Compared to my salary, the current cost of the models and tokens to do the work I normally would, is around 10%-25% of it.
Obviously, you still need someone to prevent the models from going insane and messing everything up, but in my experience (webdev projects, DevOps stuff, local software, well known domains), it is very much a force multiplier, as long as you acknowledge that you really need tests and various prebuild scripts.
So I predict one of two things happening:
A) de-valuation of software development work in well-explored domains (and perhaps some changes in regards to outsourcing, as long as cultural and communication differences can be compensated for); with the implications for those learning programming now
B) the squeeze coming in the other direction, making inference 3-5x more expensive, though maybe not with how every big org out there is trying to be a loss leader
Either way, it's an interesting direction - instead of ever becoming "proper" engineering (outside of RFCs and foundational stuff), we went from React/Vue/Angular/Svelte/Express.js/Laravel/Django/Rails/ASP.NET/Spring wild west and frameworks of the day (never being able to nail down what "good practices" are and stick to them for decades, but chasing the new thing forevermore), to even closer to producing non-deterministic slop, except that the slop kinda sorta works. Wild times.
But it is true, the cost is effectively zero. There will be, for a long time, free models available and any one of them will give you code back, always!
They never refuse. Worst case scenario the good models ask for clarification.
The cost for producing code is zero and code producers are in a really bad spot.
I beg to differ. Let's say you're right. Code producers should turn to agriculture and let their managers and product owners prompt AI to produce code. How about code maintainers? Ever heard the mantra "You build it, you run it"? Lets say that AI can build it. Can it run it though? All alone, safely, securely and reliably? No. It can't. We can keep dreaming though, and when will AI code production services turn profitable? Is there a single one which turned profitable?
Agents are monitoring metrics and logs.
A bug is introduced into the system.
Error rates go up and the agents notify diagnostic agents.
These agents look at recent commits and figure out the problem.
They instruct another agent about how to fix the issue and deploy the change.
The problem is fixed before an engineer even has time to start looking at logs.
If you aren’t seeing this, you’re not keeping up with what others are already doing. It’s not just people vibe coding ToDo apps.
Calm down buddy, maybe you're confusing code producers with something else. It's 2026 we don't bother with maintenance no more, we /new to keep context clean and start over.
Just don't forget to comment - never delete - old code. Always keep dead code around to please shareholders, line numbers up always.
We produce code, that is the main thing, never forget.
One could argue we could achieve the same goals by appending \n to a file in a loop, but this is inefficient nowadays with generous token offerings (but could change in the future than I highly suggest just outputting \n to a file an call it productivity increase)
I didn't understand your point about product owners. Who the fuck would ever need one when code produces itself?
Zero as long as your time is worth nothing, and bad code and security issues cost you nothing maybe.
"Getting code" has always been dead simple and cheap. Getting actually good code that works and doesn't turn into a problem for you down the road is the expensive part
i can't remember who said it but a long time ago i remember reading "Linux is free if your time is worthless". Now we all use Linux one way or the other.
That's still very much true, but at least in the case of Linux the cost is getting lower and lower all the time. The time investment for many has reached about the same as the cost needed for Windows and as a result we see more and more people using linux. At this point it's a perfectly viable gaming platform!
Maybe one day LLMs will eventually make good code at a low cost, and that will allow non-programmers to write programs with few problems but the cost will never be zero, and I think we're a long long way from making human programers obsolete.
All of the intelligence that LLMs mimic came directly from the work of human minds which got fed into them, but what LLMs output is a lossy conversion filled with error and hallucination.
My guess is that the LLMs producing code will improve for a
short time, but as they start to slurp up more and more of their own slop they'll start performing worse.
I am thinking about this a lot right now. Pretty existential stuff.
I think builders are gonna be fine. The type of programmer were people would put up with just because they could really go in their cave for a few days and come out with a bug fix that nobody else on the team could figure out is going to have a hard time.
Interestingly AI coding is really good at that sort of thing and less good at fully grasping user requirements or big picture systems. Basically things that we had to sit in meetings a lot for.
This has been my experience too. That insane race condition inside the language runtime that is completely inscrutable? Claude one-shots it. Ask it to work on that same logic to add features and it will happily introduce race conditions that are obvious to an engineer but a local test will never uncover.
I’m not convinced. That sort of thing usually depends on some very specific arcana or weird interaction between systems that is not in the code. It usually requires either external knowledge or deep investigation and compilation of evidence from multiple sources. I haven’t seen AI do that much.
Look at recent examples of browsers and matrix servers. AI can’t even follow extremely detailed specs with extensive test suites.
If anything, nice and friendly but mediocre devs are in more immediate danger than rough but extremely competent devs.
But we’ve seen C-suits losing institutional knowledge at a drop of a hat for decades so who knows? Maybe knowledge and skill are not that valued.
> The type of programmer were people would put up with just because they could really go in their cave for a few days and come out with a bug fix that nobody else on the team could figure out is going to have a hard time.
meetings hardly reach anywhere. most of the details are eventually figured out by developers when interacting with the code. If all ideas from PMs are implemented in a software, it would eventually turn into bloatware before even reaching MVP stage.
> The type of programmer were people would put up with just because they could really go in their cave for a few days and come out with a bug fix that nobody else on the team could figure out is going to have a hard time.
This is the exact type of programmer that isn't going to have any issues - ones who actually know what they're doing and aren't just going to vibecode react slop.
Not really, in my experience you still have to be good at solving problems to use it effectively. Claude (and other AI) can help folks find a "fix", but a lot of times it's a band-aid if the user doesn't understand how to debug / solve things themselves.
So the type of programmers you're talking about, who could solve complex problems, are actually just enhanced by it.
With all due respect to the author, this is a lot of words for not much substance. Rehashing the same thoughts everyone already thinks but not being bold enough to make a concrete prediction.
This is the time for bold predictions, you’ve just told us we’re in a crucible moment yet you end the article passively….
I have a theory: I think the recent advance in coding agent has shocked everyone. It's something of such unheard-of novelty that everyone thinks they've discovered something profound. Naturally, they all think they are the only ones in on it and feel the need to share. But in reality, it's already public knowledge, so they add no value. I've been in this trap many times in the last couple years.
- Small companies using AI are going to kick the sh*t out of large companies that are slow to adapt.
- LLMs will penetrate more areas of our lives. Closer to the STTNG computer. They will be agents in the real life sense and possibly in the physical world as well (robots).
- ASICs will eat nVidia's lunch.
- We will see an explosion of software and we will also see more jobs for people who are able to maintain all this software (using AI tools). There is going to be a lot more custom software for very specific purposes.
> Small companies using AI are going to kick the sh*t out of large companies that are slow to adapt.
Big companies are sales machines and their products have been terrible for ages. Microsoft enjoys the top spot in software sales only due to their sales staff pushing impossible deals every year.
Well, I made my predictions. Let's come back in a few years.
Netscape didn't attack Microsoft's business software, operating systems or other pieces of their offerings.
Google also didn't seriously attack Microsoft's business.
And neither had the capability to build large software very fast.
Google is both a software company and an infrastructure company as is Microsoft today. Their software is going to become more of a commodity but their data centers still have value (even perhaps more value since all this new software needs a place to run). It's true that if you're in the business of hosting software and selling SaaS you have an advantage over a competitor who does not host their own software.
> Netscape didn't attack Microsoft's business software, operating systems or other pieces of their offerings.
That's not how it was interpreted at the time: Netscape threatened to route around the desktop operating system (Win32) to deliver applications via the browser. "Over the top" as they say in television land.
Netscape didn't succeed, but that's precisely what happened (along with the separate incursion of mobile platforms, spearheaded by Apple... followed quickly by Google, who realised they had to follow suit very quickly).
> And neither had the capability to build large software very fast.
Here is my bold prediction: 2026 is the year where companies start the lay offs.
2026 is the year where we all realise that we can be our own company and build the stuff in our dreams rather than the mundane crap we do at work.
Honestly I am optimistic about computing in general. Llms will open things up for novices and experts alike. We can move into I the fields where we can use our brain power... But all we need is enough memory and compute to control our destiny....
I don't know, its a bit of a hellscape in tech right now as thousands of people with deep domain knowledge and people knowledge and business knowledge (ie experienced engineers managers and product owners), were laid off by C Suites desperate to keep the AI funded mandates going
Do you know how hard it to make a successful company or even make money? Its like saying any actor can goto hollywood and be a star
Yes it is. But I am an optimist for human nature. I personally believe smaller companies doing different things is the future... Scaling as they need. It is a hellscape but people can and will adapt.
> Do you know how hard it to make a successful company or even make money?
Yes I have failed to do it before. I get this.
> VCs wont fund everyone
And? Do you need VCs? Economics mean that scale matters but what if we don't need it. What if we can make efficient startups with our own funding??
But heres the reality from me- I’m in my 50s and I don't have it in me to grind at the level of 20 year olds to achieve some level of security in an untried business model - and this is someone who has launched 2 AI startups in the past 2 years
In one we got VC funding but I left after setting up their agent platform and tons of AI assisted coding only to not meet impossible deadlines and over-promised AI value to enterprise customers - I was literally working 20 hr days at stretch for 170k salary competing and half benefits against 25 year olds out of stamford with no lives- far lower than the 250+ with stock and benefits I got in my EM role at a big company which now evaporated - I was edged out of that startup role for not delivering “on time”
My second AI startup I cofounded with friends and trusted colleagues its bootstrapped (all of us are over 40) - so we have more experience and better deadlines now but its up to the business gods on how well it will do- crossing my fingers
But its a lot of pressure for sure and I currently have no health insurance and my wife was laid off in December and we lost her benefits
So I wouldn’t call myself optimistic in the end stage capitalistic hellwhole that is modern “middle class” America
I hope a better work model can be found - but having some any salary and medical benefit security would be nice
I went to an AI meetup last week and it was filled with gray hairs - i could sense the desperation as many people I met told me they were laid off recently and trying to dive in
Ironically looking at them it reminded me of those interviews they used to do in Appalachia or something when a town was out of work and advisors came in and said “learn to code instead of mining” (ok I may be exaggerating somewhat but I even know a ex Microsoft manager who had to resort to a go-fundme to keep his family afloat)
I own (cofounded) a medium-sized saas business with hundreds of employees. I maintain final decision of everything technical and still code every day because it’s important. All of the engineers use LLM tools, you’d be stupid not to. But I need good engineers, I replace good engineers when someone leaves, and the business itself is so much larger than just programming. The system is just so huge and complex, and I am the benevolent dictator that architected it and maintains the core design decisions, that the LLM does not replace the need for engineers nor my own expertise.
Furthermore if we were truly in the utopia the author describes, why do all the LLM companies employ (and pay top dollar) for so many engineers? Why does OpenAI pay for slack when they could just vibe code a chat app in an hour?
The challenge of building a real, valuable software business (or any business) is so much harder than using LLM to prompt “build me a successful software business”
Has there been any good and useful software created with LLMs or any increase in software quality that we can actually look at?
So far it's just AI doom posting, hype bloggers that haven't shipped anything, anecdotes without evidence, increase in CVEs, increase in outages, and degraded software quality.
Software quality has been degrading for decades without LLMs though.
I only have anecdotal evidence from some engineers I know that they don't write software by hand any more. Provided the software they are working on was useful before, we can say that LLMs are writing useful software now.
If all you were doing is taking requirements from someone else and poorly coding them up (and yes I know a decent % of the industry matches close to this) then yes you are obsolete. Something just as useless but much faster now here.
If you are part of the requirements process. If you find problems to solve and solve them. If you push back on requirements when they are not reasonable. Etc. Then you still have a career and I don't see anything coming for you soon.
> If all you were doing is taking requirements from someone else and poorly coding them up
So, in your entire career, you've always worked in companies where you were a subject matter expert on everything the company did? Always knew the business domain inside out? You were running the numbers, sitting with customers, and determining yourself what they really wanted?
> If you push back on requirements when they are not reasonable. Etc
I did, because the requirements had a cost, which I had to balance with limited resources.
If widget A would make 10 customers happy, but would cost two weeks of work, that could be better spent making widget B that'd make 20 customers happy, then it would not be reasonable.
If widget A and B are free, then it becomes unreasonable to say no.
> So, in your entire career, you've always worked in companies where you were a subject matter expert on everything the company did? Always knew the business domain inside out? You were running the numbers, sitting with customers, and determining yourself what they really wanted?
You don't have to be the only person involved in the requirements for you to be involved. So yes, I've been involved from my very first internship where I pitched a new product to the CEO in my second month on the job and got told to go make that happen.
Yes. Every platform offers free tokens generously.
That is a true statement. Might not be much, but is enough for you to produce some code and shit out a readme and then show on hacker news that your capable of pushing to git with the help of llms
Sure, the tiny rushes of dopamine are free, but that's just enough to get your foot into the door and pretty soon they won't hit the same. If you don't quit while you're ahead, it's either gonna financially ruin you or turn you into a zombie that never uses its brain for an extended period of time and you end up writing posts like this one to self-rationalise your completely irrational behaviour because you've been scared into thinking your career is gonna be ruined if you don't keep diving head-first into this addiction.
Not a coincidence that this is happening at the same time as "bet on anything" apps.
The zero cost argument is so hollow it’s got an echo.
The systems that allow this are notoriously underpriced, and yet nobody wants to confront the simple business logic that today’s free beer can’t and won’t last.
Elevating this phase as the normal state of LLM usage costs is plainly dumb and a terrible business take.
> The cost of turning written business logic into code has dropped to zero
It hasn't. Large enterprises currently footing the bill, essentially subsidizing AI for now.
I constantly see comparisons between the 200$ Claude-Code Max subscription vs 6-figure (100k$) salary of an engineer.
The comparison here is, first of all, not apples-to-apples. Let's correct CC subscription to the yearly amount first; 12x200=2400$. Still more than 10x difference compared to the human engineer.
Although when you have the human engineer, you also pay for the experience, responsibility, and you somewhat transfer liability (especially when regulations come into play)
Moreover, creation by a human engineer, unless stolen IP or was plagiarized, owned by you as the employer/company. Meanwhile, whatever AI generated is guaranteed to be somewhat plagiarized in the first place. The ownership of the IP is questionable.
This is like when a layman complains when the electrician comes to their house, identifies the breaker problem, replaces the breaker which costs 5$ and charges 100$ for 10-minute job. Which is complete underestimation of skill, experience, and safety. A wrong one may cause constant circuit-breaks, causing malfunction in multitude of electronic devices in the household. Or worse, may cause a fire. When you think you paid 100$ for 10-minutes, in fact it was years of education, exams, certification, and experience you had paid for your future safety.
The same principle applies to the AI. It seems like it had accumulated more and more experience, but failing at the first prompt-injection. It seems like getting better at benchmarks because they are now part of their dataset. All these are hidden-costs 99% does not talk about. All these hidden costs are liabilities.
You may save an engineer's yearly salary today, at the cost of losing ten times more to the civil-lawsuits tomorrow. (Of course, depending on a field/business)
If your business was not that critical to get a civil-lawsuit in the first place, then you probably didn't needed to hire an engineer yourself. You could hire an agency/contractor to do that in much cheaper way, while still sharing liability...
If written business logic can become code at near-zero cost, people with good product sense can build without buying a big engineering org. That pressure shows up first in commodity SaaS getting repriced (already happening). Meanwhile hyperscalers are leaning harder into being the platform layer by making huge investments, because if apps are easier to create, the durable value shifts to infrastructure, operations, and distribution. Great times.
A - the substitution via AI is good enough to equal everything a business needs to work properly and be sustainable long term. I would argue it does not. A business is NOT just its core tech element.
B - the cost will always remain subsidized by investors’ stupid money. It will not
For sure, because copy pasting from stackoverflow was already to difficult. Everyone loves writing business logic in computer understandable words. Name it LLM, programming language, calculator, there is still parser and interpreter. The only switch is attempt to replace deterministic machines with not deterministic machines by reducing machine error rate to acceptable percentage.
blogs like this seem like they’re in the right direction with LLMs being “here to stay” and a near indispensable part of people’s daily toolkit, but the near certainty that programming as a job or skillset is dead in the water seems just wrong?
like ok the cost for anyone to generate almost
always working code has dropped to zero but how does a lay person verify the code satisfies business logic? asking the same set to generate tests to that just seems to move the goalposts
or like what happens when the next few years of junior engineers (or whatever replaces programming as a field)who’ve been spoon fed coding through LLMs need to actually decipher LLM output and pinpoint something the machine can’t get right after hours of prompting? a whole generation blindly following a tool they cant reliably control?
but maybe I am just coping because it feels like the ladder on the rest of my already short career , but some humility m
There is something poignant about the change taking place in the software industry occurring alongside mass deportations. I can’t put my finger on just what it is that makes it so...
The Pollyanas have a point but overstate it. The naysayers should be more cautious though.
If you're a professional code producer, you shit out code as fast as possible. Don't give anyone time to analyze the disgusting pile of shit you generated, just shit out code as fast as you can and call it a win! Who would prove you wrong?
Would someone waste their biological precious resources reviewing machine generated slop, when your cadence is super human?
Would someone use the same machine you used to evaluate itself? Lol
Saying it again, I think we're in need of a moratorium on "AI Has Changed The World Forever" posts. All of them read the same and offer nothing past "I asked a LLM to make a midsize feature, I haven't looked at the code but it compiles on my machine and that should terrify you" - buddy, we've had people pushing code that compiles on their machine and occasionally goes quickly read or unread in PRs, that terrifies me now.
When these vibe coded projects realise that maintenance, security updates, API changes are still needed. Get ready for a massive swing back to senior software developers being in demand.
Playing software maintainer while many vibe coded web apps aren't built with proper software architecture or practices only makes the swing back to senior engineers being in demand a possibility.
Good luck to those who are building 600K-LOC vibe coded web apps with 40+ APIs stitched together.
I personally love this development. Sure, I find some pleasure in writing code. But what I love most is mapping out a gnarly problem on pen and paper. Then the code is "just" an implementation detail. Guess I'm an ideas guy as per the author?
I predict it's going to be a bloodbath. People who worked for Big Tech have no idea what's coming. Some of us software engineers who have been outside have been experiencing issues for almost a decade. The industry is extremely anti-competitive.
Whatever you produce, nobody is going to use unless you produce it under the banner of Big Tech. There are no real opportunities for real founders.
The problem is spreading beyond software. The other day, I found out there is a billion dollar company whose main product is a sponge... Yes, a sponge, for cleaning. We're fast moving towards a communist-like centrally planned economy, but with a facade of meritocracy where there is only one product of each kind and no room for competition.
This feeling of doom that software engineers started to feel after LLMs is how I was feeling 5 years earlier. People are confused because they think the problem is that AI is automating them but reality is that our problems arise from a deeper systemic issue at the core of our economic system. AI is just a convenient cover story, it's not the reason why we are doomed. Once people accept that we can start to work towards a political solution like UBI or better...
We've reached the conclusion of Goodhart's Law "When a measure becomes a target, it ceases to be a good measure" - Our economic system has been so heavily monitored and controlled in every aspect that is has begun to fail in every aspect. Somebody has managed to profit from every blindspot, every exploit exposed by the measurement and control apparatus. Everything is getting qualitatively worse in every way that is not explicitly measured and the measurement apparatus is becoming increasingly unreliable... Most problems we're experiencing are what people experienced during the fall of communism except filter bubbles are making us completely blind to the experience of other people.
I think if we don't address the root causes, things will get worse for everyone. People's problems will get bigger, become highly personalized, pervasive, inexplicable, unrelatable. Everyone will waste their energy trying to resolve their own experience of the symptoms but the root causes would remain.
To me it seems like the big question for the future will be how to achieve political relevance as "the little guy". It seems like with LLMs the typical "get educated" pathway for the lower class is closing quick. I dread to think of a world where large portions of society are essentially "useless".
The speed and scale are different. The power loom took a while to replace a subset of jobs. It didn't make "manual work" obsolete overnight.
"Going into trades" isn't gonna save you. Knowledge work is 40% of the workforce. Whose electricity are you gonna fix when people can't afford a house?
*THE word ''Luddite'' continues to be applied with contempt to anyone with doubts about technology, especially the nuclear kind. Luddites today are no longer faced with human factory owners and vulnerable machines. As well-known President and unintentional Luddite D. D. Eisenhower prophesied when he left office, there is now a permanent power establishment of admirals, generals and corporate CEO's, up against whom us average poor bastards are completely outclassed, although Ike didn't put it quite that way. We are all supposed to keep tranquil and allow it to go on, even though, because of the data revolution, it becomes every day less possible to fool any of the people any of the time. If our world survives, the next great challenge to watch out for will come - you heard it here first - when the curves of research and development in artificial intelligence, molecular biology and robotics all converge. Oboy. It will be amazing and unpredictable, and even the biggest of brass, let us devoutly hope, are going to be caught flat-footed. It is certainly something for all good Luddites to look forward to if, God willing, we should live so long. Meantime, as Americans, we can take comfort, however minimal and cold, from Lord Byron's mischievously improvised song, in which he, like other observers of the time, saw clear identification between the first Luddites and our own revolutionary origins.
It begins :
As the Liberty lads o'er the sea
Bought their freedom, and cheaply, with blood,
So we, boys, we
Will die fighting, or live free,
And down with all kings but King Ludd!*
What went viral?
To me it just seems like people are pretty divided on the topic which makes sense as it’s an emerging technology. I feel I see as many posts against AI as glazing it.
Which side of that argument has lots of marketing dollars behind it. I suspect what we are all seeing is marketing plus a few useful idiots against everyone else. I will change my mind when I start seeing actual apps created by LLMs which people actually use. What I do see is LLMs replacing search engines and lots of failed software projects. The basic problem here is that the powers that be think the economics of software are like manufacturing. They aren't; they are closer to music publishing, just a lot bigger. And AI isn't having any real impact there.
I agree its not a flat out replacement, but this doesn't stop the marketing from infecting C -suite and leading to them cutting engineering payroll. 400k layoffs in the USA in the last 16 months, and most of those were before this ramp up in "agentic marketing". The tools will 100% be leveraged as a weapon to devalue the expected salaries of tech workers.
Then when they do need to rehire everyone, they will have improved these tools and people will be desperate. They'll say SWE is vastly easier now, and they'll either hire less qualified or push down the value of SWE salaries, I expect the latter.
I think the lawyer money for SWE is gone, so glad I spent 100k in school, years of study and practice. Instead of getting to do the thing I love, I'm going to be expected to basically play project manager with a half literate bot.
> The cost of turning written business logic into code has dropped to zero
Didn’t realize this was science fiction.
I appreciate the author making that the first sentence.
I swear all of these are coming from the prompt "hey chatgpt rewrite this article that got a lot of views"
I've seen non-technical people vibe code with agents. They're capable of producing janky, very basic CRUD apps. Code cost definitely ain't zero
> Didn’t realize this was science fiction.
Compared to my salary, the current cost of the models and tokens to do the work I normally would, is around 10%-25% of it.
Obviously, you still need someone to prevent the models from going insane and messing everything up, but in my experience (webdev projects, DevOps stuff, local software, well known domains), it is very much a force multiplier, as long as you acknowledge that you really need tests and various prebuild scripts.
So I predict one of two things happening:
Either way, it's an interesting direction - instead of ever becoming "proper" engineering (outside of RFCs and foundational stuff), we went from React/Vue/Angular/Svelte/Express.js/Laravel/Django/Rails/ASP.NET/Spring wild west and frameworks of the day (never being able to nail down what "good practices" are and stick to them for decades, but chasing the new thing forevermore), to even closer to producing non-deterministic slop, except that the slop kinda sorta works. Wild times.I think the author forgot that code has to compile and be useful.
And how much is technical debt worth?
What coding agent are you using where the code doesn't even compile!?
The one that cursor used to build their famous browser.
[flagged]
AI misinformation? Please do provide some examples.
Your whole history is AI psychosis btw, seek help.
> AI misinformation? Please do provide some examples.
Like saying they can't generate compiling code in this very thread?
Go look at the 40k failing CI/CD runs on the famous cursor browser.
Ah, it couldn't generate a 100% complete working browser from scratch in a week. I guess the technology is cooked.
Ah, so no longer misinformation?
Depends if I can bundle the technical debt, get a triple AAA rating on it and then sell it
But it is true, the cost is effectively zero. There will be, for a long time, free models available and any one of them will give you code back, always!
They never refuse. Worst case scenario the good models ask for clarification.
The cost for producing code is zero and code producers are in a really bad spot.
I beg to differ. Let's say you're right. Code producers should turn to agriculture and let their managers and product owners prompt AI to produce code. How about code maintainers? Ever heard the mantra "You build it, you run it"? Lets say that AI can build it. Can it run it though? All alone, safely, securely and reliably? No. It can't. We can keep dreaming though, and when will AI code production services turn profitable? Is there a single one which turned profitable?
There are already agents doing this in SRE roles.
Agents are monitoring metrics and logs. A bug is introduced into the system. Error rates go up and the agents notify diagnostic agents. These agents look at recent commits and figure out the problem. They instruct another agent about how to fix the issue and deploy the change. The problem is fixed before an engineer even has time to start looking at logs.
If you aren’t seeing this, you’re not keeping up with what others are already doing. It’s not just people vibe coding ToDo apps.
Calm down buddy, maybe you're confusing code producers with something else. It's 2026 we don't bother with maintenance no more, we /new to keep context clean and start over. Just don't forget to comment - never delete - old code. Always keep dead code around to please shareholders, line numbers up always. We produce code, that is the main thing, never forget.
One could argue we could achieve the same goals by appending \n to a file in a loop, but this is inefficient nowadays with generous token offerings (but could change in the future than I highly suggest just outputting \n to a file an call it productivity increase)
I didn't understand your point about product owners. Who the fuck would ever need one when code produces itself?
Right but memory is expensive now so where do I keep all of this new code that I’ve produced??
Good grief, it's like watching people consuming Soylent meal substitute shakes and proclaiming that chefs and cooking are obsolete.
Because who cares about correct and compilable code, any code will do!
Exactly!
> The cost for producing code is zero
Zero as long as your time is worth nothing, and bad code and security issues cost you nothing maybe.
"Getting code" has always been dead simple and cheap. Getting actually good code that works and doesn't turn into a problem for you down the road is the expensive part
> Zero as long as your time is worth nothing
i can't remember who said it but a long time ago i remember reading "Linux is free if your time is worthless". Now we all use Linux one way or the other.
That's still very much true, but at least in the case of Linux the cost is getting lower and lower all the time. The time investment for many has reached about the same as the cost needed for Windows and as a result we see more and more people using linux. At this point it's a perfectly viable gaming platform!
Maybe one day LLMs will eventually make good code at a low cost, and that will allow non-programmers to write programs with few problems but the cost will never be zero, and I think we're a long long way from making human programers obsolete.
All of the intelligence that LLMs mimic came directly from the work of human minds which got fed into them, but what LLMs output is a lossy conversion filled with error and hallucination.
My guess is that the LLMs producing code will improve for a short time, but as they start to slurp up more and more of their own slop they'll start performing worse.
I am thinking about this a lot right now. Pretty existential stuff.
I think builders are gonna be fine. The type of programmer were people would put up with just because they could really go in their cave for a few days and come out with a bug fix that nobody else on the team could figure out is going to have a hard time.
Interestingly AI coding is really good at that sort of thing and less good at fully grasping user requirements or big picture systems. Basically things that we had to sit in meetings a lot for.
This has been my experience too. That insane race condition inside the language runtime that is completely inscrutable? Claude one-shots it. Ask it to work on that same logic to add features and it will happily introduce race conditions that are obvious to an engineer but a local test will never uncover.
I’m not convinced. That sort of thing usually depends on some very specific arcana or weird interaction between systems that is not in the code. It usually requires either external knowledge or deep investigation and compilation of evidence from multiple sources. I haven’t seen AI do that much.
Look at recent examples of browsers and matrix servers. AI can’t even follow extremely detailed specs with extensive test suites.
If anything, nice and friendly but mediocre devs are in more immediate danger than rough but extremely competent devs.
But we’ve seen C-suits losing institutional knowledge at a drop of a hat for decades so who knows? Maybe knowledge and skill are not that valued.
> The type of programmer were people would put up with just because they could really go in their cave for a few days and come out with a bug fix that nobody else on the team could figure out is going to have a hard time.
Amen. It was a good time while it lasted.
All software engineers become pretty much the same in this world though. Anyone can sit in the meetings.
meetings hardly reach anywhere. most of the details are eventually figured out by developers when interacting with the code. If all ideas from PMs are implemented in a software, it would eventually turn into bloatware before even reaching MVP stage.
> The type of programmer were people would put up with just because they could really go in their cave for a few days and come out with a bug fix that nobody else on the team could figure out is going to have a hard time.
This is the exact type of programmer that isn't going to have any issues - ones who actually know what they're doing and aren't just going to vibecode react slop.
Not really, in my experience you still have to be good at solving problems to use it effectively. Claude (and other AI) can help folks find a "fix", but a lot of times it's a band-aid if the user doesn't understand how to debug / solve things themselves.
So the type of programmers you're talking about, who could solve complex problems, are actually just enhanced by it.
With all due respect to the author, this is a lot of words for not much substance. Rehashing the same thoughts everyone already thinks but not being bold enough to make a concrete prediction.
This is the time for bold predictions, you’ve just told us we’re in a crucible moment yet you end the article passively….
I have a theory: I think the recent advance in coding agent has shocked everyone. It's something of such unheard-of novelty that everyone thinks they've discovered something profound. Naturally, they all think they are the only ones in on it and feel the need to share. But in reality, it's already public knowledge, so they add no value. I've been in this trap many times in the last couple years.
Predictions
- Small companies using AI are going to kick the sh*t out of large companies that are slow to adapt.
- LLMs will penetrate more areas of our lives. Closer to the STTNG computer. They will be agents in the real life sense and possibly in the physical world as well (robots).
- ASICs will eat nVidia's lunch.
- We will see an explosion of software and we will also see more jobs for people who are able to maintain all this software (using AI tools). There is going to be a lot more custom software for very specific purposes.
> Small companies using AI are going to kick the sh*t out of large companies that are slow to adapt.
Big companies are sales machines and their products have been terrible for ages. Microsoft enjoys the top spot in software sales only due to their sales staff pushing impossible deals every year.
It's true the big company products have been terrible but they also enjoyed a moat that made it harder for competitors to enter.
With this moat reduced I think you'll find this approach doesn't work any more. The smaller companies will also hire the good sales people away.
History suggests otherwise, and there's nothing particularly special about this moment.
Microsoft survived (and even, for a little while, dominated) after missing the web. Netscape didn't eat its lunch.
Then Google broke out on a completely different front.
Now there's billions of dollars of investment in "AI", hoping to break out like the next Google... while competing directly with Google.
(This is why we should be more ambitious about constraining large companies and billionaires.)
Well, I made my predictions. Let's come back in a few years.
Netscape didn't attack Microsoft's business software, operating systems or other pieces of their offerings.
Google also didn't seriously attack Microsoft's business.
And neither had the capability to build large software very fast.
Google is both a software company and an infrastructure company as is Microsoft today. Their software is going to become more of a commodity but their data centers still have value (even perhaps more value since all this new software needs a place to run). It's true that if you're in the business of hosting software and selling SaaS you have an advantage over a competitor who does not host their own software.
> Netscape didn't attack Microsoft's business software, operating systems or other pieces of their offerings.
That's not how it was interpreted at the time: Netscape threatened to route around the desktop operating system (Win32) to deliver applications via the browser. "Over the top" as they say in television land.
Netscape didn't succeed, but that's precisely what happened (along with the separate incursion of mobile platforms, spearheaded by Apple... followed quickly by Google, who realised they had to follow suit very quickly).
> And neither had the capability to build large software very fast.
Internet Explorer. Android. Gemini.
I also predict an explosion of work for qualified devs. And I predict there will be an undersupply of them.
Here is my bold prediction: 2026 is the year where companies start the lay offs.
2026 is the year where we all realise that we can be our own company and build the stuff in our dreams rather than the mundane crap we do at work.
Honestly I am optimistic about computing in general. Llms will open things up for novices and experts alike. We can move into I the fields where we can use our brain power... But all we need is enough memory and compute to control our destiny....
I don't know, its a bit of a hellscape in tech right now as thousands of people with deep domain knowledge and people knowledge and business knowledge (ie experienced engineers managers and product owners), were laid off by C Suites desperate to keep the AI funded mandates going
Do you know how hard it to make a successful company or even make money? Its like saying any actor can goto hollywood and be a star
VCs wont fund everyone
Nobody is sure of anything
Yes it is. But I am an optimist for human nature. I personally believe smaller companies doing different things is the future... Scaling as they need. It is a hellscape but people can and will adapt.
> Do you know how hard it to make a successful company or even make money?
Yes I have failed to do it before. I get this.
> VCs wont fund everyone
And? Do you need VCs? Economics mean that scale matters but what if we don't need it. What if we can make efficient startups with our own funding??
I’d like to say its possible
But heres the reality from me- I’m in my 50s and I don't have it in me to grind at the level of 20 year olds to achieve some level of security in an untried business model - and this is someone who has launched 2 AI startups in the past 2 years
In one we got VC funding but I left after setting up their agent platform and tons of AI assisted coding only to not meet impossible deadlines and over-promised AI value to enterprise customers - I was literally working 20 hr days at stretch for 170k salary competing and half benefits against 25 year olds out of stamford with no lives- far lower than the 250+ with stock and benefits I got in my EM role at a big company which now evaporated - I was edged out of that startup role for not delivering “on time”
My second AI startup I cofounded with friends and trusted colleagues its bootstrapped (all of us are over 40) - so we have more experience and better deadlines now but its up to the business gods on how well it will do- crossing my fingers
But its a lot of pressure for sure and I currently have no health insurance and my wife was laid off in December and we lost her benefits
So I wouldn’t call myself optimistic in the end stage capitalistic hellwhole that is modern “middle class” America
I hope a better work model can be found - but having some any salary and medical benefit security would be nice
I went to an AI meetup last week and it was filled with gray hairs - i could sense the desperation as many people I met told me they were laid off recently and trying to dive in
Ironically looking at them it reminded me of those interviews they used to do in Appalachia or something when a town was out of work and advisors came in and said “learn to code instead of mining” (ok I may be exaggerating somewhat but I even know a ex Microsoft manager who had to resort to a go-fundme to keep his family afloat)
update: this reddit thread is somewhat amusing satirical post on this topic
https://www.reddit.com/r/ClaudeCode/comments/1qzj9ve/67_no_f...
>Here is my bold prediction: 2026 is the year where companies start the lay offs.
Start? Excuse moi
Yeah fair... But now it is different I.e. they won't regret it
What gives you the idea that they regretted it?
Except it started in 2023, we are in the middle of layoff waves.
[flagged]
I'm human?
Oytis: I can't reply to you directly, but yes I am sure I am human.
Not sure how to prove it to you.
Are you sure?
I own (cofounded) a medium-sized saas business with hundreds of employees. I maintain final decision of everything technical and still code every day because it’s important. All of the engineers use LLM tools, you’d be stupid not to. But I need good engineers, I replace good engineers when someone leaves, and the business itself is so much larger than just programming. The system is just so huge and complex, and I am the benevolent dictator that architected it and maintains the core design decisions, that the LLM does not replace the need for engineers nor my own expertise.
Furthermore if we were truly in the utopia the author describes, why do all the LLM companies employ (and pay top dollar) for so many engineers? Why does OpenAI pay for slack when they could just vibe code a chat app in an hour?
The challenge of building a real, valuable software business (or any business) is so much harder than using LLM to prompt “build me a successful software business”
Has there been any good and useful software created with LLMs or any increase in software quality that we can actually look at?
So far it's just AI doom posting, hype bloggers that haven't shipped anything, anecdotes without evidence, increase in CVEs, increase in outages, and degraded software quality.
Software quality has been degrading for decades without LLMs though.
I only have anecdotal evidence from some engineers I know that they don't write software by hand any more. Provided the software they are working on was useful before, we can say that LLMs are writing useful software now.
If all you were doing is taking requirements from someone else and poorly coding them up (and yes I know a decent % of the industry matches close to this) then yes you are obsolete. Something just as useless but much faster now here.
If you are part of the requirements process. If you find problems to solve and solve them. If you push back on requirements when they are not reasonable. Etc. Then you still have a career and I don't see anything coming for you soon.
> If all you were doing is taking requirements from someone else and poorly coding them up
So, in your entire career, you've always worked in companies where you were a subject matter expert on everything the company did? Always knew the business domain inside out? You were running the numbers, sitting with customers, and determining yourself what they really wanted?
> If you push back on requirements when they are not reasonable. Etc
I did, because the requirements had a cost, which I had to balance with limited resources.
If widget A would make 10 customers happy, but would cost two weeks of work, that could be better spent making widget B that'd make 20 customers happy, then it would not be reasonable.
If widget A and B are free, then it becomes unreasonable to say no.
> So, in your entire career, you've always worked in companies where you were a subject matter expert on everything the company did? Always knew the business domain inside out? You were running the numbers, sitting with customers, and determining yourself what they really wanted?
You don't have to be the only person involved in the requirements for you to be involved. So yes, I've been involved from my very first internship where I pitched a new product to the CEO in my second month on the job and got told to go make that happen.
Recent models are pretty good at pointing out unreasonable requirements.
> The cost of turning written business logic into code has dropped to zero
Tokens are free now?
Yes. Every platform offers free tokens generously.
That is a true statement. Might not be much, but is enough for you to produce some code and shit out a readme and then show on hacker news that your capable of pushing to git with the help of llms
It's pretty much like drugs or gambling.
Sure, the tiny rushes of dopamine are free, but that's just enough to get your foot into the door and pretty soon they won't hit the same. If you don't quit while you're ahead, it's either gonna financially ruin you or turn you into a zombie that never uses its brain for an extended period of time and you end up writing posts like this one to self-rationalise your completely irrational behaviour because you've been scared into thinking your career is gonna be ruined if you don't keep diving head-first into this addiction.
Not a coincidence that this is happening at the same time as "bet on anything" apps.
> Or, at best, near-zero.
I mean, lots of numbers are near zero depending on your definition of near.
The zero cost argument is so hollow it’s got an echo. The systems that allow this are notoriously underpriced, and yet nobody wants to confront the simple business logic that today’s free beer can’t and won’t last. Elevating this phase as the normal state of LLM usage costs is plainly dumb and a terrible business take.
> The cost of turning written business logic into code has dropped to zero
It hasn't. Large enterprises currently footing the bill, essentially subsidizing AI for now.
I constantly see comparisons between the 200$ Claude-Code Max subscription vs 6-figure (100k$) salary of an engineer.
The comparison here is, first of all, not apples-to-apples. Let's correct CC subscription to the yearly amount first; 12x200=2400$. Still more than 10x difference compared to the human engineer.
Although when you have the human engineer, you also pay for the experience, responsibility, and you somewhat transfer liability (especially when regulations come into play)
Moreover, creation by a human engineer, unless stolen IP or was plagiarized, owned by you as the employer/company. Meanwhile, whatever AI generated is guaranteed to be somewhat plagiarized in the first place. The ownership of the IP is questionable.
This is like when a layman complains when the electrician comes to their house, identifies the breaker problem, replaces the breaker which costs 5$ and charges 100$ for 10-minute job. Which is complete underestimation of skill, experience, and safety. A wrong one may cause constant circuit-breaks, causing malfunction in multitude of electronic devices in the household. Or worse, may cause a fire. When you think you paid 100$ for 10-minutes, in fact it was years of education, exams, certification, and experience you had paid for your future safety.
The same principle applies to the AI. It seems like it had accumulated more and more experience, but failing at the first prompt-injection. It seems like getting better at benchmarks because they are now part of their dataset. All these are hidden-costs 99% does not talk about. All these hidden costs are liabilities.
You may save an engineer's yearly salary today, at the cost of losing ten times more to the civil-lawsuits tomorrow. (Of course, depending on a field/business)
If your business was not that critical to get a civil-lawsuit in the first place, then you probably didn't needed to hire an engineer yourself. You could hire an agency/contractor to do that in much cheaper way, while still sharing liability...
> The cost of turning written business logic into code has dropped to zero
Then go and throw your $0 to fix some real bugs on GitHub.. really, if AI works so well, why are all those issues still open?
Look, almost 2K issues open here: https://github.com/emscripten-core/emscripten/issues
If AI really works like non-technical people think it does, why doesn't Google just throw their AI tool to fix them all?
If written business logic can become code at near-zero cost, people with good product sense can build without buying a big engineering org. That pressure shows up first in commodity SaaS getting repriced (already happening). Meanwhile hyperscalers are leaning harder into being the platform layer by making huge investments, because if apps are easier to create, the durable value shifts to infrastructure, operations, and distribution. Great times.
But it is zero only if you accept that:
A - the substitution via AI is good enough to equal everything a business needs to work properly and be sustainable long term. I would argue it does not. A business is NOT just its core tech element.
B - the cost will always remain subsidized by investors’ stupid money. It will not
For sure, because copy pasting from stackoverflow was already to difficult. Everyone loves writing business logic in computer understandable words. Name it LLM, programming language, calculator, there is still parser and interpreter. The only switch is attempt to replace deterministic machines with not deterministic machines by reducing machine error rate to acceptable percentage.
blogs like this seem like they’re in the right direction with LLMs being “here to stay” and a near indispensable part of people’s daily toolkit, but the near certainty that programming as a job or skillset is dead in the water seems just wrong?
like ok the cost for anyone to generate almost always working code has dropped to zero but how does a lay person verify the code satisfies business logic? asking the same set to generate tests to that just seems to move the goalposts
or like what happens when the next few years of junior engineers (or whatever replaces programming as a field)who’ve been spoon fed coding through LLMs need to actually decipher LLM output and pinpoint something the machine can’t get right after hours of prompting? a whole generation blindly following a tool they cant reliably control?
but maybe I am just coping because it feels like the ladder on the rest of my already short career , but some humility m
I think you’re ignoring the growth trajectory. Surex AI generates almost always working code now, but what till it generate in 1-5 years?
There is something poignant about the change taking place in the software industry occurring alongside mass deportations. I can’t put my finger on just what it is that makes it so...
The Pollyanas have a point but overstate it. The naysayers should be more cautious though.
Lost me at the first sentence. That is an insanely large claim to make with no evidence.
> The cost of turning written business logic into code has dropped to zero. Or, at best, near-zero.
Zero if you dont consider Anthropic's API pricing, the prompter's hourly rate and verification bottleneck.
Everybody offers a generous free tier.
Verification? LoL, lmao even. Your vibes are low.
If you're a professional code producer, you shit out code as fast as possible. Don't give anyone time to analyze the disgusting pile of shit you generated, just shit out code as fast as you can and call it a win! Who would prove you wrong?
Would someone waste their biological precious resources reviewing machine generated slop, when your cadence is super human? Would someone use the same machine you used to evaluate itself? Lol
I personally believe (and so far, my evidence suggests) that AI doesn't anywhere near as well as claimed for almost any of its use cases.
However, let's suppose the alternate case:
If AI works as claimed, people in their tens of millions will be out of work.
New jobs won't be created quickly enough to keep them occupied (and fed).
Billionaires own the media and the social media and will use it to attempt to prevent change (i.e. apocalyptic taxation)
What, then, will those people do? Well, they say "the devil makes work for idle hands", and I'm curious what that's going to look like.
Saying it again, I think we're in need of a moratorium on "AI Has Changed The World Forever" posts. All of them read the same and offer nothing past "I asked a LLM to make a midsize feature, I haven't looked at the code but it compiles on my machine and that should terrify you" - buddy, we've had people pushing code that compiles on their machine and occasionally goes quickly read or unread in PRs, that terrifies me now.
When these vibe coded projects realise that maintenance, security updates, API changes are still needed. Get ready for a massive swing back to senior software developers being in demand.
Playing software maintainer while many vibe coded web apps aren't built with proper software architecture or practices only makes the swing back to senior engineers being in demand a possibility.
Good luck to those who are building 600K-LOC vibe coded web apps with 40+ APIs stitched together.
Honestly, "vibe-code-cleaners" are already out there and in demand!
I even expect "vibe-code-scalers" will come soon, to be able to fix and scale up the spaghetti AI agents plopped in the first place.
The author seems to be an Amazonian, it also seems that they are good at "Invent", but not at "Simplify" bit.
Big-Tech has invented LLMs, that is great. Big-Tech hasn't been great when it comes to "Simplify"-ing things. Actually, notoriously bad at it.
That is the opportunity here; "Simplifying" these workflows, making AaaS (Agent as a Service or AI as a Service)
I personally love this development. Sure, I find some pleasure in writing code. But what I love most is mapping out a gnarly problem on pen and paper. Then the code is "just" an implementation detail. Guess I'm an ideas guy as per the author?
I predict it's going to be a bloodbath. People who worked for Big Tech have no idea what's coming. Some of us software engineers who have been outside have been experiencing issues for almost a decade. The industry is extremely anti-competitive.
Whatever you produce, nobody is going to use unless you produce it under the banner of Big Tech. There are no real opportunities for real founders.
The problem is spreading beyond software. The other day, I found out there is a billion dollar company whose main product is a sponge... Yes, a sponge, for cleaning. We're fast moving towards a communist-like centrally planned economy, but with a facade of meritocracy where there is only one product of each kind and no room for competition.
This feeling of doom that software engineers started to feel after LLMs is how I was feeling 5 years earlier. People are confused because they think the problem is that AI is automating them but reality is that our problems arise from a deeper systemic issue at the core of our economic system. AI is just a convenient cover story, it's not the reason why we are doomed. Once people accept that we can start to work towards a political solution like UBI or better...
We've reached the conclusion of Goodhart's Law "When a measure becomes a target, it ceases to be a good measure" - Our economic system has been so heavily monitored and controlled in every aspect that is has begun to fail in every aspect. Somebody has managed to profit from every blindspot, every exploit exposed by the measurement and control apparatus. Everything is getting qualitatively worse in every way that is not explicitly measured and the measurement apparatus is becoming increasingly unreliable... Most problems we're experiencing are what people experienced during the fall of communism except filter bubbles are making us completely blind to the experience of other people.
I think if we don't address the root causes, things will get worse for everyone. People's problems will get bigger, become highly personalized, pervasive, inexplicable, unrelatable. Everyone will waste their energy trying to resolve their own experience of the symptoms but the root causes would remain.
[flagged]
To me it seems like the big question for the future will be how to achieve political relevance as "the little guy". It seems like with LLMs the typical "get educated" pathway for the lower class is closing quick. I dread to think of a world where large portions of society are essentially "useless".
That's literally what automation is. You could make the same argument against the power loom. People did!
The speed and scale are different. The power loom took a while to replace a subset of jobs. It didn't make "manual work" obsolete overnight.
"Going into trades" isn't gonna save you. Knowledge work is 40% of the workforce. Whose electricity are you gonna fix when people can't afford a house?
I don't agree, but at least that's an argument.
[dead]
*THE word ''Luddite'' continues to be applied with contempt to anyone with doubts about technology, especially the nuclear kind. Luddites today are no longer faced with human factory owners and vulnerable machines. As well-known President and unintentional Luddite D. D. Eisenhower prophesied when he left office, there is now a permanent power establishment of admirals, generals and corporate CEO's, up against whom us average poor bastards are completely outclassed, although Ike didn't put it quite that way. We are all supposed to keep tranquil and allow it to go on, even though, because of the data revolution, it becomes every day less possible to fool any of the people any of the time. If our world survives, the next great challenge to watch out for will come - you heard it here first - when the curves of research and development in artificial intelligence, molecular biology and robotics all converge. Oboy. It will be amazing and unpredictable, and even the biggest of brass, let us devoutly hope, are going to be caught flat-footed. It is certainly something for all good Luddites to look forward to if, God willing, we should live so long. Meantime, as Americans, we can take comfort, however minimal and cold, from Lord Byron's mischievously improvised song, in which he, like other observers of the time, saw clear identification between the first Luddites and our own revolutionary origins.
It begins : As the Liberty lads o'er the sea Bought their freedom, and cheaply, with blood, So we, boys, we Will die fighting, or live free, And down with all kings but King Ludd!*
Your homework:
https://archive.nytimes.com/www.nytimes.com/books/97/05/18/r...
What went viral? To me it just seems like people are pretty divided on the topic which makes sense as it’s an emerging technology. I feel I see as many posts against AI as glazing it.
The recent frontpage post I see is https://news.ycombinator.com/item?id=46926245 (not on frontpage anymore, probably downranked by flamewar detector since it has tons of comments)
Which side of that argument has lots of marketing dollars behind it. I suspect what we are all seeing is marketing plus a few useful idiots against everyone else. I will change my mind when I start seeing actual apps created by LLMs which people actually use. What I do see is LLMs replacing search engines and lots of failed software projects. The basic problem here is that the powers that be think the economics of software are like manufacturing. They aren't; they are closer to music publishing, just a lot bigger. And AI isn't having any real impact there.
I agree its not a flat out replacement, but this doesn't stop the marketing from infecting C -suite and leading to them cutting engineering payroll. 400k layoffs in the USA in the last 16 months, and most of those were before this ramp up in "agentic marketing". The tools will 100% be leveraged as a weapon to devalue the expected salaries of tech workers.
Then when they do need to rehire everyone, they will have improved these tools and people will be desperate. They'll say SWE is vastly easier now, and they'll either hire less qualified or push down the value of SWE salaries, I expect the latter.
I think the lawyer money for SWE is gone, so glad I spent 100k in school, years of study and practice. Instead of getting to do the thing I love, I'm going to be expected to basically play project manager with a half literate bot.