For better or for worse, when everyone is a "potential software founder" nobody is because your potential customers can just use AI the same way you did.
I've more or less accepted this, and I think my future is in making software more resilient, secure, and fault tolerant. These people will likely want to scale these solutions up, tie different solutions together, and generally make their lives increasingly difficult. Often without realizing it.
My experience is that Claude starts to make quite a mess in this context, and it'll often cause as many problems as it solves unless you have the technical and domain knowledge to redirect and correct it frequently. Perhaps training will solve this, and it'll certainly get better, but I'm not sure how far it'll go and how fast.
My gut feeling is that software will only become more ambitious and interface with hardware and other systems in increasingly sophisticated ways. Things that seemed infeasible due to time and cost constraints will be on the table. It'll reveal new challenges, I think. I have a feeling it'll be humans with deep technical skills who are at the forefront of solving those challenges for a while yet.
Not claiming I have the skills and to be one of those people, just that it's where I'm pushing my career at the moment.
I'm stoked that people like this have the resources and newfound capabilities to create solutions like this. The reality is that previously, many people have been underserved due to the economics of software and inherent risks of trying things like this as a smaller business owner. So this is great. We can find more ways that software can be valuable, and people can do their jobs better in ways they've literally only imagined before. It's great.
What could happen is a reduction in the amount of programs used, with a smaller set of more sophisticated programs doing more work. This maps to what we saw pre-industrial revolution: lots of small family operations doing menial manufacturing work (woodworking, textiles, cooking). This got replaced by large factories. A smaller amount of companies producing the a larger volume of goods. With AI, a smaller group of engineers could handle more local complexity, thus allowing more sophisticated, general purpose software to be created, deleting the sea of small pieces of software we have today.
Will this means many will be jobless? No, they would do other things. They'd be using this software to support society, operating at a high level. Think low-code, but incredibly complex stuff; just not raw code anymore. Instead of making circuit boards out of descrete components, you now slap a few ICs on a board with some supporting passives and the work is then all done in software. Engineers use more high-level components rather then welding and machinijng things from scratch; you buy T-slot profiles and bolts rather than casting and milling steel from billets.
So the job of programmer may disappear simmilar to how we don't have bakers anymore, baking is done in factories, operated by a small staff. Current-day programmers will then increasingly shift to whatever high-level constructs we'll come up with, this high level work will be supported by the base infrastructure that those who still touch raw code will build.
I had to strike more of a balance between moving quickly and focusing strictly on resilience. The difference moving forward would be that the features already exist, and my job will increasingly be focused on ensuring they work properly, and the underlying systems supporting features are consistent, cohesive, and not duct taped together like an LLM built it.
It reminds me a lot of my early career spent remediating offshored PHP applications.
Are we sure Claude Scale™ won’t appear next month? A specialist agent that turns your vibe coded mess into a production grade scaled solution on their infrastructure.
Expect anthropic to want to capture more of the supply chain over time
scaling these solutions will prove to be counter productive. If you are thinking of scaling you are still trapped in the current paradigm. This plumber guy is only unique in that he read the news and pushed a little harder to see what the actual f is going on. In the days that come, every single person who is serious about their job will do and experience the same thing.
I'm not saying this particular individual is wrong in trying to build his solution to the market. Maybe there is some VC money to be made in this moment. But as AI in the workplace gets normalised, most people will either come up with solutions for their problems, or they will ask someone they know to help them with this.
scale will only matter if you are explicitly building a platform. That will still require real software engineering skills.
As for hardware interfacing, if I am not mistaken, almost all companies selling hardware right now still behave like babies when it comes to users getting access to the software inside it. They void warrantees, sue them, so on etc. For ambitious user driven software innovations in the hardware space the companies should open up their interfaces. I don't see this happening at all not only because of the companies' greed but also for regulatory and safety reasons.
You assume that the agents can produce okayish software that isn't very resilient, but that's not quite what we're seeing. Remember Anthropic's attempt to have agents write a C compiler (which isn't a super-complicated task by any means). Despite preparation work that is well beyond what most software can have (specs available to the agent and that the model trained on, thousands of human-written tests available to the agent and that the model trained on, and a human-written reference implementation available to the agent and that the model trained on) the agents still failed to converge. In other words, under conditions that are so favourable as to be unrealistic for almost any kind of real software, the AI agents couldn't even produce so much a workable C compiler. And this is what I also see: the agents take the code in a direction that doesn't converge, let alone make it resilient.
"Mechanical engineer uses code to improve engineering process". Okay, this has been going on forever. Other engineering disciplines and various fields using software to solve problems. Programming doesn't exist in a vacuum of theory.
I think there are many moats that non-experts won't attempt to cross even with AI assistance.
For example, we've built in a lot of complexity to areas like authentication. And for good reason. It's like electrical code. I'd pay good money to watch a muggle attempt to configure OIDC infrastructure. Even with the AI explaining everything to you, it's too much information to digest at once. You'd need an entire afternoon just to wrap your head around the idea of asymmetric cryptography. That's a lot of time not spent doing the thing your business is actually about.
I absolutely love this, because to me, this is what software development should be about, solving actual problems and providing faster calculations, improving the workflow for people.
It does strike me as a little odd that they didn't hire a developer earlier and got the code written. Sitting back and waiting for someone to drop by and present a solution is a little naive, but it's also the world we built in the IT industry over the past 20 years. When I started my first job, we frequently had customers ask for bespoke solution, most of which was small one week to a few months of work. Multiple co-workers in the mid 2000s has side businesses, where they did contract development, most of which was these types of small one off solutions. Most of the software companies, in my area, that did these types of jobs are all gone now.
If AI accidentally created an environment where people can once again solve small programming problems on their own and massively improve the workflows I'm all for it. Serves the industry right for abandoning these customers.
In my experience, I've worked with a number of people in non-tech industries who tried to pivot their company into software, and a huge obstacle was that ultimately the code became unmaintainable.
Maybe they didn't have the expertise to pick a software stack that would serve them in the long run, or they just didn't have the budget to hire a SWE or team full time, or their contractor team just wasn't super invested in the project.
So tech people look at "vibeslop" as unmaintainable technical debt, but they ignore that in a lot of situations their own salary is what makes the tech debt unmaintainable. Maybe that's uncharitable, but I do think many techs are very far removed from the "solve a problem and then dogfood it" cycle
It's a really interesting case study, but the summary seems to lean into the AI hype to an extent that borders on lying.
> His fabrication shop uses it daily, and he built the entire thing in 8 weeks. During those 8 weeks he also had to learn everything about Claude Code, the terminal, VS Code, everything.
I don't see how he can give this summary with a straight face after posting the interview that CLEARLY contradicts it.
In the interview the engineer says "When Claud Code came out almost a year ago, I started dabbling with web based tools ..." and "When it first came out I had so many ideas and tried all these different things", so he had clearly already used extensively it for a year. I would also guess the engineer was somewhat technically minded from the get-go, since he claims he was "really good with excel" before starting with Claude Code, but that is beside the point.
The interviewer later asks "How much of those 8 weeks was learning Claude Code versus actually building the thing?", and the interviewee answers "Well, I started Claude Code when it first came out so the learning curve has really gone down for me now..." and then trails off to a different subject. Which further confirms that the summary in the post is false.
It really seems like the engineer has spent the year prior learning Claude Code and then spent 8 weeks on solely building this specific application.
The interviewer also claims "This would normally have taken a developer a year to build", which seems really unsubstantiated. It's of course hard to judge without all the details, but looking at the short demo in the video, 8 weeks of regular development time from a somewhat experienced developer doesn't seem too far fetched if the objective is "don't make it pretty, just make it work".
As I said, it's a really interesting case study about a paradigm shift in how software is developed, and it's clear this app would never have existed without Claude Code. So I don't really see the need for the blatant lying.
I feel both great and awful about this. For over a decade I’ve said that nearly anyone that uses a computer could benefit from some programming understanding. A little bit can go a long way to solving problems like this. Problems that collectively slow down and block the ambitions of a huge number of people worldwide.
But instead we’ve found a way to circumvent the process. Losing the understanding of your own problem and the new ideas that come off the back of it.
I’m reminded of the story that NASA had a research project to make pens that would work in space, and Roscosmos just used pencils. I always thought NASA came off worse in that anecdote, but I wonder what they learnt while making the pen…
The guy from the story, it’s just another developer starting from a different trade, pretty normal across our history, musicians, lawyers that discovered that they were good at computers. The conclusion is flawed, not anyone can endure what this person did, sit at a terminal, going back and forward until something is finished. That’s what a SW dev does. My conclusion, many more people will discover that they are good at software, not everybody, but some of them will discover this new powers, thanks to a new lower barrier provided by LLM.
People like this will create a net increase in software jobs. Once his software makes enough money so he doesn't have to sit in front of a computer, he will employ someone. It will initially be a gig fixing slop. https://www.slater.dev/2025/09/about-that-gig-fixing-vibe-co...
People in the trades have a ruthless pragmatism that SV has forgotten.
I think it's awesome that AI is enabling this. I think the the future of software engineering is in helping make this kind of thing resilient and removing the fragility that AI generated code always seems to inject
Trades / SWE overlap is interesting as I think they are analogous to each other.
I went to college with a lot of actual engineers - mechanical, electrical, chemical, etc. In those fields you are designing products and then engineering processes to output a cog of some sort (drug, car, GPU, iPhone, etc) in the thousands to millions.
In our fields as SWEs, a lot of our job it's like the trades going into a house to install HVAC, fix a burst pipe, upgrade a circuit breaker, replace a furnace, etc. No two setups are exactly alike, no requirements are exactly alike, etc.
Even in the age of LLMs I think the industry remains more artisanal than engineering. And that's not a knock on us, I think it's because what we do is essentially automate business processes.. and no two businesses are alike. I don't think LLMs replace the role, it just makes parts of our job faster. The mindset of how you automate something doesn't generally exist in the minds of people who want the automation.
I didn’t read the article but if he is really an actual mechanical engineer then this is not that interesting. Mechanical engineers build a lot of technical software. Most fluid dynamics code or vibration code is probably written by mechanical engineers. Now, some times people say “mechanical engineer” but they don’t mean someone that studied FFTs.
As part of my "well shit what's next" arc I've been checking out machine shops in my area. By and large they are dumb manual 30 year contracts or highly automated job shops..
And the owners of those job shops aim for 3 shifts per worker via automation, and mash their own software with AI already. They are ruthless at cost cutting and automation and AI tools are perfect for them.
Unfortunately the vibe I get talking to them is essentially a triumphant "why would I need you, I have AI" or "yeah you're screwed".
I can't blame them for being served expensive barely functional crap SaaS or ERP software for ages, but I was not expecting to be viewed as part of the problem coming from a robotics, automation, and optimization background myself. It's just all a block of overpaid swindlers to them.
Software development usability has always been measured in mean time before black-box failure: something misbehaves that the person can't fix or understand.
LLM's shorten that time for every application and every user, but particularly for users from professions that haven't built modeling or debugging skills because they rely on physical reality - like pipes fitting or process supervision - to weed out non-performers.
Hiring for LLM-enhanced work should focus on debugging skills in unknown situations.
Here's what I find difficult to reconcile with my own experience. I've been using Codex in anger for the past 2 months or so (with gpt5.3-codex and then gpt5.4) on projects of different complexity. It is quite good at debugging, but the (non-trivial) code it produces is really bad. And I don't mean bad stylistically, but bad in the sense that Codex clearly won't be able to maintain it for long (which is how Anthropic's C compiler experiment failed) because it uses an approach of "success at all costs" where it always prefers fixes that treat the symptom rather than going back and rethinking the architecture as features are added.
So the options are: 1. the program involved here is really trivial, 2. it hasn't evolved long enough for the agent to fail at evolution, or 3. others are not seeing what I'm seeing.
It is weird. I've had maintainable solutions on non-trivial code, but it does kind of require babysitting. Planning documents and detailed specs help. You get a feel for where the agent will want to take a shortcut and can devise ways to navigate around that.
I also find Go works really well, and generally stays, if not exceptional, than at least maintainable.
I've also enjoyed using OCaml, but I will say that I found the single worst function I've ever seen in a codebase in vibecoded OCAML.
You might just try asking - "hey I'm having trouble keeping maintainable codebases - how can I structure this project in a way where the code will be stable long term".
Sometimes getting the "software architect" role into the agent context is all it takes.
For decades, employees have been developing tools using whatever was available to them, and in most cases, this was limited to Excel macros.
AI provides access to much better tools for testing and quickly experimenting with new ideas.
The only ones who should be worried are companies that charge millions for four junior developers and an agile coach, and deliver more PowerPoints than code (I’m looking at you, Capgemini).
39 comments
[ 26.1 ms ] story [ 1361 ms ] threadMy experience is that Claude starts to make quite a mess in this context, and it'll often cause as many problems as it solves unless you have the technical and domain knowledge to redirect and correct it frequently. Perhaps training will solve this, and it'll certainly get better, but I'm not sure how far it'll go and how fast.
My gut feeling is that software will only become more ambitious and interface with hardware and other systems in increasingly sophisticated ways. Things that seemed infeasible due to time and cost constraints will be on the table. It'll reveal new challenges, I think. I have a feeling it'll be humans with deep technical skills who are at the forefront of solving those challenges for a while yet.
Not claiming I have the skills and to be one of those people, just that it's where I'm pushing my career at the moment.
I'm stoked that people like this have the resources and newfound capabilities to create solutions like this. The reality is that previously, many people have been underserved due to the economics of software and inherent risks of trying things like this as a smaller business owner. So this is great. We can find more ways that software can be valuable, and people can do their jobs better in ways they've literally only imagined before. It's great.
Will this means many will be jobless? No, they would do other things. They'd be using this software to support society, operating at a high level. Think low-code, but incredibly complex stuff; just not raw code anymore. Instead of making circuit boards out of descrete components, you now slap a few ICs on a board with some supporting passives and the work is then all done in software. Engineers use more high-level components rather then welding and machinijng things from scratch; you buy T-slot profiles and bolts rather than casting and milling steel from billets.
So the job of programmer may disappear simmilar to how we don't have bakers anymore, baking is done in factories, operated by a small staff. Current-day programmers will then increasingly shift to whatever high-level constructs we'll come up with, this high level work will be supported by the base infrastructure that those who still touch raw code will build.
I find that scalability is usually overblown because computers are fast now, which is not to say you shouldn't make it run fast on one computer.
It reminds me a lot of my early career spent remediating offshored PHP applications.
Expect anthropic to want to capture more of the supply chain over time
I'm not saying this particular individual is wrong in trying to build his solution to the market. Maybe there is some VC money to be made in this moment. But as AI in the workplace gets normalised, most people will either come up with solutions for their problems, or they will ask someone they know to help them with this.
scale will only matter if you are explicitly building a platform. That will still require real software engineering skills.
As for hardware interfacing, if I am not mistaken, almost all companies selling hardware right now still behave like babies when it comes to users getting access to the software inside it. They void warrantees, sue them, so on etc. For ambitious user driven software innovations in the hardware space the companies should open up their interfaces. I don't see this happening at all not only because of the companies' greed but also for regulatory and safety reasons.
For example, we've built in a lot of complexity to areas like authentication. And for good reason. It's like electrical code. I'd pay good money to watch a muggle attempt to configure OIDC infrastructure. Even with the AI explaining everything to you, it's too much information to digest at once. You'd need an entire afternoon just to wrap your head around the idea of asymmetric cryptography. That's a lot of time not spent doing the thing your business is actually about.
It does strike me as a little odd that they didn't hire a developer earlier and got the code written. Sitting back and waiting for someone to drop by and present a solution is a little naive, but it's also the world we built in the IT industry over the past 20 years. When I started my first job, we frequently had customers ask for bespoke solution, most of which was small one week to a few months of work. Multiple co-workers in the mid 2000s has side businesses, where they did contract development, most of which was these types of small one off solutions. Most of the software companies, in my area, that did these types of jobs are all gone now.
If AI accidentally created an environment where people can once again solve small programming problems on their own and massively improve the workflows I'm all for it. Serves the industry right for abandoning these customers.
Maybe they didn't have the expertise to pick a software stack that would serve them in the long run, or they just didn't have the budget to hire a SWE or team full time, or their contractor team just wasn't super invested in the project.
So tech people look at "vibeslop" as unmaintainable technical debt, but they ignore that in a lot of situations their own salary is what makes the tech debt unmaintainable. Maybe that's uncharitable, but I do think many techs are very far removed from the "solve a problem and then dogfood it" cycle
bullshit story always leave something like this.
> His fabrication shop uses it daily, and he built the entire thing in 8 weeks. During those 8 weeks he also had to learn everything about Claude Code, the terminal, VS Code, everything.
I don't see how he can give this summary with a straight face after posting the interview that CLEARLY contradicts it.
In the interview the engineer says "When Claud Code came out almost a year ago, I started dabbling with web based tools ..." and "When it first came out I had so many ideas and tried all these different things", so he had clearly already used extensively it for a year. I would also guess the engineer was somewhat technically minded from the get-go, since he claims he was "really good with excel" before starting with Claude Code, but that is beside the point.
The interviewer later asks "How much of those 8 weeks was learning Claude Code versus actually building the thing?", and the interviewee answers "Well, I started Claude Code when it first came out so the learning curve has really gone down for me now..." and then trails off to a different subject. Which further confirms that the summary in the post is false.
It really seems like the engineer has spent the year prior learning Claude Code and then spent 8 weeks on solely building this specific application.
The interviewer also claims "This would normally have taken a developer a year to build", which seems really unsubstantiated. It's of course hard to judge without all the details, but looking at the short demo in the video, 8 weeks of regular development time from a somewhat experienced developer doesn't seem too far fetched if the objective is "don't make it pretty, just make it work".
As I said, it's a really interesting case study about a paradigm shift in how software is developed, and it's clear this app would never have existed without Claude Code. So I don't really see the need for the blatant lying.
But instead we’ve found a way to circumvent the process. Losing the understanding of your own problem and the new ideas that come off the back of it.
I’m reminded of the story that NASA had a research project to make pens that would work in space, and Roscosmos just used pencils. I always thought NASA came off worse in that anecdote, but I wonder what they learnt while making the pen…
People in the trades have a ruthless pragmatism that SV has forgotten.
https://www.slater.dev/2025/08/oil-spills-can-create-jobs/
I went to college with a lot of actual engineers - mechanical, electrical, chemical, etc. In those fields you are designing products and then engineering processes to output a cog of some sort (drug, car, GPU, iPhone, etc) in the thousands to millions.
In our fields as SWEs, a lot of our job it's like the trades going into a house to install HVAC, fix a burst pipe, upgrade a circuit breaker, replace a furnace, etc. No two setups are exactly alike, no requirements are exactly alike, etc.
Even in the age of LLMs I think the industry remains more artisanal than engineering. And that's not a knock on us, I think it's because what we do is essentially automate business processes.. and no two businesses are alike. I don't think LLMs replace the role, it just makes parts of our job faster. The mindset of how you automate something doesn't generally exist in the minds of people who want the automation.
Tbh this is nothing new; we knew technical people with Claude code would be able to program well enough that tbey would be business developers.
Most engineers have to take at least one programming class in college.
And the owners of those job shops aim for 3 shifts per worker via automation, and mash their own software with AI already. They are ruthless at cost cutting and automation and AI tools are perfect for them.
Unfortunately the vibe I get talking to them is essentially a triumphant "why would I need you, I have AI" or "yeah you're screwed".
I can't blame them for being served expensive barely functional crap SaaS or ERP software for ages, but I was not expecting to be viewed as part of the problem coming from a robotics, automation, and optimization background myself. It's just all a block of overpaid swindlers to them.
LLM's shorten that time for every application and every user, but particularly for users from professions that haven't built modeling or debugging skills because they rely on physical reality - like pipes fitting or process supervision - to weed out non-performers.
Hiring for LLM-enhanced work should focus on debugging skills in unknown situations.
So the options are: 1. the program involved here is really trivial, 2. it hasn't evolved long enough for the agent to fail at evolution, or 3. others are not seeing what I'm seeing.
I also find Go works really well, and generally stays, if not exceptional, than at least maintainable.
I've also enjoyed using OCaml, but I will say that I found the single worst function I've ever seen in a codebase in vibecoded OCAML.
You might just try asking - "hey I'm having trouble keeping maintainable codebases - how can I structure this project in a way where the code will be stable long term".
Sometimes getting the "software architect" role into the agent context is all it takes.
AI provides access to much better tools for testing and quickly experimenting with new ideas.
The only ones who should be worried are companies that charge millions for four junior developers and an agile coach, and deliver more PowerPoints than code (I’m looking at you, Capgemini).