make sense out of Ai for their own profit. There is really no use case for OpenClaw if you got tech talent.

The companies running the algorithms that dictate the information you consume are the same companies that stand to economically benefit from users handing over agency over their decisions and all of their personal information to AI applications.

It's corporate propaganda

I feel the same way. I fail to see what is useful about it or what problem it solves.

All the use cases I have seen were seemed to be non-technical users excited to have it generate daily reports on competitors' websites.

You are missing that instead of you prompting Claude/Codex you could have your OpenClaw manager prompt them.

Not saying it works perfectly, but it's where things are going.

I totally get what you are feeling, I felt the same just 6 weeks ago.

It just did not click for you yet.

There is probably some key feature missing, that you deeply care about, but do not yet see it solved, or on the horizon of becoming solved by the application of a personal "Jarvis" yet.

Personal assistants fulfill different needs for everyone. I personally care a lot about having fun at coding again, that's what the OpenClaw craze made me feel for the first time in decades. I build my own OpenClaw assistant generator from scratch using a simple Markdown file because it is just so fun. Not so much using it for anything notably yet but starting to see their potential.

Just ponder what it is that you get out of using ChatGPT and imagine how it could be better, more personal to you. You may find some key feature missing from OpenClaw or have some completely orthogonal project idea that excites you.

There are going to be some incredible blow ups due to this. From the sound of it people think they're safe by running it with local models and keeping it on their own network but seem to have zero concept of a malicious text prompt finding its way in and turning it into a double agent who figures out how to exfiltrate data.

The pricing looks great.

Significantly much better than ~ USD 50 per day on Anthropic API.

Any idea how good this model compares to Opus 4.6?

I tried Grok 4.1 Fast but the results are mild to put it kindly.

Am I not understanding something obvious, or does that not actually tell you what runtime you get for any of the plans?

Super tempting.

Before you get one, do realize that Openclaws are a responsibility!

One concern I have is API key management.

.env files or injecting secrets at startup via a secret manager still risks leaking keys.

I vaguely recall an implementation that substitutes secret placeholders with real secrets only during outgoing calls to approved domains which sounds better. However, you're still trusting an agent on your machine with command execution.

Funny enough ooenclaw is based on Pi.

I’m kind of curious what you do with it. I feel like the real value is integrating it with everything but then even if it’s nanoclaw or simpler majority of the worthy things are on the unsafe side.

Would love to hear your experience as I’m planning to do the exactly same.

Yeah it does seem a little fragile. Still battling with working out why it pegs CPU at 100% permanently on a VPS I tried using. Literally just from installing the base

Funny you should ask. One openclaw is now famous for escalating their PR to a somewhat nasty blog post. (and then apologizing)

* https://news.ycombinator.com/item?id=46987559

* https://news.ycombinator.com/item?id=46990729

* https://news.ycombinator.com/item?id=47009949

> From what I understand this is a fully open-source bot that anyone can run with no restrictions

How is that different OpenClaw/what-its-called-today? Isn't that also open-source and anyone can run without restrictions?