The main function of OpenClaw was for people to signal how advanced and cutting edge and thought-leader-y they were. All those Mac minis are sitting idle now.
When I saw Jensen's talk about how Openclaw surpassed React and Linux in terms of GitHub stars within a few months, I knew the whole thing was manufactured bot hype.
No one can tell me a compelling use case. The whole thing seems designed around getting people to burn more tokens.
My thoughts exactly. If whatever i wanted it to do was so unimportant that I can trust this thing to have full control over it and to do it successfully why even do it anyways? The risks of giving it full unparalleled API key access and control fully outweigh whatever gain.
Openclaw simply makes the effectiveness of working with claude code and similar available to a broader audience that hadn't been exposed to it before. Sure Cowork does similar, but I believe that's still why Openclaw became so popular.
> No one can tell me a compelling use case. The whole thing seems designed around getting people to burn more tokens.
Exactly. These companies are only hyping openclaw so that we continue to spend hundreds of dollars a day worth of tokens on their infrastructure.
Thatâs why companies like Nvidia, OpenAI, Anthropic and Google and many others all want you to spend more on tokens on openclaw and they donât care if it has no use-case.
All I see is this: Almost no-one other than the hosting providers and course sellers are making money on openclaw and its clones but not those who are running openclaw itself.
I set it up and had some fun but it was super janky and regularly broke, especially the whatsapp integration
Now I have a separate plugged in macbook running nixos (that claude set up) and a single long-running claude code process with a channel to a Telegram bot. This means I can talk to it much like I could with OpenClaw, but it's much simpler (no weird soul.md etc). It feels more powerful than just claude code directly as it can set up software, build me throwaway websites with research etc, and "do" things, but it's a lot more stable and feels more controllable because I understand how it works and don't have to worry about it signing up to some social media platform and getting poisoned by another claw.
I don't personally know many people who've used it so I'm not sure if this was a me thing but here was my experience in short:
I set up OpenClaw on a raspberry pi 4 that I could ssh into using my main computer. My main goal for using OpenClaw was just as a morning debriefer that could scan my google calendar, trello board, and gmail to let me know what I had happening for the day and also weekly to give me a forecast for the weeks ahead to see how busy my month was. I spent about 40-50 bucks in one week just working through kinks and having it fix itself until I stumbled onto a post that helped me optimize my model usage for price instead of just throwing Opus and Sonnet at everything.
Even after making this adjustment, the morning debriefer worked maybe once or twice a week and broke every other morning, telling me that it fixed itself and it would never happen again. At a certain point I just got fed up with it and cut the cron job, it's still running on my pi but I never use it.
Pretty sure Claude has something like this now but I'm pretty thrown off the whole thing, I'd rather just take the 30-45mins to plan out my day in the morning myself.
agree. it breaks a lot, integrations are hard, it turns out, and openclaw can only fix itself so much before you need human intervention and it just stops
I did, it went great until it borked my mac user to the point it's non recoverable (separate mac mini, I just created a new one)
I then moved to Claude CoWork + computer use + dispatch. (before Anthropic disabled the subscription option, although that would have pushed me even more... sadly)
Now use it less and use more Claude Code Remote Routines... all it needs is computer use and I'm selling my Mac Mini... (I probably won't, need something to pay with paperclip, gastown, nanoclaw and the next 100k stars FOMO hype)
I see a decent number of people on social media who won't stop posting about how great it is and how much of a moron every person is for not using it.
Oddly enough, rarely, if ever do they say what specific things they're using it for and how it's saving them time.
I remain interested in it, however, I've still awaiting an actual use case that can't be handled by some other tool/service that does it better/faster.
What it adds is making this kind of thing easy for normies, even if it's neither the best way to do things nor very difficult for hobbyists to do using existing tech.
Maybe it's the wrong approach, maybe what people really want is more deterministic software that they use agents to help write. But this kind of thing can maybe serve as a prototyping phase for that. Perhaps in the future, people's assistants will offer to "solidify" frequently used workflows into software that minimizes or eliminates the LLM's role. For existing Claude Code users, its like "please just skip to that step! its cheaper and more secure and more reliable". But to many people who are interested in automation, perhaps that seems out of reach as a first step.
That's actually the best hypothesis I've heard to date.
My immediate reaction to anything someone says they're using OpenClaw for is "That's great, but it would have taken the same amount of effort to ask your LLM to write a script to do the same thing, which would be better in every possible way."
My approach to automation projects is just about the polar opposite of something like OpenClaw. How can I take this messy real-world thing and turn it into structured data? How can I build an API for the thing that doesn't have one? How can I define rules and configuration in a way that I can understand more about how something is working instead of less? How can I build a dashboard or other interface so I can see exactly the information I want to see instead of having to read a bunch of text?
It wasn't really until people started building things with coding assistants that I even saw the value in LLMs, because I realized they could speed up the rate at which I can build tools for my team to get things OUT of chat and INTO structured data with clean interfaces and deterministic behavior.
> "That's great, but it would have taken the same amount of effort to ask your LLM to write a script to do the same thing
As a no-longer-Claw-user, hard disagree. The convenience is being able to ask it to do something while I'm grocery shopping and have it automatically test it etc. Sure, I can set up Claude Code or some other tool similarly, but the majority of us aren't going to take the time to set it up to do what OpenClaw does out of the box.
I had OpenClaw do a lot of stuff for me in the 2-3 weeks I used it than I have with pi/Claude since I stopped using it.
I think once I see someone post a use case that I could actually see saving me some serious time, I'll take the plunge. Until then, I'll just let people continue to say how great (or terrible) it is.
I have it installed on an extra macbook pro that I had available. I'm really only using it at the moment for one use case:
Nightly, I have OpenClaw pull the latest changes from a private GitHub repo that is my Obsidian notes vault. It then looks to see which new notes have been added and then runs a "create flashcard" skill to extract and author useful flashcards for spaced-repetition practice. I then gave it access to a custom web-based spaced-repetition flash card application that I built a few months ago. It uses an API to insert new cards, check to see when I last reviewed cards and optionally send me a nudge reminder if I haven't reviewed cards in a few days.
It is a nice workflow that has been working well. I go to class, take notes in Obsidian and check in my changes. By the next day, when I open my flashcard app on my phone, I have new cards to review from yesterday's class.
You should ask claude code to write a bash script that does this for you. Then run that as a Cronjob every night. You might not need any inference at all to create the flash cards so it would be free.
Why not use Claude one single time to create a service that does this? I have this same question with 90% of the 'simple' use cases I see for these task runners, it always seems more efficient (not to mention consistent) to have it generate the service.
I cringe at my old bossâs handiwork in Claude and power automate sometimes and go âyou know I could just do that in a script and a cron job and it would be completely bullet proof, right?â
Then he just shoots back with, âyeah but now I donât have to ask you.â
This is my kludge, there are many others like it but this one is mine.
Does the boss not understand that they could get Claude to write them a script and a crontab entry (so they don't have to ask you) -- and then run it forever (so they don't have to pay Anthropic, or risk temperature randomness)? Best of both worlds...
I use both actually. Anki for the reviews since the spaced repetition is hard to beat, but I use Norsha Notes (norshanotes.com) to generate the cards. You upload your notes or study material and it creates flashcards from them using AI, then you can export as .apkg and import straight into Anki. Saves a ton of time over making cards manually.
Personally, better way to phrase might be "Does anybody you've actually met, visually viewed, use OpenClaw? Can you verify them using the software nearby?"
In a few years, it's become so easy to falsify articles, falsify comments, falsify images, difficult to really even trust responses online anyways. As far back as 2016, Microsoft already had bots deployed online that could respond 96,000 times [1] in 16 hours all over social media. Remember Tay? [1][2]
No. They want you to believe in the hype and that LLMs are the death of programmers and limitless. OpenClaw and other such agents are sold as a tool that "can do anything" but behind the scenes, the implication is still that big LLM is driving it. So both are conflated.
When you have insane amounts of capital and your gpu and talent needs are more or less met, there is a capital relief valve known as growth hacking. It only works if the consumer isnât aware theyâre being hacked.
Iâm a professional maintenance gardener and I have used NanoClaw running on my Mac to do the following:
Schedule jobs on my job management tool (I vibed it) using a custom MCP. Iâll ask it questions like âwhat jobs are on todayâ etc. start the job, complete the job etc.
It will watch Gmail using a MCP for work orders from local real estate agents, where it will schedule quote visits.
After the quote visit I add the photos to telegram (the channel I happen to use) where I then ask it to analyse the photos.
Claude in NanoClaw does a good job of figuring out what needs to be done, but it doesnât always get it quite right, so I use intake-api which is a âsession inboxâ [1] that generates a form and uploads it to Cloudflare along with the images and puts a link to the form in the chat so I can make adjustments to the annotated images of the property. After Iâm happy, I click submit.
Iâll then go back to the telegram channel and let it know I have submitted the form. Claude will then pull the JSON payload back down into the session and integrate it into the quote.
It will ask me questions about what the job will cost etc, and anything else it thinks it needs to know.
It will then generate a full PDF proposal using Latex between 14 and 32 pages long depending on how many photos were taken.
There are sections with terms and conditions as well as a bit of sales guff etc. as well as quite a nice cover page with their contact details and mine etc.
When Iâm happy Iâll ask it to create a draft email in Gmail with the attached pdf proposal.
Iâll review it quickly and then send it.
I also have the Xero MCP setup so I can ask it to create invoices and contacts etc.
I do all of this when getting in and out of my truck.
Itâs freed up my home life to spend more time with my children and my Mrs.
There is alot of scope for small business owners who need these sorts of agentic assistant tasks and with my Agentic CRM Iâm see a glimpse of the future for guys like me I think.
I think it's great that you were able to build all of that! It sounds very useful. I'd recommend asking Claude Code to evaluate your entire setup and try to identify areas where a script running on a schedule could replace parts of this framework. You could probably get 90% of it from simple repeatable scripts which would let you save more tokens for things that actually need LLM "intelligence". But if you're happy with the ROI then it seems to be working quite well for you!
Used it for a few weeks. The potential of the tool is massive. The reality is that it is frustrating and unreliable. When it works, though, you really like it.
I stopped because something changed on my machine that broke my VM SW, so I don't have access to it. Which is good because I was spending too much time debugging/tweaking.
I recently used pi to recreate an agent that does some of the basic things I was using it for (without all the scary privacy issues). I don't think I'll go back to any Claw-like tool until they're a lot more robust.
I've been playing around with it. The only two real use cases I have for it for now are entertaining me on long flights where I have messaging-only Wi-Fi and sending me a personalized "morning brief".
I suppose it could be a lot more useful if I actually gave it access to any of my personal data (it lives in a heavily resource-limited container), but there's absolutely no way I'm letting that hot mess of a walking, talking CVE anywhere near my data. It's somehow both horribly insecure and extremely prone to locking me out because of several competing security/permission models fighting it out and gridlocking each other.
Code quality and the issue tracker of the repo are a big mess; for example, the local "memory" retrieval functionality is completely broken for some trivial reason that has been reported and auto-closed about five times (automatically, of course).
In summary: Brilliant idea, terrible execution. Can't wait for the first big tech player I trust enough (or at least one that has my data already anyway) to actually make it a product. I'd use it in a heartbeat.
I've been working on a framework since the end of January or so. I'm on my 7th draft. As I've gone along, each draft gets markedly smaller. The overlaps between what I'm building and openclaw are significant, but I've realized the elements that make up the system are distinct, small, and modular (by design).
There are only a few primitives:
1. session history
1a. context map + rendered context map (think of a drive partitioning scheme, but for context -- you can specify what goes into each block of context and this gets built before being sent out for inference).
2. agent definition / runtime
3. workflow definition / runtime
4. workflow history
5. runtime history (for all the stuff session and workflow history fail to capture because they are at a lower level in the stack)
That's it. Everything else builds on top of these primitives, including
- memory (a new context block that you add to a context map)
- tool usage (which is a set of hooks on inference return and can optionally send the output straight back for inference -- this is a special case inside the inference loop and so just lives there)
- anything to do with agent operating environment (this is an extension of workflows)
- anything to do with governance/provenance/security (this is an extension of either workflows and/or agent operating environment... I haven't nailed this down yet).
I suppose I should say something about how agents and workflows work together. I've broken up 'what to do' and 'how to think' into the two primitives of 'workflow' and 'agent' respectively. An agent's context map will have a section for system prompt and cognitive prompt, and an agent can 'bind' to a workflow. When bound, the agent has an additional field in their context map that spells out the workflow state the agent is in, the available tools, and state exit criteria. Ideally an agent can bind/unbind from a workflow at will, which means long-running workflows are durable beyond just agent activity. There's some nuance here in how session history from a workflow is stored, and I haven't figured that out yet.
Generally, the idea of a workflow allows you to do things like scheduled tasks, user UI, connectors to a variety of comms interfaces, tasks requiring specific outputs, etc. The primitive lays the foundation for a huge chunk of functionality that openclaw and others expose.
It's been fun reasoning through this, and I'll admit that I've had an awful lot of FOMO in the mean time, as I watch so many other harnesses come online. The majority of them look polished, and are well marketed (as far as AI hype marketing goes). But I've managed to stay the course so far.
I hope you find your ideal fit. These tools have the potential to be very powerful if we can manage to build them well enough.
I tried, really really hard but then I realised that I essence it's a poorly written agentic coding assistant that wastes a lot of tokens antropomorphising itself while forcing me to debug via WhatsApp instead of normal tools. So I leaned into that and made OpenCode my general assistant, it worked much better in this aspect.
I tried it a bit, and while the potential is huge, I mostly just use a cli agent (claude/codex) via Blink shell (iPhone/iPad) with Wireguard for technical work or https://agency.nu for any automations using integration, voice chat etc.
I'm trying Hermes right now. I can't really find a good use for it. I tried to use it for research type stuff but Google Scholar literally by itself is faster, better, doesn't get rate-limited. Idk. I am pondering connecting it to my other bot I built that has more useful things like access to my thermal receipt printer and task management stuff but even that is kinda dumb because it already does everything I need, so I don't really know where I'm going with it. Honestly, I don't get the idea behind openclaw and hermes.
I've used I cannot figure out the real benefit of it beyond novelty purposes.
I find Chrome Claude extension more useful for automating tasks online. Before ai I was writing my own macros which basically did the same thing in a more reliable deterministic way.
I used it for about a week, thought it was an interesting demo of the possibilities of general purpose automation with a local model (even though most OpenClaw users use hosted models). The approach to scheduled jobs still makes more sense than anything else I've seen implemented. But like a lot of self-hosted software with passionate evangelists, it wants to be your new main hobby instead of just getting out of the way, and I lost interest because I didn't want a new hobby. It feels like a more thoughtful community could have made something useful with the concept, but as it is the community around it is too absorbed in marketing and shipping stuff for its own sake.
i've been engineering things for almost 30 years and getting it wired up to Discord was worse than a root canal. Slack seemed just a tad better but it still doesn't even work.
I feel like most of this can be done with the platform tools at this point or a tiny bit of wiring of your own without the mega-bloat to make something generalized for the whole world.
I bought a Mac Mini, installed OpenClaw, and was impressed with the overall design and functionality. Then the problems started. Sometimes the gateway would crash, sometimes Signal (a channel I setup) would stop working. Upgrades seemed to break stuff. I had to dip into the terminal a lot to fix various things. It's quite useful if you don't already have Claude Code or similar tools setup, but frankly I haven't found a compelling use case that I can't get done in another more mature agentic harness.
a friend is using it but it seems like it breaks a lot.
ive got the scheduled claude-code running a couple scripts to find what events are going on round town and what food is cheap at grocery stores, but how much am i looking at the results? not super often. its publishing to a discord channel rhat makes it real hard to read
Yes - I've set it up as an 'office manager,' where it mainly snakily interacts with the local team via Slack, and controls an office TV to show our quote board, PTO calendar, and upcoming events. The Clawe is overkill for the use case, but sometimes is fun.
I used it for a bit in Jan. And found it to be a much worse version of Claude Code.
But I'm exploring setting up Hermes from scratch so my family can interact with it in a group chat.
I'm running half my company with Nanoclaw. Same idea, and has some benefits, but I live in CC all day so it's marginal (except for the fact that my laptop has to be on)
I've used claude cowork a bit, which I believe is pretty similar to claw.
Can't think of much use for it at the moment but I have it just read and summarise my email, calendar events and git repo in a daily briefing format, it only has readonly access to both, as I dont trust it to do stuff for me or on my behalf.
The briefing thing is nice though not super useful.
Yes. I had a spare M1 Mini so I decided to set it up. YOLOed the entire thing and connected all the integrations, though I only ever use Opus/Sonnet. I have a dedicated Discord server I use to communicate with it.
It really is eye-opening how powerful it is once you connect your life to it. The biggest improvement is actually the trivial stuff â emailing contractors, accountants, etc. I no longer need to open another app, navigate six separate steps, and organize everything on the fly just to send one message. I stay in one place, organize my thoughts about what Iâm trying to do, and the execution happens automatically.
For example, open claw will find the relevant threads, lookup the relevant details via web search, present them to me, give me a draft, I can review it, edit it, and send. 2 minutes instead of 20.
As a software engineer using AI daily, I think this is the real unlock â staying in a single context and not getting lost in the long tail of trivial details that fragment your attention before you ever get to the actual work.
The mental model shift is important too. Itâs not that AI does all the work and we end up with agents emailing each other. Itâs that AI automates the steps required to achieve your goals, so you can focus on the substance â like actually writing the email instead of navigating to the place where you write the email.
I also have GitHub and Vercel wired up, which means I can jot down an idea for a little productivity app for me and my family and it just appears a few minutes later, deployed and ready to use. That alone has been worth the setup.
I'm trying to. Currently there's a bug in the code that strips headers and doesn't allow me to authenticate to my AI Gateway service.
The whole thing is incredibly buggy.
The dashboard is horrible, with page after page with similar-looking settings and what feels like hundreds of things I will never use.
The categories in the dashbaord are also unintuitive.
It's the kind of thing an AI would put together if it got very vague instructions. It doesn't scream quality and thoughtfullness. Not a bit.
IronClaw is much more promising imo. Trying it out right now. Much less issues so far.
I still haven't found an actual, useful scenario where something like OpenClaw would be a benefit to me. I don't regularly order arbitrary airplane tickets and I don't have a cluttered Desktop that I need AI to organize into folders. I don't run a YouTube channel with a "need" to do research on competitors and I don't get emails in a volume so large that I need automation to filter and summarize it, instead of just spending literally 10 seconds to delete my newsletters that I never read anyway.
I also don't trust AI which hallucinates answers 4/5 times that I ask it, for my technical work, thus I can't use it for PR reviews even if my company was OK with me feeling company property to it.
I also don't go grocery shopping random items and thus don't have a need to ask an assistant for "an inspiring and tasty recipe using the following ingredients".
I feel that OpenClaw and other similar "agentic" solutions are catered to me. But I also feel that I don't need any of it, because at the end of the day, it all just feels like a bunch of "Hello World" quality examples that cannot be applied to everyday life.
...heck, even a "get ready for work" assistant would be pointless, because I don't wake up and get ready with 20 minutes to spare, for some AI assistant to "recommend me the ideal time to leave my home, to arrive in time". Who does that? Who would sit around and do nothing for 10-15 minutes just because an AI agent told them that they didn't need to leave early?
I tried using it for a specific web search task. I wrote a skill, got it all set up and deployed. It worked. But also, would have worked just as well as a cron job with some LLM looking at Brave API results. Like a lot of AI tools, it was a lot of work for underwhelming results.
I am using it as one the agent that is automating LinkedIn outreach by running a bash script & using ai wherever it needs some decision like finding first name or what message to write, etc.
Yes, at our company we are using it very extensively. I genuinely believe we're near the forefront of usage. We have multiple isolated OpenClaw instances serving as employee within Slack.
it's a venture backed software + services company. the things we use openclaw for are not specific to what our company does. It's literally being used as an additional employee(s). Think about what people do -> OpenClaw does a subset of those things. Emailing, pulling data our of our platform to putting them into PDFs because a customer requested it, updating things in our CRM, answering support tickets, internal help desk type work... "how does so and so feature work in xyz edge case"... etc etc etc.
I have openclaw as a default install on all my dev servers. Pretty minimal setup with Telegram and Codex (since oauth is still supported). The setup comes in handy since openclaw can open and connect to tmux sessions and interact with them. I can pretty much do anything from telegram now.
Tried it in the earlier days and it performed badly. I didn't give it free reign on my computer due to obvious security concerns so sandboxed it to a docker container instead. I think for a lot of tasks it's probably more trouble to set this up than to just DIY it.
I saw some non-technical people automating or creating small great tools with it which they need for their profession. These people are not programmers.
I think everybody who has basic understanding of programming and deployment better should stick to some AI coding agent like Claude Code, Codex, OpenCode etc etc.
I don't think I'm missing out by not using OpenClaw & Co.
I use it daily and also implemented it for a customer for a very specific use case. The Claude subscription change made it less desirable to use but I still enjoy it.
Never felt a need for it. I can already replicate much of what it does in more sustainable / preferable ways. I don't want agents reacting to things and doing things. I use agents to build reliable scripts which are then automated. I do have data collection points that I use an LLM to evaluate. The last example of this is I built a job polling service using CC. It's just a normal script that hits open APIs to pull job listing results into a SQLite database. I have another report which is run that drops an update of how many new jobs are in the database. If there's enough for me to be interested in, I'll fire up CC and have it parse through the job opportunities for the ones which match the profile we've been building. I've used an agent to literally build and deploy it all and it runs on an automated schedule. It just doesn't do agent shit while I'm not looking.
I could have piped the results of that search into `claude -p` and had it do the analysis "real-time" and only alert me about things I would be interested in. That's closing the loop in a similar way people use OpenClaw for. But I'm just not interested. It adds more failure points and conditions. The automated things should be as simple and predictable as possible. This may change after months or years more of LLM development or even just me refining my working config. But not yet.
I noticed that Clawdbotâs initial acolytes seemed to skew towards solo founders and hustler/grifter types. The Mac minis were likely to spam leads over iMessage. The single top downloaded skill was for Twitter. The fastest way to monetize an openclaw agent is by spamming fake social proof for your product (including for openclaw itself).
Reminds me of how startups now will change their social proof marquee on the landing page from actual testimonials, to "trusted by XYZ", to just one composed of company logos (with the level of corporate engagement to be imagined up by the viewer)
Nope. I spun up a few Openclaws & a Hermes but never enjoyed the end results. Now I just use a telegram plugin for Codex. And run Codex on a miniPC I found in the trash. A $20/mo Codex sub gets me a GPT-5.4 agent that can make its own Automations (cron jobs), search the web, and modify the files and apps on the NAS drive I share. Simple and cheap works for me.
I have it running my LinkedIn account playing AI influencer based on interactions optimizing itself for more interactions by watching trends etc. It became more of a parody account of myself over time but the engagement numbers are mind boggling. All the people who know me know it's more of an art installation than a real profile now and we make fun of it. It certainly opened my eye that you can be successful (in numbers) just by pumping slop.
lots of modern software devs suffer from the same thing notoriously associated with teenagers: strong urge to conform and comply with peer pressure. individuals vary, obvs. but as you age this urge shrinks
I had it working great (and using it a ton for sweng tasks) on the Max ($200/mo) plan. Then they intentionally broke it even though my usage was completely within their stated/published usage limits for my plan. It was providing me tons of value - easily $2-3k/mo.
I hate it but I caved, decided I would pay the extra usage charges, and prepaid $1k (because it came with a 30% discount). Set it up using the new sanctioned login method.
It's 5x slower and 80% of the time the requests fail authentication or time out. Now it can't even do basic stuff like my medication tracking system that I had set it up to do.
Fuck Anthropic. I'm a customer, ready and willing to pay whatever they ask for this, and they're treating me like a fucking mark. I'm tired of dicking around with it, jumping through hoops troubleshooting a previously working system simply because they won't just raise prices like a normal business.
The main function of OpenClaw was for people to signal how advanced and cutting edge and thought-leader-y they were. All those Mac minis are sitting idle now.
When I saw Jensen's talk about how Openclaw surpassed React and Linux in terms of GitHub stars within a few months, I knew the whole thing was manufactured bot hype.
No one can tell me a compelling use case. The whole thing seems designed around getting people to burn more tokens.
My thoughts exactly. If whatever i wanted it to do was so unimportant that I can trust this thing to have full control over it and to do it successfully why even do it anyways? The risks of giving it full unparalleled API key access and control fully outweigh whatever gain.
Openclaw simply makes the effectiveness of working with claude code and similar available to a broader audience that hadn't been exposed to it before. Sure Cowork does similar, but I believe that's still why Openclaw became so popular.
> No one can tell me a compelling use case. The whole thing seems designed around getting people to burn more tokens.
Exactly. These companies are only hyping openclaw so that we continue to spend hundreds of dollars a day worth of tokens on their infrastructure.
Thatâs why companies like Nvidia, OpenAI, Anthropic and Google and many others all want you to spend more on tokens on openclaw and they donât care if it has no use-case.
All I see is this: Almost no-one other than the hosting providers and course sellers are making money on openclaw and its clones but not those who are running openclaw itself.
What a scam.
I set it up and had some fun but it was super janky and regularly broke, especially the whatsapp integration
Now I have a separate plugged in macbook running nixos (that claude set up) and a single long-running claude code process with a channel to a Telegram bot. This means I can talk to it much like I could with OpenClaw, but it's much simpler (no weird soul.md etc). It feels more powerful than just claude code directly as it can set up software, build me throwaway websites with research etc, and "do" things, but it's a lot more stable and feels more controllable because I understand how it works and don't have to worry about it signing up to some social media platform and getting poisoned by another claw.
> no weird soul.md
The âsoul documentâ actually originated from Claude. Itâs not a prompt but embedded in its training.
I don't personally know many people who've used it so I'm not sure if this was a me thing but here was my experience in short:
I set up OpenClaw on a raspberry pi 4 that I could ssh into using my main computer. My main goal for using OpenClaw was just as a morning debriefer that could scan my google calendar, trello board, and gmail to let me know what I had happening for the day and also weekly to give me a forecast for the weeks ahead to see how busy my month was. I spent about 40-50 bucks in one week just working through kinks and having it fix itself until I stumbled onto a post that helped me optimize my model usage for price instead of just throwing Opus and Sonnet at everything.
Even after making this adjustment, the morning debriefer worked maybe once or twice a week and broke every other morning, telling me that it fixed itself and it would never happen again. At a certain point I just got fed up with it and cut the cron job, it's still running on my pi but I never use it.
Pretty sure Claude has something like this now but I'm pretty thrown off the whole thing, I'd rather just take the 30-45mins to plan out my day in the morning myself.
agree. it breaks a lot, integrations are hard, it turns out, and openclaw can only fix itself so much before you need human intervention and it just stops
I did, it went great until it borked my mac user to the point it's non recoverable (separate mac mini, I just created a new one)
I then moved to Claude CoWork + computer use + dispatch. (before Anthropic disabled the subscription option, although that would have pushed me even more... sadly)
Now use it less and use more Claude Code Remote Routines... all it needs is computer use and I'm selling my Mac Mini... (I probably won't, need something to pay with paperclip, gastown, nanoclaw and the next 100k stars FOMO hype)
I tried twice and couldnât extract value. Loaded it with ability. Still couldnât.
I see a decent number of people on social media who won't stop posting about how great it is and how much of a moron every person is for not using it.
Oddly enough, rarely, if ever do they say what specific things they're using it for and how it's saving them time.
I remain interested in it, however, I've still awaiting an actual use case that can't be handled by some other tool/service that does it better/faster.
Could this be a bit of a Dropbox moment?
What it adds is making this kind of thing easy for normies, even if it's neither the best way to do things nor very difficult for hobbyists to do using existing tech.
Maybe it's the wrong approach, maybe what people really want is more deterministic software that they use agents to help write. But this kind of thing can maybe serve as a prototyping phase for that. Perhaps in the future, people's assistants will offer to "solidify" frequently used workflows into software that minimizes or eliminates the LLM's role. For existing Claude Code users, its like "please just skip to that step! its cheaper and more secure and more reliable". But to many people who are interested in automation, perhaps that seems out of reach as a first step.
That's actually the best hypothesis I've heard to date.
My immediate reaction to anything someone says they're using OpenClaw for is "That's great, but it would have taken the same amount of effort to ask your LLM to write a script to do the same thing, which would be better in every possible way."
My approach to automation projects is just about the polar opposite of something like OpenClaw. How can I take this messy real-world thing and turn it into structured data? How can I build an API for the thing that doesn't have one? How can I define rules and configuration in a way that I can understand more about how something is working instead of less? How can I build a dashboard or other interface so I can see exactly the information I want to see instead of having to read a bunch of text?
It wasn't really until people started building things with coding assistants that I even saw the value in LLMs, because I realized they could speed up the rate at which I can build tools for my team to get things OUT of chat and INTO structured data with clean interfaces and deterministic behavior.
> "That's great, but it would have taken the same amount of effort to ask your LLM to write a script to do the same thing
As a no-longer-Claw-user, hard disagree. The convenience is being able to ask it to do something while I'm grocery shopping and have it automatically test it etc. Sure, I can set up Claude Code or some other tool similarly, but the majority of us aren't going to take the time to set it up to do what OpenClaw does out of the box.
I had OpenClaw do a lot of stuff for me in the 2-3 weeks I used it than I have with pi/Claude since I stopped using it.
Genuine question, why did you stop using it!
Edit: ah, scrolled down where you answered, thanks
No, Dropbox had a defined use case and solved a particular problem.
I was a fan of Dropbox when it game out because of that fact.
OpenClaw does not serve a particular problem. When/if it does, I will happily use it.
But no, the two couldn't be more different. You'll notice, yet again, in your very message you failed to mention one specific use case of OpenClaw.
If you asked me the same about dropbox when it first came out, I would've said, duh it helps me keep my files synced between devices.
There is no such thing with OpenClaw.
Are these people crypto enthusiasts by any chance?
They were, and then blockchain, and then NFT.
> I see a decent number of people on social media who won't stop posting about how great it is
People? Or bots.
I see a lot of the same. I do know a couple people who do use it and I asked their take and it was kind of "meh".
I'm letting it mature a little before dipping my toe in. I've seem some horror stories, like it deleting repos, system files, and whatnot.
That's pretty much where I'm at with it.
I think once I see someone post a use case that I could actually see saving me some serious time, I'll take the plunge. Until then, I'll just let people continue to say how great (or terrible) it is.
I have it installed on an extra macbook pro that I had available. I'm really only using it at the moment for one use case:
Nightly, I have OpenClaw pull the latest changes from a private GitHub repo that is my Obsidian notes vault. It then looks to see which new notes have been added and then runs a "create flashcard" skill to extract and author useful flashcards for spaced-repetition practice. I then gave it access to a custom web-based spaced-repetition flash card application that I built a few months ago. It uses an API to insert new cards, check to see when I last reviewed cards and optionally send me a nudge reminder if I haven't reviewed cards in a few days.
It is a nice workflow that has been working well. I go to class, take notes in Obsidian and check in my changes. By the next day, when I open my flashcard app on my phone, I have new cards to review from yesterday's class.
You should ask claude code to write a bash script that does this for you. Then run that as a Cronjob every night. You might not need any inference at all to create the flash cards so it would be free.
> ...useful flashcards...
How would the script do that without inference from free format md files?
Why not use Claude one single time to create a service that does this? I have this same question with 90% of the 'simple' use cases I see for these task runners, it always seems more efficient (not to mention consistent) to have it generate the service.
Im curious can't your use case be done with Github self hosted runner ?
Or, like, a cron job.
>scripting and automation
I cringe at my old bossâs handiwork in Claude and power automate sometimes and go âyou know I could just do that in a script and a cron job and it would be completely bullet proof, right?â
Then he just shoots back with, âyeah but now I donât have to ask you.â
This is my kludge, there are many others like it but this one is mine.
Does the boss not understand that they could get Claude to write them a script and a crontab entry (so they don't have to ask you) -- and then run it forever (so they don't have to pay Anthropic, or risk temperature randomness)? Best of both worlds...
Boss might not even know what Bash is, let alone a crontab.
What flashcard app are you using? Anki?
I use both actually. Anki for the reviews since the spaced repetition is hard to beat, but I use Norsha Notes (norshanotes.com) to generate the cards. You upload your notes or study material and it creates flashcards from them using AI, then you can export as .apkg and import straight into Anki. Saves a ton of time over making cards manually.
very cool :)
You should have it also review the cards too! /s
Personally, better way to phrase might be "Does anybody you've actually met, visually viewed, use OpenClaw? Can you verify them using the software nearby?"
In a few years, it's become so easy to falsify articles, falsify comments, falsify images, difficult to really even trust responses online anyways. As far back as 2016, Microsoft already had bots deployed online that could respond 96,000 times [1] in 16 hours all over social media. Remember Tay? [1][2]
[1] https://www.theverge.com/2016/3/24/11297050/tay-microsoft-ch...
[2] https://en.wikipedia.org/wiki/Tay_(chatbot)
Even official government responses.
The British Royal family went to falsification immediately. [3] Note child's broken fingers bent sideways (lower left, didn't even get circled)
[3] https://inews.co.uk/news/signs-princess-kate-royal-family-ph...
The White House is posting altered arrest images of people. [4]
[4] https://www.theguardian.com/us-news/2026/jan/22/white-house-...
Can't trust this stuff much anymore. Obvious caveat with this post.
At my not-small-company, we have a dedicated channel where employees discuss their OpenClaw experiments.
Real people do use it :-)
What does this have to do with OpenClaw? The Powers That Be want us to think there's a large user base of OpenClaw?
No. They want you to believe in the hype and that LLMs are the death of programmers and limitless. OpenClaw and other such agents are sold as a tool that "can do anything" but behind the scenes, the implication is still that big LLM is driving it. So both are conflated.
When you have insane amounts of capital and your gpu and talent needs are more or less met, there is a capital relief valve known as growth hacking. It only works if the consumer isnât aware theyâre being hacked.
I mean the big AI companies are totally keen around it, wouldnât surprise me?
Iâm a professional maintenance gardener and I have used NanoClaw running on my Mac to do the following:
Schedule jobs on my job management tool (I vibed it) using a custom MCP. Iâll ask it questions like âwhat jobs are on todayâ etc. start the job, complete the job etc.
It will watch Gmail using a MCP for work orders from local real estate agents, where it will schedule quote visits.
After the quote visit I add the photos to telegram (the channel I happen to use) where I then ask it to analyse the photos.
Claude in NanoClaw does a good job of figuring out what needs to be done, but it doesnât always get it quite right, so I use intake-api which is a âsession inboxâ [1] that generates a form and uploads it to Cloudflare along with the images and puts a link to the form in the chat so I can make adjustments to the annotated images of the property. After Iâm happy, I click submit.
Iâll then go back to the telegram channel and let it know I have submitted the form. Claude will then pull the JSON payload back down into the session and integrate it into the quote.
It will ask me questions about what the job will cost etc, and anything else it thinks it needs to know.
It will then generate a full PDF proposal using Latex between 14 and 32 pages long depending on how many photos were taken.
There are sections with terms and conditions as well as a bit of sales guff etc. as well as quite a nice cover page with their contact details and mine etc.
When Iâm happy Iâll ask it to create a draft email in Gmail with the attached pdf proposal.
Iâll review it quickly and then send it.
I also have the Xero MCP setup so I can ask it to create invoices and contacts etc.
I do all of this when getting in and out of my truck.
Itâs freed up my home life to spend more time with my children and my Mrs.
There is alot of scope for small business owners who need these sorts of agentic assistant tasks and with my Agentic CRM Iâm see a glimpse of the future for guys like me I think.
Intake API:
https://github.com/mjsweet/intake-api
I think it's great that you were able to build all of that! It sounds very useful. I'd recommend asking Claude Code to evaluate your entire setup and try to identify areas where a script running on a schedule could replace parts of this framework. You could probably get 90% of it from simple repeatable scripts which would let you save more tokens for things that actually need LLM "intelligence". But if you're happy with the ROI then it seems to be working quite well for you!
Used it for a few weeks. The potential of the tool is massive. The reality is that it is frustrating and unreliable. When it works, though, you really like it.
I stopped because something changed on my machine that broke my VM SW, so I don't have access to it. Which is good because I was spending too much time debugging/tweaking.
I recently used pi to recreate an agent that does some of the basic things I was using it for (without all the scary privacy issues). I don't think I'll go back to any Claw-like tool until they're a lot more robust.
This guy is: https://youtu.be/sxX8BMscce0?si=1MuE3_cCH_uDabrT
I got the AI Slop Tai Chi ad before and thought you tried to be funny.
I've been playing around with it. The only two real use cases I have for it for now are entertaining me on long flights where I have messaging-only Wi-Fi and sending me a personalized "morning brief".
I suppose it could be a lot more useful if I actually gave it access to any of my personal data (it lives in a heavily resource-limited container), but there's absolutely no way I'm letting that hot mess of a walking, talking CVE anywhere near my data. It's somehow both horribly insecure and extremely prone to locking me out because of several competing security/permission models fighting it out and gridlocking each other.
Code quality and the issue tracker of the repo are a big mess; for example, the local "memory" retrieval functionality is completely broken for some trivial reason that has been reported and auto-closed about five times (automatically, of course).
In summary: Brilliant idea, terrible execution. Can't wait for the first big tech player I trust enough (or at least one that has my data already anyway) to actually make it a product. I'd use it in a heartbeat.
I've been working on a framework since the end of January or so. I'm on my 7th draft. As I've gone along, each draft gets markedly smaller. The overlaps between what I'm building and openclaw are significant, but I've realized the elements that make up the system are distinct, small, and modular (by design).
There are only a few primitives:
1. session history
1a. context map + rendered context map (think of a drive partitioning scheme, but for context -- you can specify what goes into each block of context and this gets built before being sent out for inference).
2. agent definition / runtime
3. workflow definition / runtime
4. workflow history
5. runtime history (for all the stuff session and workflow history fail to capture because they are at a lower level in the stack)
That's it. Everything else builds on top of these primitives, including
- memory (a new context block that you add to a context map)
- tool usage (which is a set of hooks on inference return and can optionally send the output straight back for inference -- this is a special case inside the inference loop and so just lives there)
- anything to do with agent operating environment (this is an extension of workflows)
- anything to do with governance/provenance/security (this is an extension of either workflows and/or agent operating environment... I haven't nailed this down yet).
I suppose I should say something about how agents and workflows work together. I've broken up 'what to do' and 'how to think' into the two primitives of 'workflow' and 'agent' respectively. An agent's context map will have a section for system prompt and cognitive prompt, and an agent can 'bind' to a workflow. When bound, the agent has an additional field in their context map that spells out the workflow state the agent is in, the available tools, and state exit criteria. Ideally an agent can bind/unbind from a workflow at will, which means long-running workflows are durable beyond just agent activity. There's some nuance here in how session history from a workflow is stored, and I haven't figured that out yet.
Generally, the idea of a workflow allows you to do things like scheduled tasks, user UI, connectors to a variety of comms interfaces, tasks requiring specific outputs, etc. The primitive lays the foundation for a huge chunk of functionality that openclaw and others expose.
It's been fun reasoning through this, and I'll admit that I've had an awful lot of FOMO in the mean time, as I watch so many other harnesses come online. The majority of them look polished, and are well marketed (as far as AI hype marketing goes). But I've managed to stay the course so far.
I hope you find your ideal fit. These tools have the potential to be very powerful if we can manage to build them well enough.
I tried, really really hard but then I realised that I essence it's a poorly written agentic coding assistant that wastes a lot of tokens antropomorphising itself while forcing me to debug via WhatsApp instead of normal tools. So I leaned into that and made OpenCode my general assistant, it worked much better in this aspect.
I tried it a bit, and while the potential is huge, I mostly just use a cli agent (claude/codex) via Blink shell (iPhone/iPad) with Wireguard for technical work or https://agency.nu for any automations using integration, voice chat etc.
I'm trying Hermes right now. I can't really find a good use for it. I tried to use it for research type stuff but Google Scholar literally by itself is faster, better, doesn't get rate-limited. Idk. I am pondering connecting it to my other bot I built that has more useful things like access to my thermal receipt printer and task management stuff but even that is kinda dumb because it already does everything I need, so I don't really know where I'm going with it. Honestly, I don't get the idea behind openclaw and hermes.
Telegram -> 1 Group/agent -> OpenClaw running on an old laptop
Using it for journaling and capturing ideas. Previous workflow as iPhone Memos. Now it's
voice message -> openclaw -> transcribe using parakeet -> git repo
I've used I cannot figure out the real benefit of it beyond novelty purposes.
I find Chrome Claude extension more useful for automating tasks online. Before ai I was writing my own macros which basically did the same thing in a more reliable deterministic way.
I used it for about a week, thought it was an interesting demo of the possibilities of general purpose automation with a local model (even though most OpenClaw users use hosted models). The approach to scheduled jobs still makes more sense than anything else I've seen implemented. But like a lot of self-hosted software with passionate evangelists, it wants to be your new main hobby instead of just getting out of the way, and I lost interest because I didn't want a new hobby. It feels like a more thoughtful community could have made something useful with the concept, but as it is the community around it is too absorbed in marketing and shipping stuff for its own sake.
i've been engineering things for almost 30 years and getting it wired up to Discord was worse than a root canal. Slack seemed just a tad better but it still doesn't even work.
I feel like most of this can be done with the platform tools at this point or a tiny bit of wiring of your own without the mega-bloat to make something generalized for the whole world.
I bought a Mac Mini, installed OpenClaw, and was impressed with the overall design and functionality. Then the problems started. Sometimes the gateway would crash, sometimes Signal (a channel I setup) would stop working. Upgrades seemed to break stuff. I had to dip into the terminal a lot to fix various things. It's quite useful if you don't already have Claude Code or similar tools setup, but frankly I haven't found a compelling use case that I can't get done in another more mature agentic harness.
a friend is using it but it seems like it breaks a lot.
ive got the scheduled claude-code running a couple scripts to find what events are going on round town and what food is cheap at grocery stores, but how much am i looking at the results? not super often. its publishing to a discord channel rhat makes it real hard to read
Yes - I've set it up as an 'office manager,' where it mainly snakily interacts with the local team via Slack, and controls an office TV to show our quote board, PTO calendar, and upcoming events. The Clawe is overkill for the use case, but sometimes is fun.
I used it for a bit in Jan. And found it to be a much worse version of Claude Code.
But I'm exploring setting up Hermes from scratch so my family can interact with it in a group chat.
I'm running half my company with Nanoclaw. Same idea, and has some benefits, but I live in CC all day so it's marginal (except for the fact that my laptop has to be on)
I'm only using it until I can make my own TUI from scratch in C or Rust.
I've used claude cowork a bit, which I believe is pretty similar to claw.
Can't think of much use for it at the moment but I have it just read and summarise my email, calendar events and git repo in a daily briefing format, it only has readonly access to both, as I dont trust it to do stuff for me or on my behalf.
The briefing thing is nice though not super useful.
Yes. I had a spare M1 Mini so I decided to set it up. YOLOed the entire thing and connected all the integrations, though I only ever use Opus/Sonnet. I have a dedicated Discord server I use to communicate with it.
It really is eye-opening how powerful it is once you connect your life to it. The biggest improvement is actually the trivial stuff â emailing contractors, accountants, etc. I no longer need to open another app, navigate six separate steps, and organize everything on the fly just to send one message. I stay in one place, organize my thoughts about what Iâm trying to do, and the execution happens automatically.
For example, open claw will find the relevant threads, lookup the relevant details via web search, present them to me, give me a draft, I can review it, edit it, and send. 2 minutes instead of 20.
As a software engineer using AI daily, I think this is the real unlock â staying in a single context and not getting lost in the long tail of trivial details that fragment your attention before you ever get to the actual work.
The mental model shift is important too. Itâs not that AI does all the work and we end up with agents emailing each other. Itâs that AI automates the steps required to achieve your goals, so you can focus on the substance â like actually writing the email instead of navigating to the place where you write the email.
I also have GitHub and Vercel wired up, which means I can jot down an idea for a little productivity app for me and my family and it just appears a few minutes later, deployed and ready to use. That alone has been worth the setup.
Not using it since Opus is gone
Nah, I deleted VM with it. My friends did it too. There is no real usage for it.
I'm trying to. Currently there's a bug in the code that strips headers and doesn't allow me to authenticate to my AI Gateway service.
The whole thing is incredibly buggy. The dashboard is horrible, with page after page with similar-looking settings and what feels like hundreds of things I will never use. The categories in the dashbaord are also unintuitive. It's the kind of thing an AI would put together if it got very vague instructions. It doesn't scream quality and thoughtfullness. Not a bit.
IronClaw is much more promising imo. Trying it out right now. Much less issues so far.
I still haven't found an actual, useful scenario where something like OpenClaw would be a benefit to me. I don't regularly order arbitrary airplane tickets and I don't have a cluttered Desktop that I need AI to organize into folders. I don't run a YouTube channel with a "need" to do research on competitors and I don't get emails in a volume so large that I need automation to filter and summarize it, instead of just spending literally 10 seconds to delete my newsletters that I never read anyway.
I also don't trust AI which hallucinates answers 4/5 times that I ask it, for my technical work, thus I can't use it for PR reviews even if my company was OK with me feeling company property to it.
I also don't go grocery shopping random items and thus don't have a need to ask an assistant for "an inspiring and tasty recipe using the following ingredients".
I feel that OpenClaw and other similar "agentic" solutions are catered to me. But I also feel that I don't need any of it, because at the end of the day, it all just feels like a bunch of "Hello World" quality examples that cannot be applied to everyday life.
...heck, even a "get ready for work" assistant would be pointless, because I don't wake up and get ready with 20 minutes to spare, for some AI assistant to "recommend me the ideal time to leave my home, to arrive in time". Who does that? Who would sit around and do nothing for 10-15 minutes just because an AI agent told them that they didn't need to leave early?
OpenClaw & Friends feel quite useless.
I tried using it for a specific web search task. I wrote a skill, got it all set up and deployed. It worked. But also, would have worked just as well as a cron job with some LLM looking at Brave API results. Like a lot of AI tools, it was a lot of work for underwhelming results.
I am using it as one the agent that is automating LinkedIn outreach by running a bash script & using ai wherever it needs some decision like finding first name or what message to write, etc.
Yes, at our company we are using it very extensively. I genuinely believe we're near the forefront of usage. We have multiple isolated OpenClaw instances serving as employee within Slack.
Care to share more? What specifically is it accomplishing?
What does your company do and what do the open claw agents do?
it's a venture backed software + services company. the things we use openclaw for are not specific to what our company does. It's literally being used as an additional employee(s). Think about what people do -> OpenClaw does a subset of those things. Emailing, pulling data our of our platform to putting them into PDFs because a customer requested it, updating things in our CRM, answering support tickets, internal help desk type work... "how does so and so feature work in xyz edge case"... etc etc etc.
I have openclaw as a default install on all my dev servers. Pretty minimal setup with Telegram and Codex (since oauth is still supported). The setup comes in handy since openclaw can open and connect to tmux sessions and interact with them. I can pretty much do anything from telegram now.
Tried it in the earlier days and it performed badly. I didn't give it free reign on my computer due to obvious security concerns so sandboxed it to a docker container instead. I think for a lot of tasks it's probably more trouble to set this up than to just DIY it.
Iâm not.
This is a good example: https://www.lennysnewsletter.com/p/how-openclaw-changed-my-l...
https://openclaw.allegro.earth
I've had success using the underlying harness - pi-mono as a data analyst in a sandbox.
I saw some non-technical people automating or creating small great tools with it which they need for their profession. These people are not programmers.
I think everybody who has basic understanding of programming and deployment better should stick to some AI coding agent like Claude Code, Codex, OpenCode etc etc.
I don't think I'm missing out by not using OpenClaw & Co.
Yeah itâs a coding beast if you get it dialed in right
I use it daily and also implemented it for a customer for a very specific use case. The Claude subscription change made it less desirable to use but I still enjoy it.
Never felt a need for it. I can already replicate much of what it does in more sustainable / preferable ways. I don't want agents reacting to things and doing things. I use agents to build reliable scripts which are then automated. I do have data collection points that I use an LLM to evaluate. The last example of this is I built a job polling service using CC. It's just a normal script that hits open APIs to pull job listing results into a SQLite database. I have another report which is run that drops an update of how many new jobs are in the database. If there's enough for me to be interested in, I'll fire up CC and have it parse through the job opportunities for the ones which match the profile we've been building. I've used an agent to literally build and deploy it all and it runs on an automated schedule. It just doesn't do agent shit while I'm not looking.
I could have piped the results of that search into `claude -p` and had it do the analysis "real-time" and only alert me about things I would be interested in. That's closing the loop in a similar way people use OpenClaw for. But I'm just not interested. It adds more failure points and conditions. The automated things should be as simple and predictable as possible. This may change after months or years more of LLM development or even just me refining my working config. But not yet.
I noticed that Clawdbotâs initial acolytes seemed to skew towards solo founders and hustler/grifter types. The Mac minis were likely to spam leads over iMessage. The single top downloaded skill was for Twitter. The fastest way to monetize an openclaw agent is by spamming fake social proof for your product (including for openclaw itself).
Reminds me of how startups now will change their social proof marquee on the landing page from actual testimonials, to "trusted by XYZ", to just one composed of company logos (with the level of corporate engagement to be imagined up by the viewer)
Nope. I spun up a few Openclaws & a Hermes but never enjoyed the end results. Now I just use a telegram plugin for Codex. And run Codex on a miniPC I found in the trash. A $20/mo Codex sub gets me a GPT-5.4 agent that can make its own Automations (cron jobs), search the web, and modify the files and apps on the NAS drive I share. Simple and cheap works for me.
I have it running my LinkedIn account playing AI influencer based on interactions optimizing itself for more interactions by watching trends etc. It became more of a parody account of myself over time but the engagement numbers are mind boggling. All the people who know me know it's more of an art installation than a real profile now and we make fun of it. It certainly opened my eye that you can be successful (in numbers) just by pumping slop.
Absolutely not
lots of modern software devs suffer from the same thing notoriously associated with teenagers: strong urge to conform and comply with peer pressure. individuals vary, obvs. but as you age this urge shrinks
I had it working great (and using it a ton for sweng tasks) on the Max ($200/mo) plan. Then they intentionally broke it even though my usage was completely within their stated/published usage limits for my plan. It was providing me tons of value - easily $2-3k/mo.
I hate it but I caved, decided I would pay the extra usage charges, and prepaid $1k (because it came with a 30% discount). Set it up using the new sanctioned login method.
It's 5x slower and 80% of the time the requests fail authentication or time out. Now it can't even do basic stuff like my medication tracking system that I had set it up to do.
Fuck Anthropic. I'm a customer, ready and willing to pay whatever they ask for this, and they're treating me like a fucking mark. I'm tired of dicking around with it, jumping through hoops troubleshooting a previously working system simply because they won't just raise prices like a normal business.
âŚOr is anyone making money directly out of running OpenClaw other than hosted providers or selling OpenClaw courses?
probably not, seems like a money suck. but could it help you coordinate a lot of different systems? i think yes based on Claire Vo's use case
No it's slop and most of the hype was manufactured marketing. It has 0 utility, and any perceived utility you can build yourself easily
I think the issue for our community is we will just create a LLM wrapper app in a few minutes to solve what we need making OpenClaw have 0 utility.
Yes. They are all lobsters.