I (and I'm sure many others) have been thinking about this a lot over the last couple of months. I called it "Extremely Personal Software" in a blog post a few months ago (https://redfloatplane.lol/blog/14-releasing-software-now/) but there are lots of names and concepts floating about for the same basic idea.
I think it's possible the amount of new software that will be written for an audience of 1-10 will be greater in 2026 than in any previous year, and then the same again for many years to come. I also think a lot of this software will be essentially 'hidden' - people just writing this stuff for themselves because the cost to say things to an agent is very low compared with the cost of actually planning out a software design and so forth.
Interoperability will probably be important in the next few years and I wonder if this is something solvable at the agent/LLM level (standing instructions like 'typically, use sqlite, use plaintext, use open standards' or whatever). I also think observability and ops will be pretty important - many people who want personal software but don't care for the maintenance and upkeep.
The effort was not to onboard future professional software developers but rather to make the personal in personal computer, or PC, meaningful. It's YOUR computer, you can put YOUR software on it. In fact even pocket calculator do that.
> It's YOUR computer, you can put YOUR software on it. In fact even pocket calculator do that.
I'm pretty sure this exists. It's called OSS or, more ubiquitously, Linux.
The problem is, of course, no one wants to publish software for your PC/handmade OS. Which makes it a huge problem. You can't write every piece of your OS, without wasting huge amount of time. Nor do people generally want this.
OSS/Linux is "our" software. It's made by us for us (or others if you don't contribute).
Your software can be made by you, for you. It can be open source/free software if you want. Others can contribute to it, if you want but it can be open source without accepting external contributions also.
My point was to highlight that having software made by you for your machine is not new. Arguably the way to do so changed but I would say the principle remains.
Agreed I’ve already started writing software for myself using Claude. I would never have done this if it weren’t for AI - I simply don’t have the time otherwise .
I now have tailor made apps with all kinds of bells and whistles that commercial products can’t offer easily ( I fall under non commercial usage which opens a lot of doors ), and that free software might offer, but later.
I have also learnt a lot technically in the process, since I’ve been able to venture into what was for me unknown territory but at controlled cost
I plan to create more such apps in the future. What is certain though is that my cooking app has immediately displaced all the others on the market , because none of the others cater to my requirements.
The production side is indeed of specific interest - most users don’t run production software so I had to think about that one. Tailscale and Cloudflare came in quite handy and there is indeed a market here
I don't know how to tell you this, but people have been writing custom software for personal use for decades. I've been doing it since at least 2009! I find it hard to believe that there is a demographic of people that were yearning to write code, but simply could not because they lacked LLMs. Is it the price? Are people simply too cheap to buy books? Or have they simply "forgotten" how to patiently and thoughtfully read them? Or has the quality of tutorials/documentation of languages/libraries/framework online decayed in the last decade? Or is it really that people have struggled to type characters of code into their text editors[1]?
Basically, I am prepared to accept that there is a friction that LLMs lubricate away, but what is the source of the friction, and why am I (and a bunch of other colleagues) not feeling that friction daily in our practice?
[1]: And if so, where did we programmers and computer scientists go wrong? Were subroutines and macros not sufficient for automating all of that excess typing? Were Emacs and Vim simply not saving enough keystrokes? Did people forget how to touch-type?
> Basically, I am prepared to accept that there is a friction that LLMs lubricate away, but what is the source of the friction, and why am I (and a bunch of other colleagues) not feeling that friction daily in our practice?
You must be extremely talented and fast if LLMs make no difference for you.
For people like me though, it's another story: I've been doing this professionally for 25 years and of course, like many, I have been writing custom software for my own use all this time, on personal time. But with LLMs I get better results, faster and with very little effort. And that is the difference between another item in my list of unfinished software that consumed too much of my weekends and a cool utility/toy/useful thing I got after a few fun and interesting chat sessions.
> I find it hard to believe that there is a demographic of people that were yearning to write code, but simply could not because they lacked LLMs.
I still vaguely remember how difficult man pages were to understand when I first started reading them. I'm pretty sure the biggest obstacle is the fact that most documentation is written for people who already know the standard computer science terminology. I have a generally negative opinion of LLMs, but one thing they do very well is function as a "reverse dictionary". You can input a idiosyncratic description of something you want and get the standard terminology. This is a new and valuable capability.
There is a universe out there, where most of the world is reading Solaris man pages, instead of Linux man pages. Whatever your thoughts on the Solaris OS, I think it is fair to say that no operating system has ever matched the quality of its man pages.
Interestingly, I also converged on the "reverse dictionary" usage of LLMs, in around 2024[1], mostly to indulge in (human) language-learning.
An excerpt from the post below:
```
It is a phenomenal reverse dictionary (i.e. which English words mean "of a specific but unspecified character, quality, or degree"). It not only works for English, but also for Esperanto (i.e. which Esperanto words mean "of a specific but unspecified character, quality, or degree"), as well as my own obscure native language. This is a huge time-saver when learning languages (normal dictionaries won't cut it, and bi-lingual dictionaries are limited, if they are available at all). Even if you are just using a language you are fluent in, a reverse-dictionary-prompt can help you find words and usages, and can also help you find "dark spots" in the language's lexicon.
```
I've commented on this subject before, but the fact of the matter is that kids getting into high tech and programming mostly don't read books anymore. How do I know? Recently I was hanging out with a bunch of high school students who asked me how I learned. I said it was mostly via books and man pages. "Yeah, don't sleep on high quality written material. O'Reilly. Wiley. Addison-Wesley. Manning. MIT. No Starch Press. &c..."
Well. You should have seen the look on their faces. I might as well have morphed into the Steve Buscemi meme "How do you do, fellow kids?" They looked at me like I was a total relic or greybeard and said things like "Nah, nobody reads tech books anymore; I learned Typescript from YouTube videos."
Already in 2008, as a millennial teen without internet at home, I was learning C# and XNA without a single book, just tutorials and official docs I downloaded from the library alongside Visual Studio Express. I couldn't have afforded books on it anyway, but I can't imagine teens in 2026 using anything other than Youtube and some tutorials to learn this stuff.
I learned programming from tutorials :) Only after I kept encountering terms in tutorials (long after I was building (badly organized) programs) that I didn't understand well did I decide to read my first book, K&R's C. This was when animated gifs were a novelty not worth the data transfer time.
I think every generation feels like their way of learning was the best, but we all make it work. There was a time when the architects of systems directly tutored programmers on how to write programs.
> I find it hard to believe that there is a demographic of people that were yearning to write code, but simply could not because they lacked LLMs. Is it the price?
Yes, because the price is measured in time.
With LLM tooling I’ve churned out idiosyncratic tools that fit my use cases quickly. Takes maybe a day instead of a week. A week instead of months. The fast turnaround changes the economics of writing custom tools for myself.
Not speaking for the OP. But my biggest constraint is time. Now with agentic coding, I can work in 5 to 15 minute bursts a few times/day, and make meaningful progress on projects, where as before I would have never been able to context shift from my day job long enough on a personal project.
Yep! Time was the biggest factor. I could have created that one tool I had for years been wanting to make, but tech moves fast, and I have a job and a family and a passion for music and yadda yadda yadda. AI has been a game changer for actually accomplishing big dreams I just didn't have the time to bring about to fruition.
Well, I’ve been writing code for decades so I know because there was a time ( when I was younger ) where I did just this.
I also know that these days, for all kinds of reasons, I do not have the time to write the tools I’m writing now without AI. I don’t lack the ability, and I could - it will simply be multi months side projects that I can’t / won’t complete.
Given how often younger people find my typing speed startling, I think it has been somewhat forgotten (US high schools had "keyboarding" classes at one point but that seems to have fallen off...)
Seriously agree. I am wildly overeducated and I often think the most useful class I ever took in high school was my senior year elective for a typing class. On old IBM typewriters. And the only class I took in high school with non-honors kids. Typing insanely fast, especially for someone who is a fast thinker, is a bit of a magic power in itself.
There's a whole lot of people who want software to do certain things but whos job isn't programming and life requirements don't allow the time for all the book reading, tutorial running, and practice to write useful code.
I'm a long time ops guy. I script, but I spend most of my time configuring, patch testing, and keeping the low level infra running much of which doesn't require "coding" per say. Infra as code is in the grand scheme relatively new and still not ubiquitous despite what silicon valley would have you believe. I never had a need to learn to code to a level to do many of the things I'd like to see happen and find useful. Now I can make those software desires a reality without having to alter my career, preferred hobbies, or much of anything else about my life.
Speaking for myself, it's less of a yearning to write more code, than it is a yearning for tools that work a specific way.
I write plenty of code at my job, and generally don't have the desire to write more code as a hobby, except in rare cases when the mood really strikes.
Interesting points. With the extreme cheapening of the cost (time/skill) for software production, we can have "Extremely Personal Software", as you mention and as demonstrated by the source. I wonder if we will reach a stage where "software" is written by a computer for an audience of 1 and for a single task, to be run once only- via an interface that works for all tasks. The very concept of software as something that users have to learn to use (memorizing keybindings, for example), might go the way of the punch card.
More like Star Trek, we would just ask "computer" to do things, and its machinations (and "software") will be invisible to us. We would just have output to deal with.
I think this would mean a lot of things. I'm sure I can't fathom all of the implications, but it sure makes me feel old! Interesting times ahead.
I had the same reaction. We're headed into a period where you can shape your tools exactly as you like them; artisanal rather than factory-created workshops, essentially.
I think the instinct that APIs, validation layers, and so on take on a much higher importance is right.. I have a few internal tools that made sense to make libraries out of, and once the first library is good, and a test suite is comprehensive, porting to a bunch of different languages is extremely simple.
Everting that, it's also going to be simple for someone to hook up to this library with custom tooling.
Since roughly 1995 or so we've been in a world where quality tooling was provided by on the order of 1,000s of developers, mostly open source. GNU, Xorg, Apache, emacs, nginx, and so on. Or you could opt in to the Microsoft ecosystem.
The ~20 years prior to that we were in a world where you chose to align with either Microsoft's tooling, IBM, or shops providing Unix tooling from proprietary vendors.
I elide a nearly infinite amount of detail, obviously.
What's new now is that you can get your own window manager written to spec in under a week, perhaps much more quickly, not just choose one of a few major window managers and configure it in accordance with the chosen configuration options delivered by the large developer team.
The reason I don’t bother writing code this days is because my use cases have been solved, and if they weren’t, I’d tweak the most suitable candidate. One of my principles is to keep my workload small. More often than not, things starts with a small script or plugin, and then grow according to my needs. Why replicate what others have already done?
Because its fun. And because your experience using a tool is fundamentally different if you made it yourself, compared to if its something someone else made for you.
I don't think I can explain the difference, but it feels really different. Even if you used claude.
I agree. I’ve noticed this phenomenon too as I’ve been building more and more tools for myself, and I’ve started calling it hyper personal software: https://paulwrites.software/articles/hyps/
This. I have written so much software recently to make my computer my own. It’s been so much fun to be able to borrow the the ideas from different tools I have used (eg vim modal behaviours etc ) and also bring them together with some completely novel ideas to produce tools for myself that are one of a kind and that “fits me like a glove“
Too bad this is all on the work computer and need to bring it to my personal one but can’t copy paste lol. It’s been thrilling building g and using them and the time from an ideating a small enhancement/ optimization to actually using it is like 5 to 15 minutes away. Soo cool.
I shudder to think about the security implications of everyone rolling thier own software. I trust my OS/browser/file system is secure because thousands of people are invovled in a complex network of interests in keeping it secure, from the kid contributing his first bit of code to the PHds at NSA writing encryption standards. The idea that any one person can replace that network is laughable.
Just to be contrarian, perhaps some measure of risk is reduced by the scale of one.
Identifying a vulnerability that can be exploited against many thousands or millions of targets is perhaps more attractive than a single one of individually low value.
This of course would assume that vulnerabilities are in fact unique (which is admittedly questionable).
I had the exact same thought. Pretty low probability that there's going to be a script-kiddie exploit for your custom tools. Pretty decent probability that there will be vulnerabilities present if someone cares enough to target you.
The counterpoint to that is that the exact same tools that are allowing this personal software creation at massive scale are also excellent at black box vulnerability analysis…
But the exploits can use AI custom tools too. "Script Kiddie" is just now "Prompt Kiddie"
Although everyone might use their own flavor of "database" or "REST API", I can't imagine every layout to be unique enough to not have similar exploit classes entirely. AI isn't known for being super original after all...
We should expect the same automated personalization to be used offensively and for that personalization to be packaged into tools anyone can run (natural language interface, likely.)
(Appreciate your counterpoint for its own sake. It’s an interesting idea.)
If they’re hosting network services, sure. I wouldn’t put vibe-coded software outside a home network, ever. But it seems low risk if people are just creating their own desktop software: especially since it’s less likely to be vulnerable to widespread malware.
(Note: I’m not an LLM fan, don’t vibe code myself at all. But I would be unconcerned about security for the kind of things I would create if I did start doing so.)
That seems like a naive view to me. Most modern software development is gluing vendor code and libraries into a CRUD app, and I don't see why that would change with agents doing the majority of programming. If anything, there's an even bigger market for solid libraries and interoperability, plugging things together like LEGO - only for real this time.
While I wouldn't do asm, I love the approach and do much the same myself but in Ruby instead.
My wm, shell, terminal, editor, file manager, pop-up menu (dmenu-like) are all pure ruby (including font rendering and X11 bindings). These all started before I started using Claude to improve them, so they're still mostly hand-written, but that is changing.
They're messy, they have bugs and "misfeatures" that works for me but likely would be painful for others.
Like OP, I don't really recommend anyone else use my code, at least not directly, and that is extremely liberating.
Overall, the projects covers the largest surface of what I use beyond the kernel, a browser, and Xorg (I'm so, so tempted, but I think an LLM will need to get a lot further first before I could fit it into my schedule).
It doesn't need to be polished because it's mostly for me. It's okay for them to have bugs as long as they work better for me than the alternatives.
I strongly believe more people should do this. It's both a great learning experience, and it gives you a system that has exactly the features you actually want and use.
This is very cool. I wonder how much time did it actually take, and how much did it cost, because Clause Code is very much not free [1] [2]. It's more like hiring a robotic contractor, very fast, but with a serious hourly rate.
I'm on Claude Max, so it didn't cost me anything more than the subscription I already have. Had to use it for Something. As for time - for the full CHasm and Fe2O3 suite of sw, I started the work 2026-03-29 and have probably spent 60h or so of my time. But then again I have a very tailored CC setup that I have fine-tuned since last summer with more than 70 CC projects helping me get it the way I need it to be since then.
So, it's at most $400 in Claude expenses for a fully custom suite of software in 2 months. Even if your time is 300/h, it's less than $2k in your own time (which, I would expect, you enjoyed spending). That's insanely impressive.
I need Claude Max in any case for my work, so the cost is effectively null. And I do creative stuff in my spare time regardless, and I don't really think about my hourly rate when I play with my kids either ;)
Can't speak for their usage, but I calculated my token use with Copilot, my employer paid 22$ for >2000$ in tokens. Given estimates margins, Github is paying 98-99% of the costs for us.
I imagine Claude max etc is similar. So it's not sustainable at all
I've spent two weeks with the cheapest tiers of Claude Pro + pi.dev+GPT-5.5 (+ some deepseek-v4 via openrouter recently) to create my own bespoke version.
I'm at 90% feature parity currently and surpassed on some levels. For ~20€-ish I've soon replaced a 60€/year subscription service.
I haven't spent a single second thinking about how someone else might run it, it doesn't have logins, security or anything - because I'm going to run it 100% behind a Tailscale node with no external access. The release and deployment processes are exactly how I like it, other people might not. I don't need to care, it's mine =)
A few months ago I did the same with Hazel[0]. That took maybe one evening to get an MVP and a week of casual updates to make it pretty. Now I have my own macOS application that does the exact things I needed Hazel for. It's mine forever and I can add or remove features as I feel like it.
Sorry I have a question that is a little off-topic: what's the value of generating an image of a laptop on a desk? That's not like it's particularly relevant, when you could have integrated a screen shot of your set-up (like the same one you put on a few of your repos) or something more unique, and even if you want to show that, it's easy to find similar images with the same vibe, so I guess it's for some fun I missed in the process?
But doesn't that make it bad? It doesn't say anything new. Unlike the software in question, which is personalized, so it's not even symbolically reflecting the topic. It's a sheer waste of pixels and time spent looking at it or scrolling past the cognitive junk food.
I've got a wrapper around tmux for an audience of one. I can operate claude code, codex, opencode, or just a shell, from any of my devices to any of my devices (via tailscale) or more commonly, operate it on my exe.dev server.
I often continue a session on my phone, sometimes with voice. I have buttons for viewing files or following links the agent has referenced, extracted from the stream of text, and I have some buttons for exactly the git stuff I need. I have a button to toggle between yolo mode and normal.
Basically, very simple UI for everything I actually use, easy to use on a phone - and maybe more importantly, no UI for anything I don't personally use. Also all my machines have the repo for the uh, harness-harness, so I just open the tab for it if I need some changes and prompt them into existence and get the changes live.
All this is great, except it enables me to work every waking hour of my life. That part might be bad.
Do you have a way to use voice on your smartphone?
I use a whisper wrapper (also "built for one") for Windows and connect through SSH but it only works well enough because of the laptop's ARC graphics card. I wish I could do that when connecting to SSH from my smartphone.
FUTO has a superb Android keyboard with on-device voice transcription using a fine-tuned Whisper model, which is quite performant for me (and actually does fairly impressive text autocorrect and prediction too): https://keyboard.futo.org/
They also do a pure play voice input keyboard, if that's your jam or you want to use it with other keuboards: https://voiceinput.futo.org/
Brilliant! I hate it. The author will surely admit that there was "joy" in creating this suite of software, but it's a different kind of joy than most of us here would recognize. I am looking forward to being a part of the group of detractors doing things the old way, similar to the "small web" or other counter cultures on the Internet. I fantasize about being here to pick up the pieces after all the others went full-on into AI-assisted everything and lost their critical thinking capacity, programming skills, knowledge of Unix command line, etc.
There is part of me that understands the appeal of the all-in on AI and personalized software approach. It's a bit cyberpunk! In terms of open-source software, the downsides outweigh the benefits in my opinion, though. Important principles like community ownership and commitment are absent, and this approach is even radically antisocial. And then there's the inevitable issues with maintainability, to say noting about dependence on big tech companies.
I read somewhere (in the myriad blog posts dealing with this Cambrian LLM explosion) that software developers could be put into two camps: those that just want the thing to exist, and those that want to build and understand the thing (in addition to wanting it to exist).
those in the first camp are having a great time.
those in the second camp (which is how you're describing yourself, and how I'd describe myself) are wary and suspicious.
it is somewhat paradoxical, we've watched/read sci-fi/cyberpunk for years and dreamed of this kind of world. after all, when did you see any members of the Enterprise writing code? they just asked the computer to "write a subroutine" and that was that. what a world!
but here we are, with the craft in danger, not entirely impressed by the idea of "just ask and walk away".
i, too, fear for my loss of critical thinking, raw skills, and design sense, as do i think about being one of the few (in 2, 3, 5, 10 years) that didn't abdicate their cognition, their craft, to the tech overlords.
but i wonder if it will matter anyway. i wonder if "source code" will be a deep abstraction that nobody thinks about anyway, similar to how 99% of us don't care/need to care what the machine code we're eventually emitting does or looks like.
I, in theory, can plane a piece of wood with a hand planer. But I'll never do it again, we did it at school in ye olden times before the millennium and it was boring then as it is boring now.
I know people who get satisfaction from it, they take one sliver off with the hand planer, feel the wood with their hand and figure out the perfect angle for the next tiny sliver of wood to come out off, repeating the process over and over again.
I, personally, will just feed the damn plank to a mechanical planer with the exact specs of the resulting board set up. I just want the board smooth so I can get to the next step of the process. I'm not doubting the "wood-slop" the machine produces, I can see and measure if it's good enough or not. I don't need to be involved in the process.
We're both making a table, mine will be done faster. It might not be hand-crafted to perfection, but it will hold the stuff I intend to put on it just fine. If I find out it sucks later on, I can make a new one that's slightly better or fix the existing one. My goal was a functional product, not a piece of handcrafted art.
I don't think the analogy works. You're focusing on the "how", not the "what". Using a mechanical planer, you still need to dial in numbers yourself. You design your own table, the more modern tools just make it easier to realize your vision.
Another example: I enjoy writing with a good pen. But whether I write by pen or on a keyboard, it's still me writing it.
However, AI does basically all the real work, only leaving you to guide it. Make a table? AI gives you one with 2 legs. More legs? Guess I can live with 5 legs.
And you wouldn't be making that table, AI is. You cannot have pride in something that you never made yourself. It's the same as 3D-printing something from Thingiverse and claiming you made it.
People who create AI blog posts are not writing. Those that prompt their way to a piece of software are not doing software engineering. The ones that generate AI images are not being artists.
Your analogy is not really apples to apples though, is it?
More close is: if there was a table making machine, you just push a button and something like a table comes out, would you still be a woodworker? You haven't planed, nor measured, nor cut, nor jointed, you've only pressed on "make me a table"
Part of it's that the whole point of going into this industry is that I love coding and have been doing it since I was 8. Part of it is that I'm a control freak and it makes me uncomfortable to have to trust AI generated code. Sure, I already trust interpreters and compilers, but those are much more deterministic, and they don't generally do anything I have to be wary of. Part of it is that anytime I've used Claude to write stuff (using Opus 4.7 via an API key), I've had to handhold it when doing simple things (telling it repeatedly that a given column doesn't exist in Snowflake's task history table and eventually just giving up and taking it out by hand) and had to remove tons of completely pointless Python code it generates. The big difference is that the people in the first camp don't seem to care enough to check. Someone at my company used Claude to write 20k lines of code this past Friday. No way he read and scrutinized all of that in one day.
The other big thing I've noticed is that a lot of the people using it extensively seem to just be spitting out API endpoint after endpoint. Just doing endless CRUD with some light business logic. Yeah, it's not too hard to automate that with AI without any major issues. Hell, back when Ruby on Rails was hot, it was so fast to write those kinds of things with it that I could spin up things as fast as AI is doing now. Full websites or APIs in an hour or two because its syntactic sugar and scaffolding did what AI does with the FastAPI codebases I see these days. You could go from an ER diagram to a working app in minutes sometimes. I don't care that much if that kind of work is automated.
My experience is that often when I think "I wish my email / browser / calendar / … did X" it turns out be a limitation of the underlying protocol. So even if you build all software yourself, you still have to make compromises when you interact with the outside world.
Some of the folks who make things will go on to make things that suit not just their preferences but also those of a small audience.
Some of those audiences will go on to grow and grow and disrupt the big players.
The capital intensive part of software construction is melting away and being converted to opex (payg token costs and your time) and that will blast open the possibility space and lead to a massive new commons.
If the thing was so cheap to create why not open source it!
And if you like someone else’s open source thing but don’t want to take it wholesale why not give it to your agent and say “put the ideas from this onto my thing”!
Absolutely and you're dead on thinking about the opening of the possibility space. The value of software as an enterprise will fall as we enter an age of abundant, and often custom, bespoke software. There will be many great apps coming and some lousy apps.
Another thing to watch for is how chatty the internet is about to become. A great many of these apps will hit APIs, ping each other, and so forth.
> The value of software as an enterprise will fall as we enter an age of abundant, and often custom, bespoke software.
Not just software. I'm predicting we'll be getting bespoke books and comics in a few years. TV and Movies after that.
Basically there will be a service with mad-libs style book skeletons and you can get your exact specifics put in with LLM writing in them.
You want a romance novel with dragons (D&D style), a red-headed princess protagonist that's a bad-ass fighter and she has three men competing for her affection, each with a very distinct look that YOU specifically like.
Done.
This already being done by actual human writers, many (not all!) people read books based on tropes they like, the rest doesn't really matter. They basically check the tags of the story and if they match well enough, they'll buy it and read it. And I mean very specific: https://www.goodreads.com/series/151379-ice-planet-barbarian... =)
Real nice to see two accounts using LLMs discuss nothing like this. They warned us this would happen, and here we are. I guess the topic is apt, but we used to customize windows (95+) and Linux like this (not down to vibe-coded insecure replacement of apps, but display/desktop/widgets/explorer/transparency of components).
I just realized I'm following the same philosophy. I use suckless tools and made so many modifications to st, dwm and such it feels like home. I'm currently in the process of implementing my own git manager so that it would nicely integrate with my workflow.
On this software itself: I’d like to know how this feels to use. It’s so very lightweight. Does it feel categorically different to what we are used to?
One of the things I miss about the 1980s home computers is that they booted into a usable command line in a handful of seconds, from a few KB in ROM. Imagine what today’s HW could do if we’d retained that level of efficiency.
Clearly havent seen what enterprise hardware is like these days... sure, the OS takes 5 seconds.. but the hardware can take 10 minutes in some cases now glares at hpe gen11 systems. Its seriously bad now. The amount of power and time backround hardware level tasks now take has significantly increased over the last 10 years. Even the ancient dell r710 i have sititng in a closet collecting dust boots faster than todays hp gen11's.
We waste a ton of energy on ineffeciencies in hardware and software today all because we managed to "just go faster".
That’s wonderful! I’ve made ultra-lightweight web apps of my own to replace bloated, slow, and poor UIs. It’s a night and day difference when the dependencies are few-to-none. And that’s on a fat browser stack. Your ASM desktop must zip!
Related tangent: https://smolmachines provides microvms with cold-start bootup times around 200ms, and a "pack" utility and format to create self-contained binaries. No affiliation, but I just discovered it a few days ago, sharing bc I find it kind of exciting.
I find this fascinating. I also like to customize my desktop experience with my own code, but it's more assembling stuff with some additional code as glue.
A word of warning: a reliable lock tool for X11 is difficult. You should look at XSecureLock, which uses a multiprocess approach to avoid leaving the desktop unprotected in case of crash. It also implements a number of countermeasure to ensure the desktop stays locked and the locker stays in the front of the display. It's small too, so easy to audit (but written in C).
> It used to be that writing your own editor, your own file manager, your own window manager, was a project of years.
This is... Not really true? Especially if you are writing just for yourself. These are week-long projects at most to get to a usable state, if you know what you're doing. This is why there are so many text editors and window managers in the first place.
This is nice, but is also leagues away from something you’re written yourself. Take LLMs out of equation, and you have piles of code that you barely recognise and barely can edit or tweak by yourself.
I dont think it matters at all to OP. Sidestepping the insult, it sounds like they very, very much want to tools that support their needs only, methodology be damned.
What about the cybersecurity aspect of bespoke software?
A cybersecurity research company can now spend a small fortune on finding zero days in iOS because of the amount of people that use it. It basically guarantees there will be clients like government agencies willing to pay through the nose for the exploits.
Software made for one might disrupt this business model.
seriously.... we already have a constellation of good deterministic tooling for taking a relatively high concept spec to low level assembly. what does an llm offer in generating optimized asm that rust wouldn't??
Less memory footprint. No reliance on libs. Pure first-person control. No wasted CPU cycles is the target here for me. And if you read the post, the asm set is only for the desktop itself. The tools I use are in Rust. Result is: Laptop now runs at between 5-6W (down from ~9W) [XPS14 latest hw] on Ubuntu 26.04 - giving me around 3.5h extra battery life.
My guess is you're likely to waste more cycles on development time, and on suboptimal algorithms because the implementation is harder, than you would waste on rust-related bloat.
Still a cool project, thanks for sharing.
I have wondered about having LLMs output machine code directly and skipping the compiler/assembler altogether. Then you'd just commit your spec/prompt and run it through the LLM to get your binary.
rust can do that. You can run a hyper stripped down rust that was made for embedded devices specifically because those devices don't have room for a runtime.
Respect. I want to do the same thing. I'm studying electronics, PCBs and CAD in order to build the literal computer that I've always wanted as well. It's my lifetime project. Who knows if I'll ever succeed? I think AI significantly upped my chances but still.
Same reason people muck about with knowledge management systems... to put off the day when you have to sit down at your desk and actually do something.
I feel like build vs buy is the conversation now. I’m not a developer but I’ve built agents I use daily. When most people can vibe code their way to a custom app, value will most likely hinge on support and other “services”. Just my 2 cents, feel free to tell me I’m wrong!
Did you have to look or review any of the code produced, to get the performance/capabilities that you wanted, or were all interactions through CC? In other words, did you hit any walls with the pure agentic workflow?
I monitored the process very tightly. I have programmed my fair share of asm (and some 30 other programming languages), but for this I did not read any code. I hit lots of obstacles on this road, lots. In the process we also created a complete TTF rasterizer on par with what kitty gives me, and that was a true deep-dive.
I started porting TurboVision from C++ to mojo with Claude as a sort of joke. Then I said, why not port the entire Turbo editor? And then like. hmm.. it doesn't have all the modern affordances like language servers, syntax highlighting for all languages, debuggers for everything, real-time find-in-files, projects, etc. Well now I have all this implemented: https://github.com/boxed/TurboKod
Not sure I can use it as a daily driver yet, but it would be pretty cool!
Finally the great solipsism singularity collapse can be mine!
ok seems a lot of fun (for those like-minded), but who seriously want to be dealing with maintenance of everything they use in the long term, in pure assembly all the more?
I was inspired by Nicklaus Wirth when he wrote his own language opearting system and CPU. LLM allow many more to do this like you so am seriously impressed. It is also fun.
OP did this:
Prompted CC for all the points I wanted included (something like a 200 word prompt) and asked CC to draft it, including all the links added to the table I furnished. Then I edited the draft (about 50% then edited). Then asked CC to spellcheck and fixed the 5 it found.
Thank you. It would have been nice to see you personalize the hook and show your storytelling voice the way you personalized your computer in the story, but we aren't all poets.
Many people care, with good reason. We learned to notice LLM-isms is because they are, in fact, a very strong predictor that a piece of text is in fact garbage that's not worth your time reading.
I usually stop reading at the first LLM-ism, but I found the premise of this post interesting enough to keep going - and guess what, the entire article was literally just "I prompt CC to make software tailored for me" blown out to 8 sections.
Instead of replacing, downsize. Do you really need a GUI for git? Do you need the JetBrains Smart Ultra Analysis Pro™? Do you even need syntax highlighting? How far can you go?
Its fun, and a lot more rewarding than replacing tool X with tool Y, realizing you actually hate it.
I think this is going to be the OS of the future. You tell the computer what you want to do, and it uses the OS's APIs to create your program for you. No more copilot embedded in notepad unless that what you ask for.
Most software is done after the first or second version and the developers just keep working on it to justify their job; adding features no one needs and just get in the way or make the program worse. It'll be nice when the software I have does exactly what I need and doesn't change until I tell it to change for something I need.
The only feature Macos has shipped in the past 10 years that I actually like is air-drop. Everything else is a PITA annoyance, or as I've found out from upgrading, just bug ridden slop that doesn't work well anymore.
I've been building an object oriented system re-imagined in a world with LLMs called Abject (https://abject.world) and one thought I had was to build an OS that boots into my project. One way to do it would be a minimal linux distro (think firefox os or similar). Has anyone done something like this with their projects?
Haha, it's funny that we've all reached the same conclusion. I, too, believe in the same idea[0][1]. What is fascinating to me is how many things can now be elided from software. I don't use configuration files or things like that. I can simply hardcode everything in because there is only one user. If I want to configure it the other way, I just modify it and rebuild it.
The other thing is that other people's applications are rarely useful. Their libraries are, the feature description READMEs are, but the software itself is full of attempts at generality that make them overly annoying for me to use. Instead I have extremely idiosyncratic software - anyone else would find it insufferable.
The wild thing, though, is that my software is outrageously useful for me. I can see why Anthropic and OpenAI are (or shortly will be) the trillion-dollar behemoths they are. They are enabling a personal productivity increase of epic proportions[2]. The highly specific functionality also means strange things performance wise. I don't need to use Electron or Tauri or whatever. Instead, my thing is Rust with objc2 and it starts instantaneously. On my M1 Max, it's the fastest text viewer I can start. 100s of megabytes of JSON and it's launching is imperceptible for my tool, pretty-printing is instantaneous, breadcrumbs are live.
Because I can make it do only the thing I want it to do. It can't do other things. I cannot edit or auto-complete or anything. And this is great. Useless to others and fantastic to me.
Likewise, my blog is on Mediawiki (which I like so anyone can edit) but the authoring flow is kind of annoying. Uploading images causes a break from writing, and requires a lot of form-filling that interrupts my thought. So I now have this software that does everything I want: link autocompletion, background image uploads, post-hoc publishing, previews and diffs, built-in Wikipedia search to interwiki link. Who would want this but me? It only brings me pleasure.
I (and I'm sure many others) have been thinking about this a lot over the last couple of months. I called it "Extremely Personal Software" in a blog post a few months ago (https://redfloatplane.lol/blog/14-releasing-software-now/) but there are lots of names and concepts floating about for the same basic idea.
I think it's possible the amount of new software that will be written for an audience of 1-10 will be greater in 2026 than in any previous year, and then the same again for many years to come. I also think a lot of this software will be essentially 'hidden' - people just writing this stuff for themselves because the cost to say things to an agent is very low compared with the cost of actually planning out a software design and so forth.
Interoperability will probably be important in the next few years and I wonder if this is something solvable at the agent/LLM level (standing instructions like 'typically, use sqlite, use plaintext, use open standards' or whatever). I also think observability and ops will be pretty important - many people who want personal software but don't care for the maintenance and upkeep.
I called it "software".
It's so strange to me that since the 1960s with BASIC then later on dozens of https://en.wikipedia.org/wiki/List_of_educational_programmin... including Logo by Feurzeig/Papert/Solomon there is effort to precisely help beginners program software.
The effort was not to onboard future professional software developers but rather to make the personal in personal computer, or PC, meaningful. It's YOUR computer, you can put YOUR software on it. In fact even pocket calculator do that.
We keep on re-discovering the foundations.
> It's YOUR computer, you can put YOUR software on it. In fact even pocket calculator do that.
I'm pretty sure this exists. It's called OSS or, more ubiquitously, Linux.
The problem is, of course, no one wants to publish software for your PC/handmade OS. Which makes it a huge problem. You can't write every piece of your OS, without wasting huge amount of time. Nor do people generally want this.
OSS/Linux is "our" software. It's made by us for us (or others if you don't contribute).
Your software can be made by you, for you. It can be open source/free software if you want. Others can contribute to it, if you want but it can be open source without accepting external contributions also.
My point was to highlight that having software made by you for your machine is not new. Arguably the way to do so changed but I would say the principle remains.
Agreed I’ve already started writing software for myself using Claude. I would never have done this if it weren’t for AI - I simply don’t have the time otherwise .
I now have tailor made apps with all kinds of bells and whistles that commercial products can’t offer easily ( I fall under non commercial usage which opens a lot of doors ), and that free software might offer, but later.
I have also learnt a lot technically in the process, since I’ve been able to venture into what was for me unknown territory but at controlled cost
I plan to create more such apps in the future. What is certain though is that my cooking app has immediately displaced all the others on the market , because none of the others cater to my requirements.
The production side is indeed of specific interest - most users don’t run production software so I had to think about that one. Tailscale and Cloudflare came in quite handy and there is indeed a market here
I don't know how to tell you this, but people have been writing custom software for personal use for decades. I've been doing it since at least 2009! I find it hard to believe that there is a demographic of people that were yearning to write code, but simply could not because they lacked LLMs. Is it the price? Are people simply too cheap to buy books? Or have they simply "forgotten" how to patiently and thoughtfully read them? Or has the quality of tutorials/documentation of languages/libraries/framework online decayed in the last decade? Or is it really that people have struggled to type characters of code into their text editors[1]?
Basically, I am prepared to accept that there is a friction that LLMs lubricate away, but what is the source of the friction, and why am I (and a bunch of other colleagues) not feeling that friction daily in our practice?
[1]: And if so, where did we programmers and computer scientists go wrong? Were subroutines and macros not sufficient for automating all of that excess typing? Were Emacs and Vim simply not saving enough keystrokes? Did people forget how to touch-type?
> Basically, I am prepared to accept that there is a friction that LLMs lubricate away, but what is the source of the friction, and why am I (and a bunch of other colleagues) not feeling that friction daily in our practice?
You must be extremely talented and fast if LLMs make no difference for you.
For people like me though, it's another story: I've been doing this professionally for 25 years and of course, like many, I have been writing custom software for my own use all this time, on personal time. But with LLMs I get better results, faster and with very little effort. And that is the difference between another item in my list of unfinished software that consumed too much of my weekends and a cool utility/toy/useful thing I got after a few fun and interesting chat sessions.
> I find it hard to believe that there is a demographic of people that were yearning to write code, but simply could not because they lacked LLMs.
We didn't lack LLMs, we lacked time and energy.
I still vaguely remember how difficult man pages were to understand when I first started reading them. I'm pretty sure the biggest obstacle is the fact that most documentation is written for people who already know the standard computer science terminology. I have a generally negative opinion of LLMs, but one thing they do very well is function as a "reverse dictionary". You can input a idiosyncratic description of something you want and get the standard terminology. This is a new and valuable capability.
There is a universe out there, where most of the world is reading Solaris man pages, instead of Linux man pages. Whatever your thoughts on the Solaris OS, I think it is fair to say that no operating system has ever matched the quality of its man pages.
Interestingly, I also converged on the "reverse dictionary" usage of LLMs, in around 2024[1], mostly to indulge in (human) language-learning.
An excerpt from the post below:
``` It is a phenomenal reverse dictionary (i.e. which English words mean "of a specific but unspecified character, quality, or degree"). It not only works for English, but also for Esperanto (i.e. which Esperanto words mean "of a specific but unspecified character, quality, or degree"), as well as my own obscure native language. This is a huge time-saver when learning languages (normal dictionaries won't cut it, and bi-lingual dictionaries are limited, if they are available at all). Even if you are just using a language you are fluent in, a reverse-dictionary-prompt can help you find words and usages, and can also help you find "dark spots" in the language's lexicon. ```
[1]: https://galacticbeyond.com/chat-room-dispatches-intelligence...
I've commented on this subject before, but the fact of the matter is that kids getting into high tech and programming mostly don't read books anymore. How do I know? Recently I was hanging out with a bunch of high school students who asked me how I learned. I said it was mostly via books and man pages. "Yeah, don't sleep on high quality written material. O'Reilly. Wiley. Addison-Wesley. Manning. MIT. No Starch Press. &c..."
Well. You should have seen the look on their faces. I might as well have morphed into the Steve Buscemi meme "How do you do, fellow kids?" They looked at me like I was a total relic or greybeard and said things like "Nah, nobody reads tech books anymore; I learned Typescript from YouTube videos."
Already in 2008, as a millennial teen without internet at home, I was learning C# and XNA without a single book, just tutorials and official docs I downloaded from the library alongside Visual Studio Express. I couldn't have afforded books on it anyway, but I can't imagine teens in 2026 using anything other than Youtube and some tutorials to learn this stuff.
I learned programming from tutorials :) Only after I kept encountering terms in tutorials (long after I was building (badly organized) programs) that I didn't understand well did I decide to read my first book, K&R's C. This was when animated gifs were a novelty not worth the data transfer time.
I think every generation feels like their way of learning was the best, but we all make it work. There was a time when the architects of systems directly tutored programmers on how to write programs.
That has been the case for a decade
This is the best use case of LLMs, the one I use it the most for.
> I find it hard to believe that there is a demographic of people that were yearning to write code, but simply could not because they lacked LLMs. Is it the price?
Yes, because the price is measured in time.
With LLM tooling I’ve churned out idiosyncratic tools that fit my use cases quickly. Takes maybe a day instead of a week. A week instead of months. The fast turnaround changes the economics of writing custom tools for myself.
Not speaking for the OP. But my biggest constraint is time. Now with agentic coding, I can work in 5 to 15 minute bursts a few times/day, and make meaningful progress on projects, where as before I would have never been able to context shift from my day job long enough on a personal project.
Yep! Time was the biggest factor. I could have created that one tool I had for years been wanting to make, but tech moves fast, and I have a job and a family and a passion for music and yadda yadda yadda. AI has been a game changer for actually accomplishing big dreams I just didn't have the time to bring about to fruition.
Well, I’ve been writing code for decades so I know because there was a time ( when I was younger ) where I did just this.
I also know that these days, for all kinds of reasons, I do not have the time to write the tools I’m writing now without AI. I don’t lack the ability, and I could - it will simply be multi months side projects that I can’t / won’t complete.
Given how often younger people find my typing speed startling, I think it has been somewhat forgotten (US high schools had "keyboarding" classes at one point but that seems to have fallen off...)
Seriously agree. I am wildly overeducated and I often think the most useful class I ever took in high school was my senior year elective for a typing class. On old IBM typewriters. And the only class I took in high school with non-honors kids. Typing insanely fast, especially for someone who is a fast thinker, is a bit of a magic power in itself.
There's a whole lot of people who want software to do certain things but whos job isn't programming and life requirements don't allow the time for all the book reading, tutorial running, and practice to write useful code.
I'm a long time ops guy. I script, but I spend most of my time configuring, patch testing, and keeping the low level infra running much of which doesn't require "coding" per say. Infra as code is in the grand scheme relatively new and still not ubiquitous despite what silicon valley would have you believe. I never had a need to learn to code to a level to do many of the things I'd like to see happen and find useful. Now I can make those software desires a reality without having to alter my career, preferred hobbies, or much of anything else about my life.
Speaking for myself, it's less of a yearning to write more code, than it is a yearning for tools that work a specific way.
I write plenty of code at my job, and generally don't have the desire to write more code as a hobby, except in rare cases when the mood really strikes.
I think I agree. But at the same time we have strength in numbers and people will find something close to what they want and fork off that.
So I think the same thesis holds for audiences of 10-100 and 100-1000.
A cambrian explosion of software.
Interesting points. With the extreme cheapening of the cost (time/skill) for software production, we can have "Extremely Personal Software", as you mention and as demonstrated by the source. I wonder if we will reach a stage where "software" is written by a computer for an audience of 1 and for a single task, to be run once only- via an interface that works for all tasks. The very concept of software as something that users have to learn to use (memorizing keybindings, for example), might go the way of the punch card.
More like Star Trek, we would just ask "computer" to do things, and its machinations (and "software") will be invisible to us. We would just have output to deal with.
I think this would mean a lot of things. I'm sure I can't fathom all of the implications, but it sure makes me feel old! Interesting times ahead.
I had the same reaction. We're headed into a period where you can shape your tools exactly as you like them; artisanal rather than factory-created workshops, essentially.
I think the instinct that APIs, validation layers, and so on take on a much higher importance is right.. I have a few internal tools that made sense to make libraries out of, and once the first library is good, and a test suite is comprehensive, porting to a bunch of different languages is extremely simple.
Everting that, it's also going to be simple for someone to hook up to this library with custom tooling.
Really interesting period in computing, for sure.
> We're headed into a period where you can shape your tools exactly as you like the
What period were we for the past 50 years?
Since roughly 1995 or so we've been in a world where quality tooling was provided by on the order of 1,000s of developers, mostly open source. GNU, Xorg, Apache, emacs, nginx, and so on. Or you could opt in to the Microsoft ecosystem.
The ~20 years prior to that we were in a world where you chose to align with either Microsoft's tooling, IBM, or shops providing Unix tooling from proprietary vendors.
I elide a nearly infinite amount of detail, obviously.
What's new now is that you can get your own window manager written to spec in under a week, perhaps much more quickly, not just choose one of a few major window managers and configure it in accordance with the chosen configuration options delivered by the large developer team.
The reason I don’t bother writing code this days is because my use cases have been solved, and if they weren’t, I’d tweak the most suitable candidate. One of my principles is to keep my workload small. More often than not, things starts with a small script or plugin, and then grow according to my needs. Why replicate what others have already done?
Because its fun. And because your experience using a tool is fundamentally different if you made it yourself, compared to if its something someone else made for you.
I don't think I can explain the difference, but it feels really different. Even if you used claude.
"Everting"?
I learned it in the math context - a sphere eversion is a 3 dimensional process that ends with the inside of the sphere becoming the outside.
I had to check it up too. Appears to be a synonym for "inverting" used in some fields like biology and medicine.
Reminds me of this blog post and conference presentation on home cooked software by Maggie Appleton.
https://maggieappleton.com/home-cooked-software
Great post, thanks for sharing. Evidently I am far behind the real thinkers! The Robin Sloan post mentioned is also very good.
I agree. I’ve noticed this phenomenon too as I’ve been building more and more tools for myself, and I’ve started calling it hyper personal software: https://paulwrites.software/articles/hyps/
This. I have written so much software recently to make my computer my own. It’s been so much fun to be able to borrow the the ideas from different tools I have used (eg vim modal behaviours etc ) and also bring them together with some completely novel ideas to produce tools for myself that are one of a kind and that “fits me like a glove“
Too bad this is all on the work computer and need to bring it to my personal one but can’t copy paste lol. It’s been thrilling building g and using them and the time from an ideating a small enhancement/ optimization to actually using it is like 5 to 15 minutes away. Soo cool.
I vibe coded an image viewer for myself to address a big pain point that had been annoying me for years. [1]
[1] https://daniel.lawrence.lu/blog/2025-10-22-sriv-simple-rust-...
A really good and thoughtful response. Thanks.
I shudder to think about the security implications of everyone rolling thier own software. I trust my OS/browser/file system is secure because thousands of people are invovled in a complex network of interests in keeping it secure, from the kid contributing his first bit of code to the PHds at NSA writing encryption standards. The idea that any one person can replace that network is laughable.
Just to be contrarian, perhaps some measure of risk is reduced by the scale of one.
Identifying a vulnerability that can be exploited against many thousands or millions of targets is perhaps more attractive than a single one of individually low value.
This of course would assume that vulnerabilities are in fact unique (which is admittedly questionable).
If a vulnerability of the common not individualized ancestor software is found, how quickly do people patch their individual versions of the software?
I had the exact same thought. Pretty low probability that there's going to be a script-kiddie exploit for your custom tools. Pretty decent probability that there will be vulnerabilities present if someone cares enough to target you.
The counterpoint to that is that the exact same tools that are allowing this personal software creation at massive scale are also excellent at black box vulnerability analysis…
There are entire vulnerability/fault/misdesign classes that are fairly general and appear to naturally emerge.
See e.g the lock screen gap that another commenter noted in a nearby thread.
But the exploits can use AI custom tools too. "Script Kiddie" is just now "Prompt Kiddie"
Although everyone might use their own flavor of "database" or "REST API", I can't imagine every layout to be unique enough to not have similar exploit classes entirely. AI isn't known for being super original after all...
Otoh, TAU will bound to get really personal now:D
We should expect the same automated personalization to be used offensively and for that personalization to be packaged into tools anyone can run (natural language interface, likely.)
(Appreciate your counterpoint for its own sake. It’s an interesting idea.)
If they’re hosting network services, sure. I wouldn’t put vibe-coded software outside a home network, ever. But it seems low risk if people are just creating their own desktop software: especially since it’s less likely to be vulnerable to widespread malware.
(Note: I’m not an LLM fan, don’t vibe code myself at all. But I would be unconcerned about security for the kind of things I would create if I did start doing so.)
But your browser will invite outside software into your network, to run on your machine. So you have to be up to speed with community knowledge.
That seems like a naive view to me. Most modern software development is gluing vendor code and libraries into a CRUD app, and I don't see why that would change with agents doing the majority of programming. If anything, there's an even bigger market for solid libraries and interoperability, plugging things together like LEGO - only for real this time.
While I wouldn't do asm, I love the approach and do much the same myself but in Ruby instead.
My wm, shell, terminal, editor, file manager, pop-up menu (dmenu-like) are all pure ruby (including font rendering and X11 bindings). These all started before I started using Claude to improve them, so they're still mostly hand-written, but that is changing.
They're messy, they have bugs and "misfeatures" that works for me but likely would be painful for others.
Like OP, I don't really recommend anyone else use my code, at least not directly, and that is extremely liberating.
Overall, the projects covers the largest surface of what I use beyond the kernel, a browser, and Xorg (I'm so, so tempted, but I think an LLM will need to get a lot further first before I could fit it into my schedule).
It doesn't need to be polished because it's mostly for me. It's okay for them to have bugs as long as they work better for me than the alternatives.
I strongly believe more people should do this. It's both a great learning experience, and it gives you a system that has exactly the features you actually want and use.
And it's only going to get easier to do this.
This is very cool. I wonder how much time did it actually take, and how much did it cost, because Clause Code is very much not free [1] [2]. It's more like hiring a robotic contractor, very fast, but with a serious hourly rate.
[1]: https://fortune.com/2026/04/28/nvidia-executive-cost-of-ai-i...
[2]: https://www.briefs.co/news/uber-torches-entire-2026-ai-budge...
I'm on Claude Max, so it didn't cost me anything more than the subscription I already have. Had to use it for Something. As for time - for the full CHasm and Fe2O3 suite of sw, I started the work 2026-03-29 and have probably spent 60h or so of my time. But then again I have a very tailored CC setup that I have fine-tuned since last summer with more than 70 CC projects helping me get it the way I need it to be since then.
So, it's at most $400 in Claude expenses for a fully custom suite of software in 2 months. Even if your time is 300/h, it's less than $2k in your own time (which, I would expect, you enjoyed spending). That's insanely impressive.
I need Claude Max in any case for my work, so the cost is effectively null. And I do creative stuff in my spare time regardless, and I don't really think about my hourly rate when I play with my kids either ;)
Did you miss a factor of 10 in that time-cost calculation?
As a hobby, normal rates don’t apply, but just not to be misleading on the equivalent cost.
Indeed, I forgot it. Must be $10-20k worth of human engineering time.
@geir_isene: This is insanely cool, thanks for writing about it.
Do you know how much it would have been at API prices?
Can't speak for their usage, but I calculated my token use with Copilot, my employer paid 22$ for >2000$ in tokens. Given estimates margins, Github is paying 98-99% of the costs for us. I imagine Claude max etc is similar. So it's not sustainable at all
No idea. Probably quite expensive. I usually run 4-5 concurrent CC sessions, so hard to pinpoint for just CHasm and Fe2O3
My irccloud.com subscription is about 60€/year.
I've spent two weeks with the cheapest tiers of Claude Pro + pi.dev+GPT-5.5 (+ some deepseek-v4 via openrouter recently) to create my own bespoke version.
I'm at 90% feature parity currently and surpassed on some levels. For ~20€-ish I've soon replaced a 60€/year subscription service.
I haven't spent a single second thinking about how someone else might run it, it doesn't have logins, security or anything - because I'm going to run it 100% behind a Tailscale node with no external access. The release and deployment processes are exactly how I like it, other people might not. I don't need to care, it's mine =)
A few months ago I did the same with Hazel[0]. That took maybe one evening to get an MVP and a week of casual updates to make it pretty. Now I have my own macOS application that does the exact things I needed Hazel for. It's mine forever and I can add or remove features as I feel like it.
[0] https://www.noodlesoft.com/whats-new-in-hazel-6/
Sorry I have a question that is a little off-topic: what's the value of generating an image of a laptop on a desk? That's not like it's particularly relevant, when you could have integrated a screen shot of your set-up (like the same one you put on a few of your repos) or something more unique, and even if you want to show that, it's easy to find similar images with the same vibe, so I guess it's for some fun I missed in the process?
I like the image. It was simple.
But doesn't that make it bad? It doesn't say anything new. Unlike the software in question, which is personalized, so it's not even symbolically reflecting the topic. It's a sheer waste of pixels and time spent looking at it or scrolling past the cognitive junk food.
I've got a wrapper around tmux for an audience of one. I can operate claude code, codex, opencode, or just a shell, from any of my devices to any of my devices (via tailscale) or more commonly, operate it on my exe.dev server.
I often continue a session on my phone, sometimes with voice. I have buttons for viewing files or following links the agent has referenced, extracted from the stream of text, and I have some buttons for exactly the git stuff I need. I have a button to toggle between yolo mode and normal.
Basically, very simple UI for everything I actually use, easy to use on a phone - and maybe more importantly, no UI for anything I don't personally use. Also all my machines have the repo for the uh, harness-harness, so I just open the tab for it if I need some changes and prompt them into existence and get the changes live.
All this is great, except it enables me to work every waking hour of my life. That part might be bad.
Do you have a way to use voice on your smartphone?
I use a whisper wrapper (also "built for one") for Windows and connect through SSH but it only works well enough because of the laptop's ARC graphics card. I wish I could do that when connecting to SSH from my smartphone.
FUTO has a superb Android keyboard with on-device voice transcription using a fine-tuned Whisper model, which is quite performant for me (and actually does fairly impressive text autocorrect and prediction too): https://keyboard.futo.org/
They also do a pure play voice input keyboard, if that's your jam or you want to use it with other keuboards: https://voiceinput.futo.org/
You could always just use whisper as a keyboard on the phone: https://github.com/woheller69/whisperIME
I do the same also
Brilliant! I hate it. The author will surely admit that there was "joy" in creating this suite of software, but it's a different kind of joy than most of us here would recognize. I am looking forward to being a part of the group of detractors doing things the old way, similar to the "small web" or other counter cultures on the Internet. I fantasize about being here to pick up the pieces after all the others went full-on into AI-assisted everything and lost their critical thinking capacity, programming skills, knowledge of Unix command line, etc.
There is part of me that understands the appeal of the all-in on AI and personalized software approach. It's a bit cyberpunk! In terms of open-source software, the downsides outweigh the benefits in my opinion, though. Important principles like community ownership and commitment are absent, and this approach is even radically antisocial. And then there's the inevitable issues with maintainability, to say noting about dependence on big tech companies.
To each their own, but this is not for me.
I read somewhere (in the myriad blog posts dealing with this Cambrian LLM explosion) that software developers could be put into two camps: those that just want the thing to exist, and those that want to build and understand the thing (in addition to wanting it to exist).
those in the first camp are having a great time.
those in the second camp (which is how you're describing yourself, and how I'd describe myself) are wary and suspicious.
it is somewhat paradoxical, we've watched/read sci-fi/cyberpunk for years and dreamed of this kind of world. after all, when did you see any members of the Enterprise writing code? they just asked the computer to "write a subroutine" and that was that. what a world!
but here we are, with the craft in danger, not entirely impressed by the idea of "just ask and walk away".
i, too, fear for my loss of critical thinking, raw skills, and design sense, as do i think about being one of the few (in 2, 3, 5, 10 years) that didn't abdicate their cognition, their craft, to the tech overlords.
but i wonder if it will matter anyway. i wonder if "source code" will be a deep abstraction that nobody thinks about anyway, similar to how 99% of us don't care/need to care what the machine code we're eventually emitting does or looks like.
in any case, i'll keep my thinking for now.
I kinda like the woodworking analogy of this.
I, in theory, can plane a piece of wood with a hand planer. But I'll never do it again, we did it at school in ye olden times before the millennium and it was boring then as it is boring now.
I know people who get satisfaction from it, they take one sliver off with the hand planer, feel the wood with their hand and figure out the perfect angle for the next tiny sliver of wood to come out off, repeating the process over and over again.
I, personally, will just feed the damn plank to a mechanical planer with the exact specs of the resulting board set up. I just want the board smooth so I can get to the next step of the process. I'm not doubting the "wood-slop" the machine produces, I can see and measure if it's good enough or not. I don't need to be involved in the process.
We're both making a table, mine will be done faster. It might not be hand-crafted to perfection, but it will hold the stuff I intend to put on it just fine. If I find out it sucks later on, I can make a new one that's slightly better or fix the existing one. My goal was a functional product, not a piece of handcrafted art.
I don't think the analogy works. You're focusing on the "how", not the "what". Using a mechanical planer, you still need to dial in numbers yourself. You design your own table, the more modern tools just make it easier to realize your vision.
Another example: I enjoy writing with a good pen. But whether I write by pen or on a keyboard, it's still me writing it.
However, AI does basically all the real work, only leaving you to guide it. Make a table? AI gives you one with 2 legs. More legs? Guess I can live with 5 legs.
And you wouldn't be making that table, AI is. You cannot have pride in something that you never made yourself. It's the same as 3D-printing something from Thingiverse and claiming you made it.
People who create AI blog posts are not writing. Those that prompt their way to a piece of software are not doing software engineering. The ones that generate AI images are not being artists.
Your analogy is not really apples to apples though, is it?
More close is: if there was a table making machine, you just push a button and something like a table comes out, would you still be a woodworker? You haven't planed, nor measured, nor cut, nor jointed, you've only pressed on "make me a table"
I'm in the second camp.
Part of it's that the whole point of going into this industry is that I love coding and have been doing it since I was 8. Part of it is that I'm a control freak and it makes me uncomfortable to have to trust AI generated code. Sure, I already trust interpreters and compilers, but those are much more deterministic, and they don't generally do anything I have to be wary of. Part of it is that anytime I've used Claude to write stuff (using Opus 4.7 via an API key), I've had to handhold it when doing simple things (telling it repeatedly that a given column doesn't exist in Snowflake's task history table and eventually just giving up and taking it out by hand) and had to remove tons of completely pointless Python code it generates. The big difference is that the people in the first camp don't seem to care enough to check. Someone at my company used Claude to write 20k lines of code this past Friday. No way he read and scrutinized all of that in one day.
The other big thing I've noticed is that a lot of the people using it extensively seem to just be spitting out API endpoint after endpoint. Just doing endless CRUD with some light business logic. Yeah, it's not too hard to automate that with AI without any major issues. Hell, back when Ruby on Rails was hot, it was so fast to write those kinds of things with it that I could spin up things as fast as AI is doing now. Full websites or APIs in an hour or two because its syntactic sugar and scaffolding did what AI does with the FastAPI codebases I see these days. You could go from an ER diagram to a working app in minutes sometimes. I don't care that much if that kind of work is automated.
My experience is that often when I think "I wish my email / browser / calendar / … did X" it turns out be a limitation of the underlying protocol. So even if you build all software yourself, you still have to make compromises when you interact with the outside world.
I sometimes wonder what will happen when people who use rust run out of punny names for their programs that reference the concept of metal rusting...
We still have crab puns. Then we're screwed
This is really exciting.
Some of the folks who make things will go on to make things that suit not just their preferences but also those of a small audience.
Some of those audiences will go on to grow and grow and disrupt the big players.
The capital intensive part of software construction is melting away and being converted to opex (payg token costs and your time) and that will blast open the possibility space and lead to a massive new commons.
If the thing was so cheap to create why not open source it!
And if you like someone else’s open source thing but don’t want to take it wholesale why not give it to your agent and say “put the ideas from this onto my thing”!
It’s a new way of thinking about code too.
Absolutely and you're dead on thinking about the opening of the possibility space. The value of software as an enterprise will fall as we enter an age of abundant, and often custom, bespoke software. There will be many great apps coming and some lousy apps.
Another thing to watch for is how chatty the internet is about to become. A great many of these apps will hit APIs, ping each other, and so forth.
> The value of software as an enterprise will fall as we enter an age of abundant, and often custom, bespoke software.
Not just software. I'm predicting we'll be getting bespoke books and comics in a few years. TV and Movies after that.
Basically there will be a service with mad-libs style book skeletons and you can get your exact specifics put in with LLM writing in them.
You want a romance novel with dragons (D&D style), a red-headed princess protagonist that's a bad-ass fighter and she has three men competing for her affection, each with a very distinct look that YOU specifically like.
Done.
This already being done by actual human writers, many (not all!) people read books based on tropes they like, the rest doesn't really matter. They basically check the tags of the story and if they match well enough, they'll buy it and read it. And I mean very specific: https://www.goodreads.com/series/151379-ice-planet-barbarian... =)
Real nice to see two accounts using LLMs discuss nothing like this. They warned us this would happen, and here we are. I guess the topic is apt, but we used to customize windows (95+) and Linux like this (not down to vibe-coded insecure replacement of apps, but display/desktop/widgets/explorer/transparency of components).
Wait, are you implying I used an LLM to write my comment? Sorry if I'm misreading what you're saying.
I just realized I'm following the same philosophy. I use suckless tools and made so many modifications to st, dwm and such it feels like home. I'm currently in the process of implementing my own git manager so that it would nicely integrate with my workflow.
I’m inspired by the message.
On this software itself: I’d like to know how this feels to use. It’s so very lightweight. Does it feel categorically different to what we are used to?
One of the things I miss about the 1980s home computers is that they booted into a usable command line in a handful of seconds, from a few KB in ROM. Imagine what today’s HW could do if we’d retained that level of efficiency.
we are there now. depending on boot loader/os combination, one can get to the sub 1-5 sec range, if its cli-only.
Clearly havent seen what enterprise hardware is like these days... sure, the OS takes 5 seconds.. but the hardware can take 10 minutes in some cases now glares at hpe gen11 systems. Its seriously bad now. The amount of power and time backround hardware level tasks now take has significantly increased over the last 10 years. Even the ancient dell r710 i have sititng in a closet collecting dust boots faster than todays hp gen11's.
We waste a ton of energy on ineffeciencies in hardware and software today all because we managed to "just go faster".
It feels very different. It's all damn instant. Me happy.
That’s wonderful! I’ve made ultra-lightweight web apps of my own to replace bloated, slow, and poor UIs. It’s a night and day difference when the dependencies are few-to-none. And that’s on a fat browser stack. Your ASM desktop must zip!
Related tangent: https://smolmachines provides microvms with cold-start bootup times around 200ms, and a "pack" utility and format to create self-contained binaries. No affiliation, but I just discovered it a few days ago, sharing bc I find it kind of exciting.
I find this fascinating. I also like to customize my desktop experience with my own code, but it's more assembling stuff with some additional code as glue.
A word of warning: a reliable lock tool for X11 is difficult. You should look at XSecureLock, which uses a multiprocess approach to avoid leaving the desktop unprotected in case of crash. It also implements a number of countermeasure to ensure the desktop stays locked and the locker stays in the front of the display. It's small too, so easy to audit (but written in C).
Thanks. I'll look into it and borrow whatever is useful there into bolt.
> It used to be that writing your own editor, your own file manager, your own window manager, was a project of years.
This is... Not really true? Especially if you are writing just for yourself. These are week-long projects at most to get to a usable state, if you know what you're doing. This is why there are so many text editors and window managers in the first place.
This is nice, but is also leagues away from something you’re written yourself. Take LLMs out of equation, and you have piles of code that you barely recognise and barely can edit or tweak by yourself.
I dont think it matters at all to OP. Sidestepping the insult, it sounds like they very, very much want to tools that support their needs only, methodology be damned.
This just doesn't matter for a lot of us. We have LLMs that can tweak it and the tools work as intended.
The whole point of this sentiment is that the personal tools wouldn't EXIST due to the time sink needed.
The tradeoff makes sense for a lot of people even if it's not a good fit for you.
One term for these is "home-cooked" software.
https://maggieappleton.com/home-cooked-software
"Home" as in "phoning home"? Because Claude Code sure isn't local, privacy-respecting software.
What about the cybersecurity aspect of bespoke software?
A cybersecurity research company can now spend a small fortune on finding zero days in iOS because of the amount of people that use it. It basically guarantees there will be clients like government agencies willing to pay through the nose for the exploits.
Software made for one might disrupt this business model.
Why did you choose to have Claude write it in assembly language?
There are big benefits to using a language that has good static analysis with LLMs.
seriously.... we already have a constellation of good deterministic tooling for taking a relatively high concept spec to low level assembly. what does an llm offer in generating optimized asm that rust wouldn't??
Less memory footprint. No reliance on libs. Pure first-person control. No wasted CPU cycles is the target here for me. And if you read the post, the asm set is only for the desktop itself. The tools I use are in Rust. Result is: Laptop now runs at between 5-6W (down from ~9W) [XPS14 latest hw] on Ubuntu 26.04 - giving me around 3.5h extra battery life.
My guess is you're likely to waste more cycles on development time, and on suboptimal algorithms because the implementation is harder, than you would waste on rust-related bloat.
Still a cool project, thanks for sharing.
I have wondered about having LLMs output machine code directly and skipping the compiler/assembler altogether. Then you'd just commit your spec/prompt and run it through the LLM to get your binary.
> Less memory footprint. No reliance on libs.
rust can do that. You can run a hyper stripped down rust that was made for embedded devices specifically because those devices don't have room for a runtime.
I'm sure I can. The original challenge was more in line of "I wonder if CC can do this now?"
And it apparently can. And very well.
One advantage seems to be that the complete asm file fits easily into CC context window.
> The original challenge was more in line of "I wonder if CC can do this now?"
well, I can respect that for sure
+3.5h extra battery life is a real measurable result! well done.
Respect. I want to do the same thing. I'm studying electronics, PCBs and CAD in order to build the literal computer that I've always wanted as well. It's my lifetime project. Who knows if I'll ever succeed? I think AI significantly upped my chances but still.
Some screenies and the code at 0…
I struggle to understand why, though.
0: https://github.com/isene/chasm
Same reason people muck about with knowledge management systems... to put off the day when you have to sit down at your desk and actually do something.
everyone is finally coming around to emacs way of doing things after all these years :D
AI generated blog. Boring…
The agent sessions (traces) would be very educational too.
Would it be possible to share the jsonl files too, like how Mario Zechner shared his chats with the AI, while working on his Pi coding agent?
https://x.com/badlogicgames/status/2041151967695634619?s=46
That would be a huge payload with a few thousand prompts...
Even better!
I feel like build vs buy is the conversation now. I’m not a developer but I’ve built agents I use daily. When most people can vibe code their way to a custom app, value will most likely hinge on support and other “services”. Just my 2 cents, feel free to tell me I’m wrong!
I use code that hooks into existing programs so that I can customize the existing programs to what I want
Did you have to look or review any of the code produced, to get the performance/capabilities that you wanted, or were all interactions through CC? In other words, did you hit any walls with the pure agentic workflow?
I monitored the process very tightly. I have programmed my fair share of asm (and some 30 other programming languages), but for this I did not read any code. I hit lots of obstacles on this road, lots. In the process we also created a complete TTF rasterizer on par with what kitty gives me, and that was a true deep-dive.
Thanks for explaining!
Note that Rust is not in fact named after Fe2O3; it’s named after a resilient fungus of the same name
Just goes to show how little I know about Rust.
Is this an advertisement for Claude Code? It sure seems like it.
The fact where all posts of this blog make sure to state that claude code is a life changer, i wouldn't doubt about it
They all are.
If it is, it’s not a good one. I just thought “You moron”.
I started porting TurboVision from C++ to mojo with Claude as a sort of joke. Then I said, why not port the entire Turbo editor? And then like. hmm.. it doesn't have all the modern affordances like language servers, syntax highlighting for all languages, debuggers for everything, real-time find-in-files, projects, etc. Well now I have all this implemented: https://github.com/boxed/TurboKod
Not sure I can use it as a daily driver yet, but it would be pretty cool!
Finally the great solipsism singularity collapse can be mine!
ok seems a lot of fun (for those like-minded), but who seriously want to be dealing with maintenance of everything they use in the long term, in pure assembly all the more?
I was inspired by Nicklaus Wirth when he wrote his own language opearting system and CPU. LLM allow many more to do this like you so am seriously impressed. It is also fun.
So how productive are you now vs. before? I assume this was the reason for doing this?
I think it's more like gardening.
I would think the reason was to enjoy using their system as much as possible.
Is there a demo video? An installer or pre built package or ISO?
Did OP write this by hand? It reads like language written by a human overfitted on GPT 4o or Claude.
OP did this: Prompted CC for all the points I wanted included (something like a 200 word prompt) and asked CC to draft it, including all the links added to the table I furnished. Then I edited the draft (about 50% then edited). Then asked CC to spellcheck and fixed the 5 it found.
Thank you. It would have been nice to see you personalize the hook and show your storytelling voice the way you personalized your computer in the story, but we aren't all poets.
If they basically generated a desktop for themselves, what’s the chance they didn’t generate the article? I think pretty slim.
Also, reading it is probably not the intended use. It’s probably: “Hey Claude, give me a TLDR of this”
Who cares? It’s their content. If they hired an editor to help them, cool. If the content doesn’t suit you, move on.
But the incessant “AI was used here, thus is it garbage” is long past time to enter the grave.
Many people care, with good reason. We learned to notice LLM-isms is because they are, in fact, a very strong predictor that a piece of text is in fact garbage that's not worth your time reading.
I usually stop reading at the first LLM-ism, but I found the premise of this post interesting enough to keep going - and guess what, the entire article was literally just "I prompt CC to make software tailored for me" blown out to 8 sections.
> Who cares?
The parent comment does. Why do you care that they care?
^^ some anti-luddism right there
Luddism is when you don't bother reading what someone didn't bother to write.
Dictionaries on the second row down that aisle.
Is that supposed to mean something?
Not a good look to be here fighting in the comments when you could have been writing something that reads like anything other than Claudeslop.
I agree, yeah
AI slop, pretty sad.
Programmers turning into mindless slop feeders.
Instead of replacing, downsize. Do you really need a GUI for git? Do you need the JetBrains Smart Ultra Analysis Pro™? Do you even need syntax highlighting? How far can you go?
Its fun, and a lot more rewarding than replacing tool X with tool Y, realizing you actually hate it.
ORDERS OF MAGNITUDE HAHA HE SAID THE THING
I think this is going to be the OS of the future. You tell the computer what you want to do, and it uses the OS's APIs to create your program for you. No more copilot embedded in notepad unless that what you ask for.
Most software is done after the first or second version and the developers just keep working on it to justify their job; adding features no one needs and just get in the way or make the program worse. It'll be nice when the software I have does exactly what I need and doesn't change until I tell it to change for something I need.
The only feature Macos has shipped in the past 10 years that I actually like is air-drop. Everything else is a PITA annoyance, or as I've found out from upgrading, just bug ridden slop that doesn't work well anymore.
Also open source tools can be “hacked” to enable features you want.
I do this because I don't want to rely on others for my basic needs.
https://github.com/nurettin/gnome-extensions-topterm
https://github.com/nurettin/gnome-extensions-toptop
Great job! This is super cool and probably feels incredibly satisfying!
I do find it curious how even after replacing all of your software, but are still using Claude Code instead of building your own coding agent.
Hmmm
I've been building an object oriented system re-imagined in a world with LLMs called Abject (https://abject.world) and one thought I had was to build an OS that boots into my project. One way to do it would be a minimal linux distro (think firefox os or similar). Has anyone done something like this with their projects?
> Nor do I have to write documentation for users who don’t exist
Brother mine, you will learn that the future you is ignorant of all the things, and every bit of documentation goes a long way
Haha, it's funny that we've all reached the same conclusion. I, too, believe in the same idea[0][1]. What is fascinating to me is how many things can now be elided from software. I don't use configuration files or things like that. I can simply hardcode everything in because there is only one user. If I want to configure it the other way, I just modify it and rebuild it.
The other thing is that other people's applications are rarely useful. Their libraries are, the feature description READMEs are, but the software itself is full of attempts at generality that make them overly annoying for me to use. Instead I have extremely idiosyncratic software - anyone else would find it insufferable.
The wild thing, though, is that my software is outrageously useful for me. I can see why Anthropic and OpenAI are (or shortly will be) the trillion-dollar behemoths they are. They are enabling a personal productivity increase of epic proportions[2]. The highly specific functionality also means strange things performance wise. I don't need to use Electron or Tauri or whatever. Instead, my thing is Rust with objc2 and it starts instantaneously. On my M1 Max, it's the fastest text viewer I can start. 100s of megabytes of JSON and it's launching is imperceptible for my tool, pretty-printing is instantaneous, breadcrumbs are live.
Because I can make it do only the thing I want it to do. It can't do other things. I cannot edit or auto-complete or anything. And this is great. Useless to others and fantastic to me.
Likewise, my blog is on Mediawiki (which I like so anyone can edit) but the authoring flow is kind of annoying. Uploading images causes a break from writing, and requires a lot of form-filling that interrupts my thought. So I now have this software that does everything I want: link autocompletion, background image uploads, post-hoc publishing, previews and diffs, built-in Wikipedia search to interwiki link. Who would want this but me? It only brings me pleasure.
What a revolution in software.
0: https://wiki.roshangeorge.dev/w/Blog/2026-04-25/The_rise_of_...
1: https://wiki.roshangeorge.dev/w/Blog/2026-04-30/Personal_Sof...
2: Predictably, I have chosen to use the spare time on leisure
Cool reply & thanks for the links.