Outside of the article's mentioned graphics development, there is no reason to isolate an agent using actual hardware. I threw together this script[0] using libvirt to give claude its own graphical desktop env to be able to do user acceptance testing with Chrome. It has full root and can do what ever. If it makes a mess, I can dump and reinstall in seconds.
Nice to give Claude their own blue bubble iMessage. Can be done in a VM of course if you have the capacity, but old Apple hardware seems to accumulate for many.
There are many ways and tools on linux to get a "VM" up and running. But with libvirt you can easy script out what the initial environment and network stack is. The heart of it is just running the 'virt-install' command, but as you can see I've got a bunch of other opinionated stuff going on before that.
You mean you had claude throw together a script because this is heavily generated. Lost interest after about 250 lines, it's just tiresome reading through this incoherent amalgamation
Many Claude Code power users don’t really use IDEs anymore, so the only purpose of them working from their laptop instead of a phone is because that is the normal way to do it.
Here is a real use case: you are are responsible for some alerting channel. You have datadog/ cloud logging/ github all connected. You see a bunch of alerts come through while you are out and about and you prompt CC to investigate - Claude triages and says “all of the sudden you are getting time outs from this bank API your company partners with, this started an hour ago. It’s happening on ~15% of requests”. So you ping the guy at your company who does vendor relationships and go back to your weekend.
This is a non hypothetical example. Obviously it would be better if your job had a real on call rotation and more robust alerting and you wouldn’t be getting slack alerts on the weekend… but I take the approach this job affords me a lot of nice flexibility so it’s ok
You don't have an on-call rotation but do have people dedicated to vendor relationships, and that guy works on the weekend? I'm not sure how you completely avoid getting alerts on the weekend for third-party payment processor issues, which can happen anytime, if you actually want to transact business on the weekend.
I don't know what exactly GP works on, but 'bank' covers a lot of ground and some things happen on weekends. 'Open Banking' APIs, and some payment rails for example.
> Many Claude Code power users don’t really use IDEs anymore
This is something we of the HN bubble take for granted. Most of us know how to type quickly and use editors and use macros and program scripting languages and compose regexes.
The vast majority of programmers do not know those things. As such, AI speeds them up tremendously.
Yeah good luck being employed in 3 years once this bubble popped when all you do is type some natural language into a phone screen. People being proud of not using an IDE anymore is such a foreign concept to me, who enjoys coding and got into the profession because of the love of that.
The “debugging” for these type of issues is looking at some logs and http responses and being like “ah if we get this error it means they restarted their firewall again and took us off the whitelist. Email that guy Joe at the bank and hope he responds”. It’s not rocket science or the majority of my job… but someone’s needs to do it. We automate all the stuff we can.
I run a lot of data science-type analyses that can take up to hours at a time to run, so Claude is « monitoring » tasks most of the time. I have it on remote-control so I get notified when a task is done or need clarification, but most importantly whenever I have a new idea, I can just ask Claude to queue it up. Most of the time my hardware is the bottleneck, not the subscription quotas.
There are so many neat use cases like this one that I'm skeptical of the AI-is-a-dot-com-bubble naysayers. Not to mention that, unlike in 2000, technical-adjacent people can now get a reasonable approximation of their ideas running quickly.
I used to have some hooks for local notification, but lately I find that claude is pretty good at notifying through the app with remote control (but definitely not perfect)
> take up to hours at a time to run, so Claude is « monitoring » tasks most of the tim
How is Claude monitoring them for hours? Claude runs out of context and extremely long sessions are prohibitively expensive even according to Anthropic (after they dispense with the marketing bullshit of long running tasks)?
It launches these tasks in the background. It became really good at it a couple of months ago, now it sets monitors on timer (not something I instructed, so I assumed it’s part of the system prompt for this kind of tasks) and then just wait for the next prompt, for the background process to be done, or for a monitor to trigger a checkup.
A single session running for multiple hours is prohibitively expensive, as per Anthropic. Regardless of whether it just waits for a prompt or does something.
Is the high expense coming from cache misses? If their workload does need to wait for a long time before it can continue, I wonder would starting new sessions and having to re-read the contexts and results anyway be any cheaper.
Prohibitively expensive is all relative. Pre-Fable, I was getting fine on the 5x plan for 1-2 concurrent long-running tasks plus interactive work (I do a lot of coding for work, but it is not my full work day). I don’t think a cache miss every hour on Opus hurts that much, even at 500-600k context. It would be nice if they got /clear working on remote-control.
I've used it for the following when I've had tokens to burn:
- Fuzzing with the goal for it to apply domain-specific and source-informed knowledge to choose specific fuzzing approaches.
- More generally, any optimization problem that benefits from domain-specific or source informed knowledge.
- Running Microsoft's SkillOpt [0].
The other day I used Claude Cowork to create a checklist for me so that I can safely re-install my laptop... It went through all ssh configuration, apps, downloads, documents etc... Once I had the checklist, I asked it to do the backup itself, along with instructions to re-install/setup everything once I had the laptop re-installed. This was something I was procrastinating for too long already...
1/ Using GUI software. My agents are using headful Google Chrome and Figma. It helps a lot to have separate environment, which is not interfering my main machine.
2/ Running long processes (1h+), so I can leave main machine closed.
3/ Running intensive processes. I use Gemma, Whisper and Qwen, which could burn main machine CPU and resources.
Running Claude code 24/7 on a code base on that “second Mac” so you can always continue after a usage limit reset, from your main device or from your phone?
I changed all smart speakers to retrofitted old radios with an amp and a pi. The hot word detection runs on the pi itself but whisper and LLM/task orchestrations goes over my local server with a 4080.
It's less about 24/7. It's more about it can't work when your laptop is in your bag and in transit and there is something that you have set up and want to run.
I think the main use case for AI bros is to setup a goofy looking dashboard, name it Jarvis to cosplay being Tony stark, and display stats for all the generated videos they're posting to social media.
Running tests that the agent created for its own hallucinations? Optimize using another hallucinating agent?
Which tests and optimizations do you propose to run after a night of supervised work when one of main things that all agents keep doing is "load all records from db , and filter them in memory"? It's now become so bad, I had to literally vibecode a separate linter for this. And that's just one of the problems.
we don't have master AI that can create a great product out of a mediocre prompt.
but we do have sufficient AI to make a great product out of a great prompt.
garbage in -> garbage out hasn't gone anywhere.
so: much like to anyone that blindly complains that their compiler hates them : if you actually want help, provide information. If you just want to complain that the compiler is mean, scream at the sky.
plenty of people have figured out how to get this to work; more than enough to confirm that a straight <gambling-machine>/<hallucinatory-psychopath>/<random-number-generator> analogy is too simplistic to explain what we're working with.
Demagoguery. Your "great product" (where is it? show me these great products) will have wildly ineficient code and things like "read the entire DB into memory" regardless of the prompt.
> plenty of people have figured out how to get this to work
Plenty of people claim they have figured it out. In reality these people are full of shit and assume that if LLMs can produce working software, it's great working software. And also assume that LoC is a measure of quality.
>It's now become so bad, I had to literally vibecode a separate linter for this.
You see, there's your problem right there. You're vibe coding, which by definition literally means you're unwilling to look at the generated code. That's not what successful ai assisted software developers are doing. YOU HAVE TO READ THE CODE. Refusing to do that means you're not a serious programmer, you're outsourcing your thought and design and implementation, trying to get something for nothing by taking the easy way out, and you're going to get terrible results no matter what prompts you "engineer". There ain't no such thing as a free lunch (yet).
And while we're at it, to elaborate on what serf said: people mindlessly parroting terms like "stochastic parrot" to criticize llms without having read the actual paper that coined the term and understanding what it really claimed and how other papers responded to it means you're just a human stochastic parrot no better than what you're criticizing -- at least the llm has read all those papers and understands what "stochastic parrot" actually means in context. Ask it, it will be glad to explain!
Emily M. Bender, Timnit Gebru, Angelina McMillan-Major, and Margaret Mitchell.
On the Dangers of Stochastic Parrots: Can Language Models Be Too Big? (FAccT 2021)
> You see, there's your problem right there. You're vibe coding, which by definition literally means you're unwilling to look at the generated code. That's not what successful ai assisted software developers are doing. YOU HAVE TO READ THE CODE.
I guess you vibe-read what I wrote. Let me write it again for you: "I always have to correct its hallucinations during the day. Why would I ever let it run unsupervised overnight?"
If this is your problem, I hate to say it, but it’s a skill issue on your side. I didnt even start developing software until LLMs even though I know all the primitives quite well and have a strong grasp of architecture. ADHD, perfection, and focus got in the way. I’m writing a pretty complex application and I don’t have the problems you’re running into. I regularly have my code reviewed by professional devs to ensure I’m not just vibecoding into slop. I’m not. I also read my code regularly and do a lot of exercises and courses to keep learning as well.
So I dunno what to say, except it’s possible to write really solid code with LLMs.
Generating leads for new work, if you are a freelance. Automatically answering customer support emails, if you own a SaaS. Monitoring competitors' socials, websites, etc for new features you have to compete with. Monitoring updates on software you depend on for breaking change / deprecation announcements.
How? I mean what could be the ultimate usefulness of Claude if not to make money, just like NFT but with extra steps. Most people isn't using Claude to make massive paradigm shift breakthrough discoveries. Nobody's curing cancer nor solving climate change. Probably the most common use case for llms it's just speed up the grind and make money. Like nfts.
If curing cancer or solving climate change is your definition of usefulness, almost anything humans do is useless, probably including your own profession. Though I don't want to speculate. Speculation is precisely why NFTs are stupid. Claude, on the other hand, if used effectively, can speed up people's work and increase productivity.
This implies a long road to catch up. But seems to be that all one needs to do to be on the cutting edge is to integrate a few scripts that support a cutting edge model. Which you then converse with in natural language. So how higher will the bar to entry be five years from now?
It's a no win trap, except for the tech bros cashing in on investors burning money. Data centers using up the available clean drinking water, people with no jobs paying higher electrical bills, stressing the electrical power grid to the breaking point, historically unparalleled IP theft to divert money from creators to their tokens...
Yeah these guys will set up claude on their third pc after they realize their second pc keeps hallucinating out code nobody cares about or needs in hopes of staying relevant. I'll just keep coding by hand because I ACTUALLY ENJOY it. What a novel concept these days.
I've done this before as a free replacement for my OpenClaw bot. It loses its connection sometimes and you need to redo the handoff with your phone. But otherwise, nice to have a use for my old Mac. It's also running my Home Bridge, which Claude can interact with. Pretty cool.
I currently have Claude Desktop installed on a separate Mac mini M4 and control it with Dispatch. Is there a reason to do this method, it still seems the way I have it setup it has full control over the local account I gave it on the Mac mini.
The gpu inference happens on anthropics side, but the output of the gpu inference is often many local tool calls. The agent is locally reading files, writing files, compiling programs, downloading artifacts, running tests, invoking CLIs, monitoring logs, inspecting web sites, grepping etc, and can be doing all of that at once, in parallel via subagents, much faster than a single human ever could.
I found Dispatch to be less capable than RC. The latter is more like the direct Code experience (though there are some gaps in RC mode vs in a Claude-owned sandbox).
I’d love to just have Claude use my machine as a sandbox host instead of having to run RC on each host session. (In case you are listening Boris ;) ).
In the meantime I have a janky master RC session that creates new tmux windows and Claude RC sessions for each new code trajectory that I want to run.
The other benefit here is you can drop down and use termux to use Code directly if you hit a RC bug, I found permissions UX to be a bit flaky in the iOS UI.
Dispatch/Cowork is basically claude code in a container. The section "Why not run it in a container?" would address your question in the post. One example I run into is that Cowork won't download and fill out or read pdfs or other files due to container permissions. Vanilla claude code has no problem using curl and wget.
Cowork gets tangled with git as well. Fails, and then can't delete lock files.
Running a helper from the terminal, making Claude work in a working directory, and then create a .commit file has been my workaround for this for a while now.
Imagine there's a better solution nowadays, but this allows me to use dispatch building on Vercel, so I can check it out from wherever, without too much pain.
Claude dispatch does things differently than this. Dispatch is very convenient for somethings (including connectors).
But increasingly it seems like dispatch was slapped on top of cowork incrementally, when there was not an integrated and cohesive strategy across cowork between mobile/desktop/laptop. This is kind of what many of us get/got to learn in our 20's.
Though it doesn't get by all the hurdles mentioned, it is alternatively possible to run Mac OS in a VM on your Mac using UTM and install Claude Code within the UTM VM. UTM can be run under a non-admin Mac account. This can allow you to use most Mac-native tooling, at least. The interactive performance of using the Claude Code ui on the VM isn't great, however. I'm not sure if you can log into the VM via terminal on the host from the non-root admin account to avoid the ui performance issues.
An argument against this method is UTM doesn't support graphics acceleration so browser support will be hobbled. Even if you don't need the acceleration, I've found browsers in UTM virtualized OS's can't get past some modern captchas and other browser fingerprinting checkpoints :\. It's terrible but that's the way it is.
I don't know about solving captchas, but if you are running macOS virtual machines on an Apple Silicon Mac, hardware graphics acceleration is supposed to be natively supported out of the box when using UTM's Apple Virtualization Framework backend.
A primary source of UI lag is how Apple's native virtualization framework processes multitouch trackpad events. There are other issues like mismatched resolutions and framerates, too. Ask your favorite ai to debug. You can try e.g. deselection of the trackpad setting and instead select basic mouse support, and that can clear things up..
I just redid my homelab/media server (switched from an old NUC I couldn’t figure out how to stop overheating even with some decent modification work, now using an hp elitedesk with an i5 processor that is handling my stack nicely). Thinking about setting up a vm on the base ubuntu install for isolation to run Claude in. May play with dispatch, may just put raw Claude code then use Moshi app on my iPhone and iPad.
I’ve been doing something similar with an old m2. It isn’t powerful enough for local models, well sufficient local models but for openclaw and Claude it’s been perfect.
MacBook m1/m2 also are cheap enough now vs an Mac mini which I was surprised about, not too surprised but yeah..
If I had one I would definitely try creating a separate VLAN for it to control, otherwise it's isolated from your files but still has access to your network and devices in it.
For me, it's mostly because I want whatever I'm running to stay on my own home network, with access to Home Assistant, my homelab, and so on. Beyond that, old laptops are hard to beat for this: I already have a few, plus spares, they sip power, the battery is a built-in UPS, and the screen and keyboard mean I have a console when something goes wrong — no hunting for a monitor. There are plenty of use cases for a cheap cloud VM or VPS too, of course.
In my case, most of my VMs live on a proxmox cluster of HP EliteDesk minis, with the laptop sitting outside that as the thing that manages the cluster and a few other odds and ends.
This can work, but increasingly there is benefit from local, including places that want data to be secure and not in the cloud. This is a thing. It's normal. It's just new to people who haven't realized how common it is.
I agree, and it's not necessary for agents to run as root and full access to data.
There has always been a trade off sold to consumers of security vs convenience and a belief that giving up a little security gets a lot of convenience.
It ends up being often about a convenience of adopting the new tech, not necessarily in a way that's in the best long term interests of each party.
Giving sudo permissions to an agent seems reckless. Claude gets his own unprivileged UNIX account, no more. I don't bother with containers or VMs however.
This is the paranoia talking, but given that Claude is going to be doing its own thing, I would put this box in its own VLAN or behind deny-all firewall rules to protect against network escapes.
It's so strange to me that tech crowd uses macs for use cases that spend all their time idle or serve some light http request.
It's a glorified API gateway...it could run on a medium sized potato
Almost like an entire generation just grew up coding on macbooks as the obvious choice for coders and just can't conceptualize hardware outside of that walled garden
Not OP but in my case I've been using this setup for quite a while (well before Claude remote control was a thing), this works with any other tool / ssh session.
I'm not entirely sure there's a big advantage to Claude remote control to be honest. Maybe it's just me being afraid of lock in (I do switch between Claude and Codex somewhat often) and/or the inertia of changing things up that keeps me from trying it.
I've been putting off learning Claude and this article had me strongly considering jumping in. Then I looked up Anthropic pricing and its 100x more convoluted than cloud services management. Its a goddamn full time job and independent skill set figuring out how to prevent going bankrupt from AI usage!
I think I'm gonna be a late adopter on this one until the industry figures out a less cumbersome pricing model.
Look at the subscriptions. That's how anyone using the stuff as an individual should be approaching it today if they are dealing directly with Anthropic or OpenAI.
With the $200 Claude subscription I was able to get around $13-15k of API equivalent usage in one month (note: this was during the "+50% usage" promotion that they have kept extending since May). When you hit your usage limit for a given time period you get cut off until the time period resets; don't bother paying for additional usage credits, you will be disappointed.
What I’m reading is this use case is for 3-5 days a week full time dev if you stick to off hours US time. That you can save significantly if you have spiky usage 1-2 days a week by going the ad hoc API route if you’re new and you need to install a bunch of monitoring tools to tell you how best to save week to eeek as your usage patterns change and you risk surprise bills in the 10s of thousands if you get it wrong.
The approach I took was to start with the $20 plan, then when it became clear it wasn't sufficient I upgraded to the $100 plan. The $20 plan didn't cover one night of coding. With the $100 plan I started bumping into the 5 hour usage limits after ramping up over a week or two.
My average API equivalent use is around $30-40/hr. I would just bite the bullet on a plan for one month, then use that to calibrate your expectations around usage and cost optimization. The plans are heavily subsidized.
I still don't understand what these freaks are doing running these agents 24/7 on machines. What are they doing? Managing a todo list? You mean crossing items off as you complete them? Research tasks? To do what?
Never really get good answers. There is no killer app. Just bikeshedding.
Swiping Tinder. It takes about 5000 matches to get a date. It’s easier to just automate it. It automatically adds dates to my calendar, all I have to do is show up. I get a summary of our chat history (well, what the agent wrote to her) in the notes section of the calendar entry and some pointers and talking points for the date.
Maybe I should have the agent also do a background check.
PS: This is a joke, but feel free to steal this idea.
The future is agents chatting to each other on tinder and automating the initial getting to know you part. I can imagine that while that's going on, we could add like a little text chat box for the humans to chit chat with each other a bit and pass the time, before they can go on a date
> TinderGPT automates the process of writing and arranging dates with girls on Tinder, enabling you to generate romantic meetings with almost zero effort. Your only role is to like the profiles that catch your eye. After that, TinderGPT comes into the play. It initiates a conversation with the girl, using details from her profile, continues by building an emotional bond and highlighting your attractive traits, and finishes by arranging a meeting and giving you a push-up on your phone with her number.
This is not really the same at all. You still have to swipe, the value add here is apparently “building an emotional bond”. Together with the explicit goal of “dates with girls”, this actually feels incredibly nefarious.
It seems the main use case is having Claude automatically write blogposts about how great using Claude is, then submit them wherever necessary.
There's lots of news about the billions AI companies spend on data center construction, but it feels like it's not even a fraction of the money they're spending on endless nonstop blogs about how great their app is at doing... things. Things that will never be defined.
It really feels to me like this OpenClaw type stuff is the new "I built a static site generator!" type blogs that just post a few articles about how they built their generator.
Exact same question as you. When the new ChatGPT app dropped it suggested to me to set up a task something like (paraphrased) “every Monday read my Gmail and Slack an make a summary and task list for the week”.
Why would I need an LLM to do this for me? That’s 5 minutes of work max, and doing it gets me in the flow of work again, to see what’s going on and needs to be done.
For a lot of folks summarizing a few days of work email and especially slack chats is way more than 5 minutes. Some work environments do not have great communication hygiene so it can be overwhelming to try to keep up with 500 emails a day and 38 Slack channels.
For the folks I talk to who use a LLM for this that seems to be the case. Takes a huge cognitive load off every morning and saves them an hour or two.
More or less a very expensive band aid over a bad work environment.
I kinda use it the same way in a sense. I have a little skill I run against our (horrible) task management system to summarize things and give me a punchlist to work through sorted by priority. This saves me thousands of clicks to do the same thing in the horrible web UI. A proper system in the first place would be a lot better!
At some point I’ll probably just take that to the next logical step and have the LLM write my own web interface to abstract and replace the horrible one entirely for me.
In such environments everyone is constantly missing things and not replying until they get followed up with.
So that sort of thing is already normalized. An LLM is unlikely to be worse than a human. There is no accountability either way since everyone is failing at the task, it really doesn’t matter much if your or your bot misses a few percentage points of high priority items. You will probably be vastly outperforming your peers who are doing it manually.
Because then OpenAI can read your emails and project communications and eventually build a model they will sell as an automated consultant. The CEOs will uncritically eat it up just long enough to cut the footing out from the industry. Once everyone is used to the sorry state of software, nobody will be able to imagine putting people to the task anymore and we'll have the new world order that Altman and Theil have been talking about creating.
> This workflow alone is immensely powerful, and takes alot of burden off the team.
ITSM those unsupervised workflows are essentially an attempt at purported productivity in the near term at the expense of meaningful incremental long term burden for teams.
The only ostensible benefit is in the eyes of the AI-psychotic tinkerer, who knows no better, or in those of the clout-chasing developer farming likes on their LinkedIn posts.
> The only ostensible benefit is in the eyes of the AI-psychotic tinkerer, who knows no better, or in those of the clout-chasing developer farming likes on their LinkedIn posts.
A company at the scale to benefit from this almost certainly has some kind of development sandbox environment and/or periodic job runner that's integrated into its environment and maintained by a team, not random Mac Minis.
None of these are things I want or need in the product I maintain with a team, there's really no point to any of this unless you run a vibe coded SaaS (?)
You want your team spending their time fixing these simple errors? The secret sauce is in the triage. We've adopted solutions alot like this, and now our team spends its time on much more meaningful work.
Why are the errors occurring, though? That's what boring analyse-and-fix addresses, through familiarity, recognition of patterns and "hang onnn..." moments.
It's like your AI agent is just plugging the leaks in the dyke each time, instead of fixing the architecture of the dam.
Yes, I want my team to be deeply familiar with the codebase and every single little bug that needs fixing both trains them and let's them learn a little bit more about the codebase.
They can use agents. Like, team members don't need to be replaced, they can simply use agents when they deem it useful. If they see a trivial bug,they can put their agent on it and go work on something else meanwhile.
Not really? Imagine for example looking for http status code 500 in an api log over the past hour. The nice thing here is it doesn’t matter if you get them all because it’s reoccur (or not).
Dude, really open your mind and stop arguing mindlessly. We're using this to great success on a serious sass product with hundreds of thousands of users. I'm not some AI influencer, or blowing smoke.
The source of errors can be whatever you like. Sentry, grep, whatever. Its not the point. The point is that many of these errors are real issues, and can be fixed automatically and safely by agentic systems. It really saves time, and work by our team leaving them to actually concentrate on the thing that delivers real value.
You started with "scanning logs for errors" with LLMs which is one of the most stupid things could do with LLMs on a "serious SASS product with hundreds of thousands of users" (aka logs easily in gigabytes to hundreds of gigabytes per minute).
But now it's suddenly not "we use LLMs to scan logs for errors", but "we use actual tools to find erots in logs, but then just hand them off to an LLM and hope for the best".
As in "let's repeat what @troupo said but pretend it somehow goes against what @troupo said" lol
Let me guess -- in your day job you don't manage people. I have agents parsing messages, building out document sets, evaluating existing document sets, one is currently fixing a giant backlog of bugs and feature requests for a multi year personal coding project, one is exploring some ideas on speeding up inference at the edge..
If you put yourself in a position where you need more leverage (technical or operating) I think you might find you get some value.
I don’t have time for much leisure coding these days. I do have time to kick off a few tasks in the morning to progress my many side projects. Nothing public / oss, just code that I find useful/interesting like home automation, content pipelines, games, etc.
There are a bunch of cases where remote control from iOS onto a Mac Mini is simply nicer than using iOS Claude Code sandboxes.
It’s the same pattern as you (hopefully) apply at $dayjob. If you are not defining a /goal and letting your agent crank you are not making full use of the models’ capabilities.
Well I am fully of the opinion that LLMs can help in software programming, it's not something that I feel provides any value unless it has a human in the loop. The overhead of having to figure out if the agent did a good job, if the agent is actually done or not, and if the thing it built is shit or not, is worth simply avoiding by having a human in the loop.
So I wouldn't agree that the agent should be cranking out code all the time, in fact that seems more like a waste of resources compared to the work it creates. But I do understand home automation software can be very one-off and simple. But then again, a properly programmed home automation suite doesn't need a SOTA model to modify it, I think.
What's a good way to give a limited amount of money to the LLM, say like 2k or 5k or something. But keep it completely separate from my identity.
Like I want the LLM to have a bank account and he can do ANYTHING with that bank account that he wants. But he can't fuck anything up that has to so with me. He only has 2 - 5k
Idk how you could at least in the US. Closest thing off the top of my head would be one of those checking accounts you can setup for kids. It would still be tied to you.
If you want Anthropic (or others) but anonymously, what you do is use https://openrouter.ai/pricing and fund your account from any of your preferred cryptos.
Outside of the article's mentioned graphics development, there is no reason to isolate an agent using actual hardware. I threw together this script[0] using libvirt to give claude its own graphical desktop env to be able to do user acceptance testing with Chrome. It has full root and can do what ever. If it makes a mess, I can dump and reinstall in seconds.
0: https://gist.github.com/smith153/04b4068b5a2d7b234f1c3d5992d...
Nice. I just have
So it can blow up its own files, but not mine.It was also doing some kind of headless Chrome stuff in there. I don't know how that works, but it was taking screenshots iirc.
I did also set up VNC at some point but didn't find it worth using.
>If it makes a mess, I can dump and reinstall in seconds.
This is also true of a $3 VPS, where I found it very amusing to give my agent root. What's the worst that could happen ;)
Patiënt zero, that’s what :-)
You can run macOS in a VM guest on a Mac host; Apple explicitly allows this.
Nice to give Claude their own blue bubble iMessage. Can be done in a VM of course if you have the capacity, but old Apple hardware seems to accumulate for many.
This is neat, and thank you for sharing!
I've been wanting to set up something exactly like this for my own use, but... You know, time is limited.
This is just enough scaffolding to have a little project for Monday morning!
I also have a light weight cross platform virtual machine with egress filtering if you're looking for an even more batteries included approach.
https://github.com/smol-machines/smolvm
There are many ways and tools on linux to get a "VM" up and running. But with libvirt you can easy script out what the initial environment and network stack is. The heart of it is just running the 'virt-install' command, but as you can see I've got a bunch of other opinionated stuff going on before that.
You mean you had claude throw together a script because this is heavily generated. Lost interest after about 250 lines, it's just tiresome reading through this incoherent amalgamation
I don't know what sort of shell scripts you've had to review in your day... I used to get far worse shell scripts, assuredly
Yeah bro, you can just use AI!
I just cannot come up with a good AI-is-actually-24/7-helping-me-out use case.
Please help: I wánt to need this!
Many Claude Code power users don’t really use IDEs anymore, so the only purpose of them working from their laptop instead of a phone is because that is the normal way to do it.
Here is a real use case: you are are responsible for some alerting channel. You have datadog/ cloud logging/ github all connected. You see a bunch of alerts come through while you are out and about and you prompt CC to investigate - Claude triages and says “all of the sudden you are getting time outs from this bank API your company partners with, this started an hour ago. It’s happening on ~15% of requests”. So you ping the guy at your company who does vendor relationships and go back to your weekend.
This is a non hypothetical example. Obviously it would be better if your job had a real on call rotation and more robust alerting and you wouldn’t be getting slack alerts on the weekend… but I take the approach this job affords me a lot of nice flexibility so it’s ok
You don't have an on-call rotation but do have people dedicated to vendor relationships, and that guy works on the weekend? I'm not sure how you completely avoid getting alerts on the weekend for third-party payment processor issues, which can happen anytime, if you actually want to transact business on the weekend.
> You don't have an on-call rotation but do have people dedicated to vendor relationships, and that guy works on the weekend?
I'm an account manager. My clients will phone at almost any time, weekends included, if they feel there issue wasn't yet looked at by the on call dev.
I said vendor here but it’s more like banks we work with. So there’s someone responsible for the technical side of banking relationships.
But yeah it’s kinda a zone where most weekends there’s no problems so it’s not a huge priority… until it is
somehow having a contact at a bank who works on the weekend is more difficult to believe than a random vendor
I don't know what exactly GP works on, but 'bank' covers a lot of ground and some things happen on weekends. 'Open Banking' APIs, and some payment rails for example.
> Many Claude Code power users don’t really use IDEs anymore
This is something we of the HN bubble take for granted. Most of us know how to type quickly and use editors and use macros and program scripting languages and compose regexes.
The vast majority of programmers do not know those things. As such, AI speeds them up tremendously.
Yeah good luck being employed in 3 years once this bubble popped when all you do is type some natural language into a phone screen. People being proud of not using an IDE anymore is such a foreign concept to me, who enjoys coding and got into the profession because of the love of that.
The “debugging” for these type of issues is looking at some logs and http responses and being like “ah if we get this error it means they restarted their firewall again and took us off the whitelist. Email that guy Joe at the bank and hope he responds”. It’s not rocket science or the majority of my job… but someone’s needs to do it. We automate all the stuff we can.
If you got into the industry due to enjoying the typing of code the future is looking pretty bleak.
I dunno.
I've been watching "How it's made" on Hulu to fall asleep at night.
I’m constantly surprised by how many things are made with human hands, despite the ability to automate.
An enormous amount of on-call debugging is just natural language reading of logs.
I run a lot of data science-type analyses that can take up to hours at a time to run, so Claude is « monitoring » tasks most of the time. I have it on remote-control so I get notified when a task is done or need clarification, but most importantly whenever I have a new idea, I can just ask Claude to queue it up. Most of the time my hardware is the bottleneck, not the subscription quotas.
There are so many neat use cases like this one that I'm skeptical of the AI-is-a-dot-com-bubble naysayers. Not to mention that, unlike in 2000, technical-adjacent people can now get a reasonable approximation of their ideas running quickly.
That makes sense - thanks. Do you use hooks for this?
I used to have some hooks for local notification, but lately I find that claude is pretty good at notifying through the app with remote control (but definitely not perfect)
> take up to hours at a time to run, so Claude is « monitoring » tasks most of the tim
How is Claude monitoring them for hours? Claude runs out of context and extremely long sessions are prohibitively expensive even according to Anthropic (after they dispense with the marketing bullshit of long running tasks)?
It launches these tasks in the background. It became really good at it a couple of months ago, now it sets monitors on timer (not something I instructed, so I assumed it’s part of the system prompt for this kind of tasks) and then just wait for the next prompt, for the background process to be done, or for a monitor to trigger a checkup.
> and then just wait for the next prompt
A single session running for multiple hours is prohibitively expensive, as per Anthropic. Regardless of whether it just waits for a prompt or does something.
Is the high expense coming from cache misses? If their workload does need to wait for a long time before it can continue, I wonder would starting new sessions and having to re-read the contexts and results anyway be any cheaper.
> Is the high expense coming from cache misses?
Yup. They can't keep your workload in cache forever, or they would run out of cache for users.
> I wonder would starting new sessions and having to re-read the contexts and results anyway be any cheaper.
Yes, that's what they recommend
Prohibitively expensive is all relative. Pre-Fable, I was getting fine on the 5x plan for 1-2 concurrent long-running tasks plus interactive work (I do a lot of coding for work, but it is not my full work day). I don’t think a cache miss every hour on Opus hurts that much, even at 500-600k context. It would be nice if they got /clear working on remote-control.
I've used it for the following when I've had tokens to burn:
[0]: https://github.com/microsoft/SkillOptThe other day I used Claude Cowork to create a checklist for me so that I can safely re-install my laptop... It went through all ssh configuration, apps, downloads, documents etc... Once I had the checklist, I asked it to do the backup itself, along with instructions to re-install/setup everything once I had the laptop re-installed. This was something I was procrastinating for too long already...
Few cases I have found very useful myself
1/ Using GUI software. My agents are using headful Google Chrome and Figma. It helps a lot to have separate environment, which is not interfering my main machine.
2/ Running long processes (1h+), so I can leave main machine closed.
3/ Running intensive processes. I use Gemma, Whisper and Qwen, which could burn main machine CPU and resources.
They help folks on fixed rate plans consistently hit their usage limits which provides them the feeling of getting their money's worth.
Letting it control a browser and searching for a pair of pants of a given size and length and color and style.
Yes, surprisingly, this is something Google cannot do yet.
i like using /remote-control to keep vibe dev running smoothly against my usage limits and deadlines
Running Claude code 24/7 on a code base on that “second Mac” so you can always continue after a usage limit reset, from your main device or from your phone?
I changed all smart speakers to retrofitted old radios with an amp and a pi. The hot word detection runs on the pi itself but whisper and LLM/task orchestrations goes over my local server with a 4080.
It's less about 24/7. It's more about it can't work when your laptop is in your bag and in transit and there is something that you have set up and want to run.
Oh yeah sounds great working in your free time and while traveling to buy groceries to feed yourself so you can continue working.
That time is otherwise worthless, so yes?
I don't value my travel time at all, but it used to be wasted on travelling.
Didn't imply or mention anything about working at a job in one's free time, but rather claude code continuing to work on whatever.
It could be for a personal project or hobby.
Having independently running processes from the computer you carry around offers benefits.
Pretty much doing everything that takes more time doing manually on computers
I think the main use case for AI bros is to setup a goofy looking dashboard, name it Jarvis to cosplay being Tony stark, and display stats for all the generated videos they're posting to social media.
(I wish I was joking)
Do you do software development? Is there any work left at the end of the day? Have it do that stuff.
I always have to correct its hallucinations during the day. Why would I ever let it run unsupervised overnight?
Running tests and optimize?
Running tests that the agent created for its own hallucinations? Optimize using another hallucinating agent?
Which tests and optimizations do you propose to run after a night of supervised work when one of main things that all agents keep doing is "load all records from db , and filter them in memory"? It's now become so bad, I had to literally vibecode a separate linter for this. And that's just one of the problems.
we don't have master AI that can create a great product out of a mediocre prompt.
but we do have sufficient AI to make a great product out of a great prompt.
garbage in -> garbage out hasn't gone anywhere.
so: much like to anyone that blindly complains that their compiler hates them : if you actually want help, provide information. If you just want to complain that the compiler is mean, scream at the sky.
plenty of people have figured out how to get this to work; more than enough to confirm that a straight <gambling-machine>/<hallucinatory-psychopath>/<random-number-generator> analogy is too simplistic to explain what we're working with.
Demagoguery. Your "great product" (where is it? show me these great products) will have wildly ineficient code and things like "read the entire DB into memory" regardless of the prompt.
> plenty of people have figured out how to get this to work
Plenty of people claim they have figured it out. In reality these people are full of shit and assume that if LLMs can produce working software, it's great working software. And also assume that LoC is a measure of quality.
Because without fail all the models keep doing this: https://news.ycombinator.com/item?id=48962703
And you can only see that in your "great product" if you actually read the code and understand what's going on.
>It's now become so bad, I had to literally vibecode a separate linter for this.
You see, there's your problem right there. You're vibe coding, which by definition literally means you're unwilling to look at the generated code. That's not what successful ai assisted software developers are doing. YOU HAVE TO READ THE CODE. Refusing to do that means you're not a serious programmer, you're outsourcing your thought and design and implementation, trying to get something for nothing by taking the easy way out, and you're going to get terrible results no matter what prompts you "engineer". There ain't no such thing as a free lunch (yet).
And while we're at it, to elaborate on what serf said: people mindlessly parroting terms like "stochastic parrot" to criticize llms without having read the actual paper that coined the term and understanding what it really claimed and how other papers responded to it means you're just a human stochastic parrot no better than what you're criticizing -- at least the llm has read all those papers and understands what "stochastic parrot" actually means in context. Ask it, it will be glad to explain!
Emily M. Bender, Timnit Gebru, Angelina McMillan-Major, and Margaret Mitchell. On the Dangers of Stochastic Parrots: Can Language Models Be Too Big? (FAccT 2021)
https://dl.acm.org/doi/10.1145/3442188.3445922
> You see, there's your problem right there. You're vibe coding, which by definition literally means you're unwilling to look at the generated code. That's not what successful ai assisted software developers are doing. YOU HAVE TO READ THE CODE.
I guess you vibe-read what I wrote. Let me write it again for you: "I always have to correct its hallucinations during the day. Why would I ever let it run unsupervised overnight?"
If this is your problem, I hate to say it, but it’s a skill issue on your side. I didnt even start developing software until LLMs even though I know all the primitives quite well and have a strong grasp of architecture. ADHD, perfection, and focus got in the way. I’m writing a pretty complex application and I don’t have the problems you’re running into. I regularly have my code reviewed by professional devs to ensure I’m not just vibecoding into slop. I’m not. I also read my code regularly and do a lot of exercises and courses to keep learning as well.
So I dunno what to say, except it’s possible to write really solid code with LLMs.
> I regularly have my code reviewed by professional devs
> I also read my code regularly
So you're literally doing what I am talking about.
Generating leads for new work, if you are a freelance. Automatically answering customer support emails, if you own a SaaS. Monitoring competitors' socials, websites, etc for new features you have to compete with. Monitoring updates on software you depend on for breaking change / deprecation announcements.
harvesting
ask AI to help
Same here
> setting up your spare Mac
as one has
i once thought the same
They seem to hang around.
You people are too far gone
You're being left behind.
For what it's worth, which may be nothing, I remember being told that by NFT people.
NFTs are not useful, claude is.
>claude is
How? I mean what could be the ultimate usefulness of Claude if not to make money, just like NFT but with extra steps. Most people isn't using Claude to make massive paradigm shift breakthrough discoveries. Nobody's curing cancer nor solving climate change. Probably the most common use case for llms it's just speed up the grind and make money. Like nfts.
If curing cancer or solving climate change is your definition of usefulness, almost anything humans do is useless, probably including your own profession. Though I don't want to speculate. Speculation is precisely why NFTs are stupid. Claude, on the other hand, if used effectively, can speed up people's work and increase productivity.
This implies a long road to catch up. But seems to be that all one needs to do to be on the cutting edge is to integrate a few scripts that support a cutting edge model. Which you then converse with in natural language. So how higher will the bar to entry be five years from now?
How am I being left behind by not using AI? I have a fully functional brain that I can use to do anything AI can and better.
You can't do it faster, certainly. But if you used it to do all the things you're slow at, you get all of the above.
Eventually everyone will be without a job other than the Bros working at AI companies.
Or the bubble will implode and we see ourselves at the job queue anyway.
It's a no win trap, except for the tech bros cashing in on investors burning money. Data centers using up the available clean drinking water, people with no jobs paying higher electrical bills, stressing the electrical power grid to the breaking point, historically unparalleled IP theft to divert money from creators to their tokens...
https://isaiprofitable.com/ (Is AI Profitable Yet?)
https://www.yahoo.com/entertainment/music/articles/largest-i... (“The largest IP theft in human history”)
https://www.recordinglaw.com/news/authors-protest-ai-empty-b... (Don't Steal This Book)
https://finance.yahoo.com/energy/articles/why-ai-boom-break-... (Why the AI Boom Is About to Break the U.S. Power Grid)
https://www.aljazeera.com/opinions/2026/1/21/ais-growing-thi... (AI’s growing thirst for water is becoming a public health risk)
Yeah these guys will set up claude on their third pc after they realize their second pc keeps hallucinating out code nobody cares about or needs in hopes of staying relevant. I'll just keep coding by hand because I ACTUALLY ENJOY it. What a novel concept these days.
Just letting you know you're not alone in coding by hand.
raises hand
I too confess to writing code by hand. It… works for me?
I've done this before as a free replacement for my OpenClaw bot. It loses its connection sometimes and you need to redo the handoff with your phone. But otherwise, nice to have a use for my old Mac. It's also running my Home Bridge, which Claude can interact with. Pretty cool.
I currently have Claude Desktop installed on a separate Mac mini M4 and control it with Dispatch. Is there a reason to do this method, it still seems the way I have it setup it has full control over the local account I gave it on the Mac mini.
Why would you need an M4 for this? If most of the "thinking" is happening on Anthropic's side, are you running particularly resource-intensive apps?
The gpu inference happens on anthropics side, but the output of the gpu inference is often many local tool calls. The agent is locally reading files, writing files, compiling programs, downloading artifacts, running tests, invoking CLIs, monitoring logs, inspecting web sites, grepping etc, and can be doing all of that at once, in parallel via subagents, much faster than a single human ever could.
I found Dispatch to be less capable than RC. The latter is more like the direct Code experience (though there are some gaps in RC mode vs in a Claude-owned sandbox).
I’d love to just have Claude use my machine as a sandbox host instead of having to run RC on each host session. (In case you are listening Boris ;) ).
In the meantime I have a janky master RC session that creates new tmux windows and Claude RC sessions for each new code trajectory that I want to run.
The other benefit here is you can drop down and use termux to use Code directly if you hit a RC bug, I found permissions UX to be a bit flaky in the iOS UI.
Dispatch/Cowork is basically claude code in a container. The section "Why not run it in a container?" would address your question in the post. One example I run into is that Cowork won't download and fill out or read pdfs or other files due to container permissions. Vanilla claude code has no problem using curl and wget.
Cowork gets tangled with git as well. Fails, and then can't delete lock files.
Running a helper from the terminal, making Claude work in a working directory, and then create a .commit file has been my workaround for this for a while now.
Imagine there's a better solution nowadays, but this allows me to use dispatch building on Vercel, so I can check it out from wherever, without too much pain.
Claude dispatch does things differently than this. Dispatch is very convenient for somethings (including connectors).
But increasingly it seems like dispatch was slapped on top of cowork incrementally, when there was not an integrated and cohesive strategy across cowork between mobile/desktop/laptop. This is kind of what many of us get/got to learn in our 20's.
Though it doesn't get by all the hurdles mentioned, it is alternatively possible to run Mac OS in a VM on your Mac using UTM and install Claude Code within the UTM VM. UTM can be run under a non-admin Mac account. This can allow you to use most Mac-native tooling, at least. The interactive performance of using the Claude Code ui on the VM isn't great, however. I'm not sure if you can log into the VM via terminal on the host from the non-root admin account to avoid the ui performance issues.
An argument against this method is UTM doesn't support graphics acceleration so browser support will be hobbled. Even if you don't need the acceleration, I've found browsers in UTM virtualized OS's can't get past some modern captchas and other browser fingerprinting checkpoints :\. It's terrible but that's the way it is.
I don't know about solving captchas, but if you are running macOS virtual machines on an Apple Silicon Mac, hardware graphics acceleration is supposed to be natively supported out of the box when using UTM's Apple Virtualization Framework backend.
A primary source of UI lag is how Apple's native virtualization framework processes multitouch trackpad events. There are other issues like mismatched resolutions and framerates, too. Ask your favorite ai to debug. You can try e.g. deselection of the trackpad setting and instead select basic mouse support, and that can clear things up..
> browsers in UTM virtualized OS's can't get past some modern captchas and other browser fingerprinting checkpoints :\
I'm using Qubes OS, where everything runs in VMs without GPU acceleration, and never experienced this.
I just redid my homelab/media server (switched from an old NUC I couldn’t figure out how to stop overheating even with some decent modification work, now using an hp elitedesk with an i5 processor that is handling my stack nicely). Thinking about setting up a vm on the base ubuntu install for isolation to run Claude in. May play with dispatch, may just put raw Claude code then use Moshi app on my iPhone and iPad.
I have a similar setup with Claude on an Ubuntu VM. It handles scheduled things that might otherwise be in a routine.
I haven’t set up dispatch yet. I wonder what a Mac gets me over this set up if I don’t need iMessage
I’ve been doing something similar with an old m2. It isn’t powerful enough for local models, well sufficient local models but for openclaw and Claude it’s been perfect.
MacBook m1/m2 also are cheap enough now vs an Mac mini which I was surprised about, not too surprised but yeah..
My setup is sort of reversed. The powerful machine (framework desktop) is my headless AI machine and M1 mbp is my daily driver. Works well!
I am sorry to report that 16GB+ MacBook Airs/MacBooks have become unreasonably expensive - probably from use cases like this.
If I had one I would definitely try creating a separate VLAN for it to control, otherwise it's isolated from your files but still has access to your network and devices in it.
Put it in a faraday cage with some kind of explosive device set to trip if it escapes containment. It's the only way to be sure.
Why a laptop at all?
Why not just use a VM in the cloud and just a CLI interface?
For me, it's mostly because I want whatever I'm running to stay on my own home network, with access to Home Assistant, my homelab, and so on. Beyond that, old laptops are hard to beat for this: I already have a few, plus spares, they sip power, the battery is a built-in UPS, and the screen and keyboard mean I have a console when something goes wrong — no hunting for a monitor. There are plenty of use cases for a cheap cloud VM or VPS too, of course.
In my case, most of my VMs live on a proxmox cluster of HP EliteDesk minis, with the laptop sitting outside that as the thing that manages the cluster and a few other odds and ends.
This can work, but increasingly there is benefit from local, including places that want data to be secure and not in the cloud. This is a thing. It's normal. It's just new to people who haven't realized how common it is.
Agents running as root and data to be secure do not fit well in the same sentence
I agree, and it's not necessary for agents to run as root and full access to data.
There has always been a trade off sold to consumers of security vs convenience and a belief that giving up a little security gets a lot of convenience.
It ends up being often about a convenience of adopting the new tech, not necessarily in a way that's in the best long term interests of each party.
Giving sudo permissions to an agent seems reckless. Claude gets his own unprivileged UNIX account, no more. I don't bother with containers or VMs however.
I mean, what's the worst that can happen? A complete takeover of the box by a hacker via prompt injection or something similar?
Dispatch works great, and I have reversed the setup so Claude can ssh-spawn sessions on my homelab for non-Mac dependent work
Why a spare machine has to be exactly a mac ? For iMessage?
New macs share memory between the cpu and the GPU so the GPU gets a lot of ram for a low cost (relative to everything else available right now.)
This is the paranoia talking, but given that Claude is going to be doing its own thing, I would put this box in its own VLAN or behind deny-all firewall rules to protect against network escapes.
It's so strange to me that tech crowd uses macs for use cases that spend all their time idle or serve some light http request.
It's a glorified API gateway...it could run on a medium sized potato
Almost like an entire generation just grew up coding on macbooks as the obvious choice for coders and just can't conceptualize hardware outside of that walled garden
for local LLM, obvious choice:
idle = doesn't have to be H200 speeds
uses macs = local LLM can use all the ram, be smarter
as for hosted models, potato is tasty
i just use terminus + wireguard + tmux and works great, i can control claude/codex from my phone while working out
Why not Claude remote control?
Not OP but in my case I've been using this setup for quite a while (well before Claude remote control was a thing), this works with any other tool / ssh session.
I'm not entirely sure there's a big advantage to Claude remote control to be honest. Maybe it's just me being afraid of lock in (I do switch between Claude and Codex somewhat often) and/or the inertia of changing things up that keeps me from trying it.
Too bad you can't let your agent accumulate your gains yet and still have to work out
same, blink iOS app + tailscale + mosh
I've been putting off learning Claude and this article had me strongly considering jumping in. Then I looked up Anthropic pricing and its 100x more convoluted than cloud services management. Its a goddamn full time job and independent skill set figuring out how to prevent going bankrupt from AI usage!
I think I'm gonna be a late adopter on this one until the industry figures out a less cumbersome pricing model.
Look at the subscriptions. That's how anyone using the stuff as an individual should be approaching it today if they are dealing directly with Anthropic or OpenAI.
With the $200 Claude subscription I was able to get around $13-15k of API equivalent usage in one month (note: this was during the "+50% usage" promotion that they have kept extending since May). When you hit your usage limit for a given time period you get cut off until the time period resets; don't bother paying for additional usage credits, you will be disappointed.
What I’m reading is this use case is for 3-5 days a week full time dev if you stick to off hours US time. That you can save significantly if you have spiky usage 1-2 days a week by going the ad hoc API route if you’re new and you need to install a bunch of monitoring tools to tell you how best to save week to eeek as your usage patterns change and you risk surprise bills in the 10s of thousands if you get it wrong.
The approach I took was to start with the $20 plan, then when it became clear it wasn't sufficient I upgraded to the $100 plan. The $20 plan didn't cover one night of coding. With the $100 plan I started bumping into the 5 hour usage limits after ramping up over a week or two.
My average API equivalent use is around $30-40/hr. I would just bite the bullet on a plan for one month, then use that to calibrate your expectations around usage and cost optimization. The plans are heavily subsidized.
Thanks for the insight! Really wish they had easier cheaper ways to learn.
I still don't understand what these freaks are doing running these agents 24/7 on machines. What are they doing? Managing a todo list? You mean crossing items off as you complete them? Research tasks? To do what?
Never really get good answers. There is no killer app. Just bikeshedding.
Swiping Tinder. It takes about 5000 matches to get a date. It’s easier to just automate it. It automatically adds dates to my calendar, all I have to do is show up. I get a summary of our chat history (well, what the agent wrote to her) in the notes section of the calendar entry and some pointers and talking points for the date.
Maybe I should have the agent also do a background check.
PS: This is a joke, but feel free to steal this idea.
Any sufficiently advanced satire...
Good news about hell: it doesn't exist. Bad news: Humans can pretty much create whatever they can imagine.
The future is agents chatting to each other on tinder and automating the initial getting to know you part. I can imagine that while that's going on, we could add like a little text chat box for the humans to chit chat with each other a bit and pass the time, before they can go on a date
It works well enough for bumble web, just make sure you have rate limiting..
Then the openclaw WhatsApp module…
Kidding of course.
Crap, I totally believed this. We live in a dystopia already.
Someone apparently made this
https://github.com/Grigorij-Dudnik/TinderGPT
> TinderGPT automates the process of writing and arranging dates with girls on Tinder, enabling you to generate romantic meetings with almost zero effort. Your only role is to like the profiles that catch your eye. After that, TinderGPT comes into the play. It initiates a conversation with the girl, using details from her profile, continues by building an emotional bond and highlighting your attractive traits, and finishes by arranging a meeting and giving you a push-up on your phone with her number.
This is not really the same at all. You still have to swipe, the value add here is apparently “building an emotional bond”. Together with the explicit goal of “dates with girls”, this actually feels incredibly nefarious.
This is a sure way to get girls! Girls love being entirely commoditized and objectified, famously that's a great way to date! /s
I’m surprised that the author didn’t even refer to them as females.
It seems the main use case is having Claude automatically write blogposts about how great using Claude is, then submit them wherever necessary.
There's lots of news about the billions AI companies spend on data center construction, but it feels like it's not even a fraction of the money they're spending on endless nonstop blogs about how great their app is at doing... things. Things that will never be defined.
It really feels to me like this OpenClaw type stuff is the new "I built a static site generator!" type blogs that just post a few articles about how they built their generator.
Exact same question as you. When the new ChatGPT app dropped it suggested to me to set up a task something like (paraphrased) “every Monday read my Gmail and Slack an make a summary and task list for the week”.
Why would I need an LLM to do this for me? That’s 5 minutes of work max, and doing it gets me in the flow of work again, to see what’s going on and needs to be done.
For a lot of folks summarizing a few days of work email and especially slack chats is way more than 5 minutes. Some work environments do not have great communication hygiene so it can be overwhelming to try to keep up with 500 emails a day and 38 Slack channels.
For the folks I talk to who use a LLM for this that seems to be the case. Takes a huge cognitive load off every morning and saves them an hour or two.
More or less a very expensive band aid over a bad work environment.
I kinda use it the same way in a sense. I have a little skill I run against our (horrible) task management system to summarize things and give me a punchlist to work through sorted by priority. This saves me thousands of clicks to do the same thing in the horrible web UI. A proper system in the first place would be a lot better!
At some point I’ll probably just take that to the next logical step and have the LLM write my own web interface to abstract and replace the horrible one entirely for me.
And how can they be sure the summary correct and doesn’t miss anything important?
In such environments everyone is constantly missing things and not replying until they get followed up with.
So that sort of thing is already normalized. An LLM is unlikely to be worse than a human. There is no accountability either way since everyone is failing at the task, it really doesn’t matter much if your or your bot misses a few percentage points of high priority items. You will probably be vastly outperforming your peers who are doing it manually.
This is very much just laundering not giving a shit through an LLM so you can blame it after the fact.
Because then OpenAI can read your emails and project communications and eventually build a model they will sell as an automated consultant. The CEOs will uncritically eat it up just long enough to cut the footing out from the industry. Once everyone is used to the sorry state of software, nobody will be able to imagine putting people to the task anymore and we'll have the new world order that Altman and Theil have been talking about creating.
I think you need to open your mind to the possibilities? For example:
- scanning logs for errors and
- opening issues which are then auto-triaged and
- PRs are opened for them and auto-reviewed and
- merged (and deployed).
This workflow alone is immensely powerful, and takes alot of burden off the team.
> This workflow alone is immensely powerful, and takes alot of burden off the team.
ITSM those unsupervised workflows are essentially an attempt at purported productivity in the near term at the expense of meaningful incremental long term burden for teams.
The only ostensible benefit is in the eyes of the AI-psychotic tinkerer, who knows no better, or in those of the clout-chasing developer farming likes on their LinkedIn posts.
Really they're not. But it seems you have decided that you, above all, know best.
I started my post with "it seems to me" precisely because I haven't decided that I know best.
> The only ostensible benefit is in the eyes of the AI-psychotic tinkerer, who knows no better, or in those of the clout-chasing developer farming likes on their LinkedIn posts.
Truly open minded
A company at the scale to benefit from this almost certainly has some kind of development sandbox environment and/or periodic job runner that's integrated into its environment and maintained by a team, not random Mac Minis.
None of these are things I want or need in the product I maintain with a team, there's really no point to any of this unless you run a vibe coded SaaS (?)
You want your team spending their time fixing these simple errors? The secret sauce is in the triage. We've adopted solutions alot like this, and now our team spends its time on much more meaningful work.
Why are the errors occurring, though? That's what boring analyse-and-fix addresses, through familiarity, recognition of patterns and "hang onnn..." moments.
It's like your AI agent is just plugging the leaks in the dyke each time, instead of fixing the architecture of the dam.
There are many sources of boring predictable errors which nonetheless are easy to miss and easy to fix. API validation errors for example.
Yes, I want my team to be deeply familiar with the codebase and every single little bug that needs fixing both trains them and let's them learn a little bit more about the codebase.
They can use agents. Like, team members don't need to be replaced, they can simply use agents when they deem it useful. If they see a trivial bug,they can put their agent on it and go work on something else meanwhile.
Your non-vibecoded non-SaaS application never has errors?
None of this requires running it 24/7.
So someone has to be in the office to press the "observe logs" button every 5 minutes?
> scanning logs for errors
famously a good job for a tool that takes 10-50k logs to run out of context and forget what it's doing.
Not really? Imagine for example looking for http status code 500 in an api log over the past hour. The nice thing here is it doesn’t matter if you get them all because it’s reoccur (or not).
> Imagine for example looking for http status code 500 in an api log over the past hour.
1. On a blog that no one visits maybe?
2. It's called a grep
3. For bigger projects it's called sentry
Dude, really open your mind and stop arguing mindlessly. We're using this to great success on a serious sass product with hundreds of thousands of users. I'm not some AI influencer, or blowing smoke.
The source of errors can be whatever you like. Sentry, grep, whatever. Its not the point. The point is that many of these errors are real issues, and can be fixed automatically and safely by agentic systems. It really saves time, and work by our team leaving them to actually concentrate on the thing that delivers real value.
You started with "scanning logs for errors" with LLMs which is one of the most stupid things could do with LLMs on a "serious SASS product with hundreds of thousands of users" (aka logs easily in gigabytes to hundreds of gigabytes per minute).
But now it's suddenly not "we use LLMs to scan logs for errors", but "we use actual tools to find erots in logs, but then just hand them off to an LLM and hope for the best".
As in "let's repeat what @troupo said but pretend it somehow goes against what @troupo said" lol
Let me guess -- in your day job you don't manage people. I have agents parsing messages, building out document sets, evaluating existing document sets, one is currently fixing a giant backlog of bugs and feature requests for a multi year personal coding project, one is exploring some ideas on speeding up inference at the edge..
If you put yourself in a position where you need more leverage (technical or operating) I think you might find you get some value.
Given all the automation you do, it sounds like you don't really manage people either.
I set it up out of curiosity a few months ago and realised I had no requirement for it whatsoever.
I’m actually very time-poor, so figured it could help be clawed back time doing… what exactly?
If you can’t think up enough coding projects to keep an agent busy in the background that’s a skill issue on your side.
I am aware this is likely sarcasm, but in case it isn't, what do you gain from doing side projects this way?
No sarcasm, I am completely serious.
I don’t have time for much leisure coding these days. I do have time to kick off a few tasks in the morning to progress my many side projects. Nothing public / oss, just code that I find useful/interesting like home automation, content pipelines, games, etc.
There are a bunch of cases where remote control from iOS onto a Mac Mini is simply nicer than using iOS Claude Code sandboxes.
It’s the same pattern as you (hopefully) apply at $dayjob. If you are not defining a /goal and letting your agent crank you are not making full use of the models’ capabilities.
Well I am fully of the opinion that LLMs can help in software programming, it's not something that I feel provides any value unless it has a human in the loop. The overhead of having to figure out if the agent did a good job, if the agent is actually done or not, and if the thing it built is shit or not, is worth simply avoiding by having a human in the loop.
So I wouldn't agree that the agent should be cranking out code all the time, in fact that seems more like a waste of resources compared to the work it creates. But I do understand home automation software can be very one-off and simple. But then again, a properly programmed home automation suite doesn't need a SOTA model to modify it, I think.
On all projects I've run any of the models they:
- infinitely duplicate any and all code, helpers and components
- infinitely duplicate CSS (because they duplicate components)
- continuously write code like "read the entire db into memory and run a filter function on retrieved data"
- continuously write code like "call db with multiple queries for each element in a list"
- etc. etc.
Why the hell would I ever want to run them unsupervised?
Have it work down my jira tickets while I’m sitting on the porcelain throne
What's a good way to give a limited amount of money to the LLM, say like 2k or 5k or something. But keep it completely separate from my identity.
Like I want the LLM to have a bank account and he can do ANYTHING with that bank account that he wants. But he can't fuck anything up that has to so with me. He only has 2 - 5k
Your LLM can fuck that up badly if your LLM starts to take loans, right? Or BNPL regardless of sandboxing?
Idk how you could at least in the US. Closest thing off the top of my head would be one of those checking accounts you can setup for kids. It would still be tied to you.
If you want Anthropic (or others) but anonymously, what you do is use https://openrouter.ai/pricing and fund your account from any of your preferred cryptos.
Yes you can anonymize from one frontier lab, but you've just now created a new source of identification for all your inference: OpenRouter.
This is the way.
Form an LLC, open bank account for the LLC, and use that.
Best idea. Thanks.
Solana/Ethereum