I now started to us AI to help review my juniors PRs, because I couldn't keep up with the amount of code they ship. It started poorly, but now I have my method: I first read the code and flag the lines I'm not sure about, then ask any frontier model (I like Claude here for analysis, even if I don't use it for the rest) to explain the PR and to put effort on the parts I flagged (basically explain in detail the code, not only the PR), and to search through the libraries. Sometimes it notices something I would have missed (like missing an 'order_by' or off by one errors, because the underlying lib wasn't coded like the original AI pretended it was).
I also changed the way I do review because it has been more than a year and the juniors/new hire are still lost, wether on domain knowledge for the older new hire, or just capabilities for the juniors, and discussing with other departments, it's the same for like 95% of them. Now, rather than correcting the PR or adding a request for change, I add a whole unit/functional test to the PR and let that as an exercise to pass the test. They can use AI but I tell them to try to find what part of the code doesn't work before generating the fix, hopefully they'll take ownership of the code if I keep doing that.
even before AI very few developers cared about the actual craft. I'd say only about 3%. For a lot of people it's just a job, no matter if it's a corperate dev or a startup dev.
One of my customers now added a policy that junior developers may use AI assistance, but are expected to write code by hand. Only proven developers can use AI to generate code, and are then expected to understand the generated code, and take ownership.
It's a different mindset. Has nothing to do with curiosity.
For me and you, "fixing" something means finding which assumption was violated and redesigning the solution in light of that. Sometimes this means adding code, sometimes it means making the solution smaller.
For a lot of people, "fixing" means adding code to make it work. This always makes the solution larger.
The latter group can still be curious! But they're curious about which addition makes it work, not about their incorrect beliefs.
What I saw in these cases, the result was a few levels removed from the root cause and they'd add the same workaround to like 5 different places (and miss one or two) instead of tracking it back to the origin where the fix was that simple.
Usually something more subtle like adding an early return somewhere before the offending line, or inheriting from that class and blanking that method, or something along those lines.
"because the underlying lib wasn't coded like the original AI pretended it was"
So, the models ARE dumb. They just are very good at finding patterns in their training data. I mean, when they code minecraft clones, it is not because they can cook up how to write minecraft clones, rather, their training data includes a lot of minecraft-like games code, and they just reuse that.
Bit of a humbling/jarring moment when I realized that people are doing real paid work using LLMs that they could not otherwise do. I mean, it's quite obvious I suppose. But up until now I just assumed it was only a (massive) catalyst for things people would already be able to do with enough time. But nope -- it seems people are right now employed in roles that they would not be able to fulfil the tasks within if AI wasn't there telling them what to write/say/produce. Nobody is really going to come out and say that ... it's not something the less-AI-literate superiors would take kindly to.
I mentioned this then other day, my university class had maybe 80% of people who could not code. No joke. And a lot have gone onto to code professionally.
At a FAANG adjacent company in 2017 we hired a guy from Google with 4 years experience who passed our very hard coding interviews. He literally couldnt code. It blew my mind.
My families business is bookkeeping, they hired someone who graduated with an accounting degree. They didn’t grasp the basics of bookkeeping nor accounting once employed.
You’d think Google who my understanding pride themselves on technical chops would weed this out with all the interviews they do. But maybe it’s just big places have more cracks for people slip into.
I’ve long been curious what you (and a lot of people that express this) feel the cutoff is to know how to code? Because obviously there’s a sliding scale between truly knowing very little, being able to adapt similar work, extensive modifications, coding everything from scratch, etc. on top of understanding certain programming concepts and organization.
Can you build a desktop/terminal app, web app/site, mobile app, or even a library without engaging AI is probably where I’d draw the line. Not everything you touch when you know how to code is doing the whole thing, but I’d expect you’d be able to piece it all together with some time and docs. Pick your own platform that suits too, I’m not saying iOS or Windows only.
For context to my above comment re university course.
The final project we had to do was make a very rudimentary room reservation server/client on the terminal. Java, however you wanted to store the data was fine, TCP sockets were the parameters given.
It took me all of a week to finish it. Lots struggled to even begin. We were given 10 weeks.
I didn’t know much then but I read the docs and figured it out.
Yeah, not the one you are responding to, but I wouldn't define it as from scratch.
Personally, "know how to code" is basically having the logic an reasoning skills to read and understand and apply documentation for whatever subject you need for this job.
If you have never written a web app in your life, but you can follow next.js tutorials and customize them beyond the tutorials afterwards, then you're at the point where the only thing you need is more time or more docs based on what you need to do at work.
> Can you build a desktop/terminal app, web app/site, mobile app, or even a library without engaging AI
If companies would start hiring people who are fully experienced and qualified but can't do the monkey dance of performing live coding, maybe they would find these types of programmers again...
You can learn to code in a few months in your free time and get good in your job. That's different from refusing to learn it, because you outsource it.
Also coding is not that hard. You kinda only write the algorithms down you make up in your head. And you do have an algorithmic understanding when you graduate. You also invented some/several languages in during your studies. The learning an actual developed language IS learning, but it's just boring learning, nothing compared to what you did for your exams.
> people are doing real paid work using LLMs that they could not otherwise do
FWIW, this is not exactly new; those same people were just using other sources like Stack Overflow, blog posts, etc. before, cobbling together random code snippets, libraries, and so on without actually understanding any of that at a relevant detail level.
Sure, with LLMs, one can naturally tailor this much closer to the current need (or at least the need one thinks they have) and iterate ("spew") faster, but it's not a new phenomenon in general.
I think this is fine as long as the success criteria are strict enough that they are forced to learn something in the end.
Any significant testing will inevitably create that situation. The LLM won't have enough context to handle the more precise business requirements. The dev will have to read the code carefully and make their changes by hand. Additional rounds of testing may cause thrashing between regressed states until something clicks for the developer. That lightbulb going off is called "learning" and they are human after all!
Promotion into management isn't so different. They still experience similar thrashing and have to learn or be fired.
My point was not only that so much human learning happens out-of-band, but that it has to be fundamentally different from how an LLM builds context. I've never seen an LLM overcome the thrashing on its own. There's never any "lightbulb moment".
Yeh but I also mean outside of software engineering. Within the gamut of 'being a programmer' it seems fair to bleed into adjacent areas without too much cheek. We've all done it; it's part of the learning curve. But I was talking more about people in other knowledge work who have to come up with a lot of prose-like material about {insert thing}. Marketing, consultants, PMs, or even domain-specific analysts, .. ya know, the types of office jobs where people basically write emails, attend meetings, discuss reports, and produce mostly text-or-data artefacts all day long.
aka there are a lot more bullshit artists around these days
i know of several engineers who produce absolute slop and who probably would have produced nothing at all in pre AI times (which would have been preferable) and probably let go or never hired (even better).
they impose such an enormous drag on productivity that they more than wipe out any gains from people using the tools responsibly.
I started manual coding again at least 1 day a week.
Its good for the brain, but man it's really slow.
And the AI results prove it: Over the past couple years I've been able to bring several work projects to completion single handedly.
Our R&D company is now handling more clients and we are returning better outcomes/products faster than before.
So...what to do? Results don't lie.
However I will say LLM's can't write multi threaded double buffers by them selves. Probably because they are so application specific.
But I think the reason I can go do fast is because I went so slow before. I have an intuition I can transfer into an LLM. Question is if that intuition will become sloppy as time goes on...
“I wanted to see if I can get in on this 10x magic. I decided to put my money where my mouth was and I bought $10 worth of DeepSeek credits to use with a project I was working on.“
I'm sure DeepSeek isn't the point here. You can change the name to whatever you prefer and the article still holds.
Actually, I think the author put DeepSeek on purpose to avoid the obvious ChatGPT/Claude comparison — because whatever he chose, there would be a question of why model A and not B, while the point of the article isn't about models comparison at all.
It is the point. Also open source model enthusiast tell you otherwise, there is a coding quality gap between these models. If I use DeepSeek, I do so knowing that I have to limit to simpler tasks on smaller, well specified prompts. What the author did, letting the model do the planning, is not something DeepSeek will excel at. I'm using GPT (Terra, Sol, Luna), Claude (Opus 5, Fable), Qwen 3.8 and GLM 5.3 Flash daily and have to vary which model I use where because there's a huge intelligence step function difference here. That's why this article is so useless:
Imagine someone trying to make the case that riding bicycles is a terrible experience and their whole argument is that they took a random cheapo bike with flat tires and rode it for 3min and that wasn't fun. Sure, but if you buy a 25k carbon bike you will have a different experience. I'd not trust that person. If someone told me they have 10 bikes they ride daily and can explain the differences, in detail, between their bikes, and what they excel at. I'd trust that person's opinion.
There’s a night-and-day difference between frontier and budget models, no question. But the issue isn't the tooling at all : if you put someone who doesn't know the rules of the road on a $15k carbon road bike, they're just gonna slam into a telephone pole at 30 mph instead of 6 mph
Excellent analogy. This paragraph invalidates the entire post and honestly just looks lazy. The author may be right anyway, but with that level of experience with these tools, he is really just guessing.
Right. Better throw money at 5 different ones and then people come and tell you that you just need MORE agents and throw MORE money at it or you're not doing it right.
> I'm also convinced the effect would not be there If I had a 10$ budget.
Sure it would be. I pay $10/month to OpenCode for a Go subscription, it's fine for day-to-day coding tasks. I wouldn't necessarily try and one-shot a production app on that budget, but with decent planning and test-driven-development, it gets the job done
The cheapest coding subscription is $20 and you can get a refund if you cancel within 7 days.
Actually, I lied, Codex is free and I developed my first AI written application using it and the free tier limits were generous enough to work on it for several months.
I got here and... I agreed with everything up to this point.
The author makes a good point. If you don't know what you're doing, AI accelerates that. No question.
But they put a whopping... ten bucks into using DeepSeek and weren't impressed with the initial results.
I know they try to cover this with "you just aren't prompting correctly!" but if, in 2026, you aren't able to have an LLM generate decent quality code... IDK what to tell you. Good luck I guess?
The workflow isnt a single prompt then you have working code from the LLM, its 30 prompts of discussion and then reading each proposed edit and having tangent discussions until you're somewhat ok with it. Then reviewing the changeset as a whole and refactoring a few times
> but if, in 2026, you aren't able to have an LLM generate decent quality code... IDK what to tell you.
I think the author's point still stands. At some point you need to determine whether the output is of decent quality. A lot of people don't have the ability and experience to do that.
In general, the frontier models are not capable of reliably authoring non-trivial code without careful oversight yet. They are great at producing code that can pass tests, but not neccessarily a code review. This means if you care about code quality you still need a human in a loop understanding what has been done, and that becomes the bottleneck. And less disciplined folks will indeed become increasingly dependent.
However, over time the complexity of problems where you can get away with less/no oversight is increasing. And the models are already great at solving certain classes of problems where one doesn't really care that much about code quality, that wouldn't have even been attempted in a pre-LLM world. Over the weekend I was using Claude to add features to the compiled (no source available) firmware of one of my audio devices, adding workflow features by patching assembly and custom DSP code.
In coding, as with other areas, what's emerging is jagged intelligence.
DeepSeek has been the most used model on open router by far, only temporarily overtaken by ox alpha when it was free, so it's the most representative experience
> Why not use actual frontier models, and you know do some real research, before writing a blog post?
Seems like they've already done more research than you. Their results not matching your expectations does not indicate a lack of research. If you feel they missed something, why don't you go spend your own money on whatever frontier model you want and go publish a post with your own conclusions.
What expectations do I have? They used one AI model and then wrote a blog post as if they have an informed opinion. It has no credibility whatsoever, why would I listen to anyone's opinion who hasn't done proper research.
It's like playing one video game and then claiming all video games are boring and have bad gameplay and stories.
It is funny that only real research on productivity gains from AI shows at best very minimal gains, but AI bros will always tell you "no no no, you have used wrong model, try a different one, there are more of them, you have to try, trust me" and call that a "research".
Ah yes I'm an AI bro because I said the author must do thorough testing before coming to a conclusion about LLM's.
Looking forward to the authors next blog post on how after driving one car they find that all cars are slow and uncomfortable. Then the follow up after trying one phone and that all phones have bad cameras and battery life.
Not at any decently sized company. Delivering features is all that matters. By the time you're cleaning up the buggy mess, they were promoted two times and work somewhere else now. If anything, AI makes it much easier to suck and still get things done to an acceptable degree.
Nope. That would require the people paying for the code (those requesting the projects) to see something wrong with the code. As a coder, you would look in an see a mess. As a user they'll probably see something relatively nice, with some maybe strange behavior and non ideal, but what software sin't non ideal? It does most of what we wanted, and you finished it how fast!?
'Or else admit this is a dopamine game that makes you feel like The Universe's Most Special Programmer™ when it's really just gamified mass-scale intellectual dependency.'
Alternatively, AI teaches you lessons about technical debt, code maintainability, and architecture faster too. Traditionally it took at least one or two years to really bump into those problems. Now any student can get into these problems within one or two weeks.
> “after four years of open source LLMs, we should have three AirBnBs, two Stripes, and three Dropboxes thanks to the power of AI.”
Airbnb, Stripe and Dropbox were created in a different time when the market was much less competitive.
Saturation of software development velocity doesn’t increase large scale product opportunities in the market. It can also mean that opportunities get filled even more quickly by niche players, and nobody gets to grow to Airbnb scale.
IMO the latter is what’s currently happening. AI-powered companies are like little mammals scurrying around between the feet of the dinosaurs, and commentators like the OP look at the evolution of the brontosaurus as evidence that the mammals don’t seem to be growing as they should.
There are some good points, and I ask the question of where is the ground breaking stuff myself, but severely weakened by
* stretching the timeline: the actual real programming ability appeared in LLMs in the last 6-8 months, not 3-4 years,
* using the weakest possible tool: and I bought $10 worth of DeepSeek credits that is a far cry from Claude with Fable.
Also, I know nothing about marathons but for most uses putting the app, database, and background processes on the same server is very much the right starting point. With the next steps being employing Cloudflare or similar solutions long before managing a fleet of servers.
I remember being told in early 2025 that only now has "real programming ability" appeared in LLMs.
And since then there's been 3 more "now real programming ability has been made available and previous stuff was just toy examples" cycles (summer 2025, winter 2025 and spring 2026)
Looking forward to the next "everything before this was trivial and bad, here's the good stuff" moment
If you haven't been using the latest NanoQwen27B-5.2.10-Turbo-KimiGLM-Pro-10.4 in the past 13 hours, your information on what LLMs can do is outdated and you're falling behind. Check your priors.
It's a bit Goomba Fallacy. For different people the threshold was crossed at different times. That you're getting told different things by different people doesn't make those statements self-contradictory.
Yeah, it's interesting how the current thing is always the end of programming, but then when a new thing comes it turns into absolute trash that can't be used for anything.
Even as someone using AI on the regular I'm starting to hate the "You didn't actually use this exact most expensive model so your point is invalid" argument.
This is fair to say if someones last experience with AI was copy-pasting code into GPT3 chat windows years ago, but Deepseek is a more than capabale model and enough for someone to get an informed opinion about the technology.
If people have actual counter argument, use those. And if some of those counter argument are "What you say isn't possible, the neweste model can do and here are examples of that", that is fine.
But a blanket "Nuh-uh, it wasn't Model X" is not only a poor argument but also automatically invalidates any criticism when a new, better model comes out - and that can't be the basis of a good argument.
In this case, the argument is valid though. Only recently have some of the models become powerful enough (for some) to actually be useful in day to day programming, without too much hand-holding. For most, this change occurred with the introduction of Claude Opus, and OpenAI and Google have caught up. Unfortunately, none of the "open" models is at this level yet.
The field is moving fast, and asking for scientific arguments is not realistic. It takes an extreme amount of effort to show what exactly is different.
We were in a similar position with static vs dynamic typing for decades. There is still no scientific proof that one is better than the other, but it is quite obvious to professionals which flavor works better in a given situation.
So, even though the argument might be sloppy, I subscribe to it. Using DeepSeek to dismiss better models is the bad argument here.
Edit: added "(for some)" as a disclaimer that you still need to be a fairly decent programmer to actually benefit.
> Only recently have some of the models become powerful enough (for some) to actually be useful in day to day programming, without too much hand-holding.
That still is very much not the case. Every model, yes, including $most_recent_model, needs supervision or you will get burned.
The models got good starting with 2026 and that isn't some attempt at excusing it. Companies like OpenAI started building dedicated models around their coding harness called Codex, there was gpt-5.3-codex and it was both cheaper and better at using the harness than the regular models. Then they started merging the two model types into their main release models. All of this happened like 6 months ago.
You don't have to pay money to use Codex, there is a very generous free tier that costs you nothing, you just have to accept being told you're out of tokens every day. Because your token limits are low, you need to make sure that you accept or reject everything manually and when it tells you that it wants to run a command you have to paste in the command into your terminal and only paste the relevant output back otherwise it floods the context window.
> Even as someone using AI on the regular I'm starting to hate the "You didn't actually use this exact most expensive model so your point is invalid" argument.
What the author of the article is doing is dismissing a technology so disruptive that it's basically all everyone's talking about in the "tech space" at the moment (I mean look at HN frontpage for the past few months), by trying a relatively mediocre (but still quite good) model for about 10 seconds.
The reality is that frontier models suddenly got very good in the past 3-6 months. It has it's problems, and you need to learn how to use this new tool (as with any tool).
But models can and do generate good code. They also can and do generate absolute garbage (even Fable).
You need a good harness, tools to help the model check it's own output, good context, and a good idea of what you actually want. If you have those 4 things, the chances of generating absolute garbage are pretty slim (but yeah, still there).
I wish I could quote so much from this. But it encapsulates my issue with AI. Yes it can help. But it's not the end all be all. We still need smart people to make decisions.
Also, AI was build on what people figured out before. What happens when people stop figuring things out? Or stop making it available to AI? We hit a wall.
AI is like a tool used by mega-corporations to change the world according to what they want. We see this with the increase in RAM prices. I don't want to pay the overprice here - AI companies owe us money. People seem still mostly in the AI hype phase, but a lot more criticism has amplified in the last months. It is only a matter of time until the hype phase is over.
His AI experience is $10 of DeepSeek - but I think he is right anyway in the main points.
You can most definitely ship crap much faster than you used to. It's obvious to anyone when you're shipping crap. And it seems everywhere I look people are shipping crap. Both software and writing.
The hardest part of shipping quality software is not and has never been "writing the code". The hard parts are product taste, architecture and ensuring your product actually solves the problems it should in an efficient and secure manner.
If you don't have an intuition of those, you will likely still be shipping crap. Just faster.
Yep, using AI to remove your thinking process and judgement, and just let it generate stuff as fast as possible, will definitely make you ship crap faster! If you did the same thing with a junior developer you would get the same result.
Where AI shines for me is accelerating the learning and exploration process. I can get up to speed with new tech fast. It is good at spotting issues in designs and code. It can knock out quick tests or benchmarks to support me. The quality of what I can produce with AI support is much higher than I could without it.
So it really just depends on what you use it for. If the goal is "replace humans and ship fast" that's one thing. If the goal is "explore the problem space in greater depth", it's another.
That's one of the few consumer-facing areas where there are still standards in place, namely PCI-DSS. As far as I know the audits require the name of a human who is responsible for payment security. Card companies can one-hit kill your startup if you're breaking those rules (maybe purely blockchain startups are exempt).
Yes, you can offload this to stripe, but then your app should never see the card number and certainly not the CVV. You end up storing these, even by accident, both stripe and the card companies will hate you.
Speaking anecdotally, at least: I don't think the lack of publicly available software is necessarily evidence that AI tools are ineffective.
From the limited perspective I have as a senior developer working in industry, I think that:
1. It has been a relatively short amount of time since the November 2025 quality increase in frontier LLMs.
2. It takes time for developers to learn how to use agents effectively, once they begin to trust them.
3. There is a significant difference in the output quality dependent on how you guide the agent.
4. Experienced developers are creating lots of quality software that is used internally (or only personally by them) but which they don't have the time or confidence to release for others to use.
I largely agree with this piece but the author isn't really giving the AI development process a fair shot. It's not a matter of "prompt better" but of establishing processes that account for and correct agent shortcomings.
For me, it's taken thousands of dollars and months of experiments to get a process that produces working and maintainable code. You can't just go in with $10 and be like "Wow it doesn't work" after playing around with it for 30 seconds, which is how far $10 takes you on models that are doing work. All that does is expose the shortcomings that need correction.
You can get around these not by just "prompting better", but by establishing rules and pipelines that are enforced with standard software engineering tools like compilers, linters, type checkers, borrow checkers, profilers, benchmarking, smoke tests, proof assistants, and so forth. Those things we know work really well, so AI coding shouldn't be viewed as AI doing all the work, but AI orchestrating all of the tools that we know work so well in software development.
The magic of agentic engineering is only possible because our software tools are so good not because agents are.
It almost feels like ages ago when the promise of AI was to be a sponge of knowledge, to help us dig in quicker and speed up the search process of looking up implementations on Google / Stackoverflow because the expectation was that we authored code and demonstrate it to our peers it was correct as the role of an earnest code contributor.
You cannot overstate how disruptive agentic powered development has had on the startup scene. There is clearly an embrace or loosening of that expectation which is upending the profession and also paradoxically giving superpowers to us who have been in the business a long time.
It’s hard to know where this is all going as we are being challenged by this transformative, powerful, yet problematic technology every day at work.
This made me think of "1,000 True Fans". I'm sure there will be more than a few business school case studies written about competition in this era of software development. Even things like niche apps singularly tailored for small skilled trades are seeing competition. AI tooling will definitely allow new/smaller/solo players to clone apps and undercut the legacy players on price more rapidly than before. Some of these very narrow apps are fairly expensive ($100s/month), if replacements of acceptable quality can be produced they will take some market share. This may be death (shrinkage) by a 1000 cuts for some established players.
> Borrowing Disesdi's logic in their arithmetic (which is hard to argue against), after four years of open source LLMs, we should have three AirBnBs,1 two Stripes,2 and three Dropboxes3 thanks to the power of AI.
>
> So. Where the fuck are they?
I think this is a great question, and I have a completely unqualified theory.
I work in a fairly specialised field, I've notice when I use AI to try to generate code in fields where I'm very experienced, it produces a worse output at a slower rate than I can produce manually[0].
When I use it in fields I'm not an expert in by any stretch, like webdev, genAI massively improves my output by a huge factor.
I think this makes the productivity gains a little fuzzy. Claude code makes me 10x faster at webdev, but I'd be a very slow webdev. Ultimately, the next innovation in a field, will probably be from an expert in that field, and they therefore won't see the kind of "10x productivity boost" that non-experts do.
[0] I still find AI helpful for exploring code etc, just not so much producing output.
At least once a week I have to drop what I'm doing and race in to stop the non dev manager from merging a fully vibe coded patch which will blow up production. Claude Code has made this guy a negative 10x developer.
Just let him take down prod on a friday night once and leave him to deal with the PagerDuty alerts solo. Cures the urge to vibe-code infrastructure real fast
They don't have permissions but if I don't rush to block the PR with a bunch of review comments, one of the other AI brained devs will just hit approve on it without reading or understanding anything.
Yes it is hell here. But it's a new kind of problem that just showed up.
I'm experiencing similar rubber stamping around my company.
CEO hired an AI forward CTO / Head of Engineering. He said at his last company his directive was NOT to look at code anymore.
I've watched thousands of lines of misguided, verbose, brittle code (but there's 24 tests that nobody reads) go essentially straight into production. Nobody understands it, we're adding so much trash documentation that is impossible to keep up with.
It's an absolute train wreck, I feel farther away from the system than I did when I started. Commit messages / PR descriptions are to be generated by AI (and pushed).
It's only broken production code a couple of times so far, but my guess is as this continues we will be unable to maintain the codebase without AI involved.
AI can be a force multiplier, we are not using it that way.
My main issue with it is that when a human developer broke the production codebase there is generally an "A-Ha" moment, like, "oh no... I know exactly what that error means, let me run a hotfix and we can change things so that doesn't happen again."
Now we're at the full behest of Claude "do you see what happened to our codebase, robot? Can you help fix it?". Not a fun place to be...
This is so maddening. I'm tech lead and I run a tight ship on the repo I'm in charge of, all PRs unofficially wait for my approval (it's mostly juniors), but if I ever go on PTO, people just approve absolute shit. It is a common occurrence that I come back from vacation and have to get people to revert merges.
At worst, needs another manager to approve. Now they can both saddle up on the velocity/time to market wave... together. The racket begins in earnest.
How does a manager do this, they ask? I answer: the same way they do everything. Lunch. If the effort fails, at least they tried! Deck chairs, Titanic, etc.
I think my 10x comes elsewhere than writing code. For ex llms are much better at bash than me. So good I don't have to leave my shell. Jira? ServiceNow? Ha! Claude you go update that ticket. Go check my history and git commits and write me my standup. I'm not leaving the terminal.
I don't disagree with the title of this weblog post, but its premise is wrong: the existence of AI has an impact on the total number of tech companies, but not on the number of companies that make it real big (the AirBnBs, Stripes and Dropboxes the author is looking for).
I tried really hard to get senior management to understand this where I work. Yes, AI is an accelerator, but that doesn't necessarily mean it's going to accelerate you in the "right" direction.
I have a coffee cup with the writing "Do stupid thing faster with cafe". That's how I feel about myself when I use AI carelessly... The speed with which I can make a mess is astronomical!
AI does make me suck more, which is a good thing. It sped me up and has me iterating on things in 3 days that normally would have taken me more than a month to do. Surprise, surprise: I'm discovering that I have a lot of blind spots and false presumptions about what good programming is. My understanding of the craft is patchy, but that wasn't as noticable when going at the speed of human-programming.
It's like cars. If you only ever drive at 30 km/h, a 30 yo car, that had its last oil change in 2008, might be great. But, if, in the same car, you start driving at 200 km/h, you might notice that there's a lot of room for improvement. My programmer-self is the 30 yo car in this metaphor.
How you use AI as a person who can actually code is up to you.
You can use AI in two ways, either…
1. set the plane on autopilot and arrive at the destination having no idea how you got there - like any old idiot who can type a prompt. And also pay a very expensive air fare for this super power auto pilot.
2. You can pilot the plane yourself. Plan the route in advance. Set waypoints. Make decisions along the way. And land the plane manually. And the reward is you know the route and you pay 1/10 the cost.
If one truly believes it's not in a bubble, then borrow and leverage an unlimited amount and bet the house & your family on creating a business with infinite growth and value. Don't know how to do your own prompts? Borrow and then hire others to do it for you. If those who you hire don't know, they can hire others too.*
[* This is not financial advice. Please don't actually do this.]
You could make the converse argument, too: why don't you borrow and leverage and bet and short these AI companies, if you so vehemently think it's a bubble?
The bottom line is that markets are complicated, useful technologies are often accompanied by bubbles, investors are not always rational, and people generally try to make the best decision with the information they have available. The answer is likely somewhere in the middle, but that's a lot more boring and a lot less inflammatory.
> You could make the converse argument, too: why don't you borrow and leverage and bet and short these AI companies, if you so vehemently think it's a bubble?
Because going long and going short are very different. For successfully shorting something you need to have a pretty precise estimate of when the crash will happen. Being a comparatively short time off can cost you everything.
Yeah, but that applies to the argument in the original comment too. It's supposedly directed at someone who denies we're in an AI bubble, telling them to put their money where their mouth is. But then suddenly they're supposed to risk literally everything they have on "creating a business with infinite growth and value" before their "unlimited" leveraged loans are called in.
Obviously there's some hyperbole there, and I'm not really sure what serious point was being made, but it does seem to depend on the anti-bubble guy having some very confident and specific beliefs...
You mean for companies that make -$2.50 for every $1 of revenue, and bury their groups $2Tn debt in a footnote?
It is not a question of if the bubble will go, but when... but you are right that Bears or Bulls always get it wrong predicting the future (if they are a legal investor.)
Personally, Shrek movie release correlation with market corrections is funny, and a new film is due June 2027. Please hedge your bets with a diversified portfolio. =3
Why does it have to lead to a 10x increase in revolutionary companies?
Next to maintaining and expanding my own network of websites I 10xd writing boring CRUD applications for companies that were otherwise unable to afford it, making all their employees more productive. There's true economic value in that.
This is fine, but software also needs maintenance, especially security updates. If we have an explosion of new software without the corresponding increase in maintenance work - very likely from clients that pay peanuts for software - the world will be taken over by botnets and malware.
I just allow access to their application only for their IP address(es) and my deployment server checks if I need to update anything. That's not expensive.
> The chatbot recommended some of the dumbest shit you could possibly do
The "I tried it and it sucked" is borderline conspiratorial at this point. There are enough talented, thoughtful developers saying there's something real here, and it is worth believing them and investing some time to understand it, even if you come out the other side and decide you don't want to use LLMs.
Use a very good model. Set up a good harness. Spend some time on your system prompts and skills. Develop your intuitions about how the model works, what it's good at, how to scope the work, and how to steer it. Recognize when it's alleviating menial work and recognize when it's making choices you really need to understand yourself. Be patient and accept that the failures are going to be very painful for a while.
Don't write it off until you've genuinely seen the upsides.
You describe my experience with Claude (except my gut feeling is that system prompts and especially skills are useless, and I've even done limited testing that agrees with that).
However, your discourse is not the usual borderline religious discourse that's used by LLM advocates. You're not extreme enough either pro or against LLMs :)
Ha, well that's good because I would hate to be labeled an "LLM advocate". I've been on the negative side of the conversation many times, and I think the downsides are very real and often ignored. But calling them useless is just not reality.
As far as system prompts and skills, my approach has been to start with nothing and add very specific instructions as I notice issues or feel like I need them. Mostly around trying to keep it terse, kill the obnoxious rhetoric ("It's not X. It's Y"), and give specific guidance on how to write code. Skills are things like "use edge-tts to generate spoken audio for this answer and send it to me" kind of thing.
> except my gut feeling is that system prompts and especially skills are useless
I tend to agree. Not with the extreme version of this statement: there are some genuinely useful instructions I give in my Claude.md and prompts. Comment style, how much to push back, which subagents to orchestrate for tests, reviews, etc.
But I've tried the single-sentence versions of those and versions with multiple files of long prompts for the orchestrator and its subagents, and I can't tell the difference in output quality. If anything, the shorter version is better
> Don't write it off until you've genuinely seen the upsides.
This is ridiculous. Basically you're saying "if you try it and find it sucks, you're wrong, keep at it until you change your mind". That isn't a tool at that point, it's a religion.
Of course you don't need to keep trying until you think it's good. But if you haven't found anything it's good at, and meanwhile lots of smart and seemingly-honest people are saying they've found it to be very good at some things, then it's pretty likely that you really are holding it wrong (and perhaps not even trying to hold it right in the first place).
How do you account for the lots of smart and seemingly-honest people who are saying they've earnestly tried it and find it not useful? These are all just personal accounts. The only empirical evidence we have is the output, but there is conspicuously no big successful AI project taking over anything, which is some minor evidence against.
No, they're saying there are upsides, even if you think they are small, or think they don't outweigh the downsides, you at least have to be honest about where the utility is.
- when are all the "I am running agents 24x7 a day on max plan" people on HN gonna learn this?
- best case scenario bro: LLMs get infinitely better and nobody needs to code anymore
"software requirement prompting is the new skill" I can write those pretty well bro
- worst case scenario: AI market crash, cognitive debt spikes across every major organization filled with vibe coding juniors that have never spent a single day in their life debugging a production setup without AI.
- Now the whole world is filled with 90% programmers that cannot add 2 numbers in c++ without using GPT
- Guess what bro? I am now one of the most valuable programmers there is :)
> and ruining the quality of internet search engines.
Tbf, I'm not sure we can really lay this one at the feet of GenAI. The SEO bros had pretty thoroughly ruined search before LLMs took off - the process just accelerated a little at that point.
it hasn't even been a year since Opus 4.5 and my work codebase is thoroughly ensloppified. From the top, the assumption is you don't edit it by hand. It was fine at first but hand editing is becoming more and more painful. LLM coding is a cancer in this way.
I think it depends on what area of the world you work. We recenty had one of our plant managers build a web portal to keep track of some of our operational tech, with features you wouldn't find in standard products. I've been turning it into a container app that can actually deploy safely into our cloud infrastrcture. It's quite frankly better quality than what most external software companies have provided us with in the past.
Don't get me wrong. It's not great. It would never pass any of our policies for things that actually operate stuff on the power grid, but as an administrative tool that can live in total isolation from the vital networks. It's perfectly fine. It's also not like we would have hired the best software companies to build it otherwise. We'd hire some low-level cheap consultant house who would then likely get cheap student labour to build it. With that in mind though, the AI is much better than what the realistic alternative would be.
Money wise it's also cheaper. It's been roughly €1000 + the time it's taken us both. If I had known they were doing it, I would have rolled out the developer cowork app/skills/whateveryoucallconfigurationsthesedays to them. This would have avoided their AI building it to be depoyed on a VM rather than in our managed k8s in our Azure. It would also have written the code a little different, used UV and maybe django rather than flask. But hey. For what it is, it's like a 90% cost saving compared to buying what would've been a less maintainable and lower quality system.
I think perhaps the greater issue will be finding people who want to extract the gold from the heap of shit and getting it to run in production. I don't personally mind, but it's not like any of my colleagues would've wanted the task.
By coincidence, we've been building software for five years that other plant managers are now considering using to keep track of some operational metrics. And I don't think we should put LLMs into our development cycle, simply because the data our software is built for can have critical meaning in some situations.
The insecurity in a vibe-coded web portal isn't that someone hacks it with XSS, it's that after the next vibe-coded release, some X quietly becomes −Y somewhere no one expects.
From this perspective, having no software at all might be better, or as in your case, safer.
I don't think you should keep AI away from software development for security reasons. We're in the European economic area and are completely NIS2 compliant and we use LLM's to aid in our software development for high risk systems. None of that is vibe coded though, so it's not like I disagree with you either.
As you point out this portal isn't that, but what protects us is the processes around compliance. This can't grow from X to Y because not even the CEO has the authority to overwrite our compliance gates. The EU is a tremendous help in this area since personal liability changed things completely.
I don't think they are opposite. We still don't let AI run anything we do. That being said, things have changed. 4 months ago I wouldn't touch the M365 Copilot thing if you paid me to, now it's basically the only AI I use professionally. Cowork changed the way we work, and I've previously spoken about how I think Microsoft sort of won the AI enterprise "war" by selling the tools and not the mode. If you'd asked me a year ago I would've been very anti AI in general. I got proven wrong. I do think it's going to be interesting to see where the EU goes with things though. Currently you can't use all features of Cowork as an example, because they are illegal in the EU. So you can't let it run your browser session for you as an example. I wonder how long that will last.
I still doubt we will ever give an AI access to run code on our systems directly though. In isolation, sure, but other than that.
This is a side note, but my personal favorite part of Cowork is that I can roll out our compliance policy to every developer as a Microsoft Teams app (no, that makes no sense to me either). So when they try to install some package that isn't pre-approved their Cowork agent won't let them and will instead explain how they might get approval. If they then continue to reference it, Cowork will even alert us.
Yes cowork is the first part of copilot 365 that's actually useful, to me too. It is however paid separately per token which means you need the expensive 30$ subscription and pay tokens on top of that.
If you just get cowork directly through claude you do get generous usage within their 20$ subscription.
I have to use it through Microsoft too because of their lobbying our company but I don't think they won any war. They're just reselling other people's stuff. The integration with office is alright but I don't really rely on that. In my personal life I avoid them.
What you're telling me now is that in the short period of 4 months, an American corporation took an employee of a company the EU relies on for critical matters from "wouldn't touch it if you paid me" to outsourcing thinking.
Are you purposely twisting things to fit a certain narrative?
If you really want to go into it, I've previously talked about how we used AI tools provided to us by one of our major investors who do so for all the companies they are invested into. These were also Anthropic and OpenAI models. The main difference is that Cowork has access to files on a users one drive (and that we get a lot more control over what goes into it).
This is the funny part, because I have nothing to twist.
You work in the extreme opposite setup to mine. VC vs bootstrapped. LLMs vs hand-coded. High margins vs open-source. If you ever look at our codebase, you will see that there are very few code dependencies, and of course bringing in another as large as LLMs is not in scope.
If you wish, we can always continue our exchange by email (mine is in hn profile).
> What you're telling me now is that in the short period of 4 months
Not OP, but I opened the link to the comment and it was six months. I get why you're skeptical, and I think it's fair to point out. But that's sort of glaring when I opened the link to contextualize your comment.
Your perspective is correct. We call them shadow developers, but it's a massive issue and security threat. The more tools Microsoft adds to their AI admin center (and put behind that ridicilous Agent365 DLC license) the more we see just how big of an issue it has become. People are using their personal credit cards to buy AI tools and use them wildly irresponsible.
In the big threat picture a vibe coded web tool that's not on the internet and runs in total isolation on it's own management group on the "this might get hacked" tenant in Azure is nothing though. I'd worry more about all that OT which was compromised from the factory which sits around in the energy sector. Especially because it's very easy to draw you a risk analysis that will tell you that unless you're running a Nuclear Powerplant then it makes no financial sense to secure your stuff. The audits are so rare and the consequences so low that it's cheaper to just pay the fine (if you ever get one).
Most actual security happens when someone on the ground decides that it's just too stupid that something clearly labeled "DO NOT PUT ON THE INTERNET" was put directly on the internet.
I doubt it would be better served as a customized off the shelf software. They certainly could have done it with our existing systems with SmartSheet, Microsoft Fabric and/or Power Apps with SharePoint for document storage.
From an enterprise perspective this becomes complicated for various reasons. RBAC is one area. In the perfect world you have a system to handle roles and rights to every system, something that you can give managers access to so they can maintain the access available to their employees, something linked with HR. In reality you have EntraID with a hieracy which is sort of automated by HR data, but not really, because sometimes HR puts everyone on the CEO level by mistake, and, if you trusted HR as authoritative that would've just broken all the EU laws. So you have all those Entra groups and you need IT Operations to maintain them and since you want to build it on job roles and not people you'll typically not be able to maintain them in the off the shelf system. Which means that you would have had to build a web portal for the plant managers manager where they could maintain a couple of Entra groups in a web interface. That or you'll have to setup an IT support flow where you add yet another system that IT has to maintain access for.
Then we get to the actual customization. Maybe you buy a custom API on top of your BC365 platform. Maybe your C-levels deciced that paying €50k a year to avoid outages on major updates isn't worth the cost. Then when things predictably and completely avoidable fail you're going to hav to deal with the literal shitstorm. You'd think that all the people being locked out of their jobs and the €150k cost of getting an immediate and prioritised update to the system would mean you'd start paying for that $50k service after this. You'd be wrong. Ok, to be fair, in this particular example it would be a different scenario. For a small system like this you'd find a cheap consultant house in your area and get them to build the customization for you. Only they would outsource it to some solo developer who will build it in a way that basically requires that specific person to alter it. Then when it breaks or needs to be customized futher a year down the line, that person is no longer a solo developer. So you reach out to another cheap consultant house and do it all over again, from scratch.
This doesn't even mention how poorly all those 300 off the shelf systems work together. I mean, I don't maintain a SDK delivering a way to use Apache-Arrow to write and read parquet files from our datalake in the same manner for fun. I do it because those 600 container apps which basically simply translate data from one system to another need it to be as slim as possible.
for an administrative tool the criticality is not determined by the network boundary but by which decisions within the plant are based on these figures
With the models we got at out disposal, you can easily deploy a kickass internal tool, it's fine even for mvp's unless you're not handling sensitive user data, shooting yourself in the foot is easier than ever, and talking about code smell for most application (especially frontend) does it even matter?
I now started to us AI to help review my juniors PRs, because I couldn't keep up with the amount of code they ship. It started poorly, but now I have my method: I first read the code and flag the lines I'm not sure about, then ask any frontier model (I like Claude here for analysis, even if I don't use it for the rest) to explain the PR and to put effort on the parts I flagged (basically explain in detail the code, not only the PR), and to search through the libraries. Sometimes it notices something I would have missed (like missing an 'order_by' or off by one errors, because the underlying lib wasn't coded like the original AI pretended it was).
I also changed the way I do review because it has been more than a year and the juniors/new hire are still lost, wether on domain knowledge for the older new hire, or just capabilities for the juniors, and discussing with other departments, it's the same for like 95% of them. Now, rather than correcting the PR or adding a request for change, I add a whole unit/functional test to the PR and let that as an exercise to pass the test. They can use AI but I tell them to try to find what part of the code doesn't work before generating the fix, hopefully they'll take ownership of the code if I keep doing that.
>I tell them to try to find what part of the code doesn't work before generating the fix
this makes me kinda sad. they don't do this on their own? are they not even a little curious
even before AI very few developers cared about the actual craft. I'd say only about 3%. For a lot of people it's just a job, no matter if it's a corperate dev or a startup dev.
One of my customers now added a policy that junior developers may use AI assistance, but are expected to write code by hand. Only proven developers can use AI to generate code, and are then expected to understand the generated code, and take ownership.
> are they not even a little curious
Some are. Some aren't. Usually a good indicator of career trajectory.
It's a different mindset. Has nothing to do with curiosity.
For me and you, "fixing" something means finding which assumption was violated and redesigning the solution in light of that. Sometimes this means adding code, sometimes it means making the solution smaller.
For a lot of people, "fixing" means adding code to make it work. This always makes the solution larger.
The latter group can still be curious! But they're curious about which addition makes it work, not about their incorrect beliefs.
An engineer that refuses to understand how their craft is constructed is a poor engineer.
> For a lot of people, "fixing" means adding code to make it work.
What if the fix requires deleting a single line of code?
Is the additive fix to enshroud it in if (0) { } ?
What I saw in these cases, the result was a few levels removed from the root cause and they'd add the same workaround to like 5 different places (and miss one or two) instead of tracking it back to the origin where the fix was that simple.
Usually something more subtle like adding an early return somewhere before the offending line, or inheriting from that class and blanking that method, or something along those lines.
Why are you paying them if they insist on remaining meat proxies?
"because the underlying lib wasn't coded like the original AI pretended it was"
So, the models ARE dumb. They just are very good at finding patterns in their training data. I mean, when they code minecraft clones, it is not because they can cook up how to write minecraft clones, rather, their training data includes a lot of minecraft-like games code, and they just reuse that.
Yes this is the reason they fail miserably when it comes to novel problems.
Wow, that painful AI aided review process, it's like you are computering with eyes closed.
where do u work that even hires juniors lol
[dead]
Bit of a humbling/jarring moment when I realized that people are doing real paid work using LLMs that they could not otherwise do. I mean, it's quite obvious I suppose. But up until now I just assumed it was only a (massive) catalyst for things people would already be able to do with enough time. But nope -- it seems people are right now employed in roles that they would not be able to fulfil the tasks within if AI wasn't there telling them what to write/say/produce. Nobody is really going to come out and say that ... it's not something the less-AI-literate superiors would take kindly to.
I mentioned this then other day, my university class had maybe 80% of people who could not code. No joke. And a lot have gone onto to code professionally.
AI is a godsend to this cohort of code monkies.
Edit: we graduated in 2006.
At a FAANG adjacent company in 2017 we hired a guy from Google with 4 years experience who passed our very hard coding interviews. He literally couldnt code. It blew my mind.
I’m sure any profession has this issue.
My families business is bookkeeping, they hired someone who graduated with an accounting degree. They didn’t grasp the basics of bookkeeping nor accounting once employed.
You’d think Google who my understanding pride themselves on technical chops would weed this out with all the interviews they do. But maybe it’s just big places have more cracks for people slip into.
I’ve long been curious what you (and a lot of people that express this) feel the cutoff is to know how to code? Because obviously there’s a sliding scale between truly knowing very little, being able to adapt similar work, extensive modifications, coding everything from scratch, etc. on top of understanding certain programming concepts and organization.
Can you build a desktop/terminal app, web app/site, mobile app, or even a library without engaging AI is probably where I’d draw the line. Not everything you touch when you know how to code is doing the whole thing, but I’d expect you’d be able to piece it all together with some time and docs. Pick your own platform that suits too, I’m not saying iOS or Windows only.
For context to my above comment re university course. The final project we had to do was make a very rudimentary room reservation server/client on the terminal. Java, however you wanted to store the data was fine, TCP sockets were the parameters given.
It took me all of a week to finish it. Lots struggled to even begin. We were given 10 weeks.
I didn’t know much then but I read the docs and figured it out.
> Can you build a desktop/terminal app, web app/site, mobile app, or even a library without engaging AI is probably where I’d draw the line.
Wasn't the workflow there always to take the example code of whatever library or framework it was and then customize it?
I could definitely never start a react or android or c# or whatever application from scratch with no resources. Nor would I want to because why.
Works with C I guess though. Given that the contract is so small.
But that also breaks down the moment you want to write C for some uC, and then there's tons of boilerplate again you just pull from the sdk example.
Yeah, not the one you are responding to, but I wouldn't define it as from scratch.
Personally, "know how to code" is basically having the logic an reasoning skills to read and understand and apply documentation for whatever subject you need for this job.
If you have never written a web app in your life, but you can follow next.js tutorials and customize them beyond the tutorials afterwards, then you're at the point where the only thing you need is more time or more docs based on what you need to do at work.
You hit the nail on the head friend :)
> Can you build a desktop/terminal app, web app/site, mobile app, or even a library without engaging AI
If companies would start hiring people who are fully experienced and qualified but can't do the monkey dance of performing live coding, maybe they would find these types of programmers again...
You can learn to code in a few months in your free time and get good in your job. That's different from refusing to learn it, because you outsource it.
Also coding is not that hard. You kinda only write the algorithms down you make up in your head. And you do have an algorithmic understanding when you graduate. You also invented some/several languages in during your studies. The learning an actual developed language IS learning, but it's just boring learning, nothing compared to what you did for your exams.
Cool story, glad it worked out for you.
That's not my story at all, almost nothing of that applies to me.
Sure, with LLMs, one can naturally tailor this much closer to the current need (or at least the need one thinks they have) and iterate ("spew") faster, but it's not a new phenomenon in general.
I think this is fine as long as the success criteria are strict enough that they are forced to learn something in the end.
Any significant testing will inevitably create that situation. The LLM won't have enough context to handle the more precise business requirements. The dev will have to read the code carefully and make their changes by hand. Additional rounds of testing may cause thrashing between regressed states until something clicks for the developer. That lightbulb going off is called "learning" and they are human after all!
> I think this is fine as long as the success criteria are strict enough that they are forced to learn something in the end.
They are fired, get promoted to management, or learn the technical aspects.
Promotion into management isn't so different. They still experience similar thrashing and have to learn or be fired.
My point was not only that so much human learning happens out-of-band, but that it has to be fundamentally different from how an LLM builds context. I've never seen an LLM overcome the thrashing on its own. There's never any "lightbulb moment".
Yeh but I also mean outside of software engineering. Within the gamut of 'being a programmer' it seems fair to bleed into adjacent areas without too much cheek. We've all done it; it's part of the learning curve. But I was talking more about people in other knowledge work who have to come up with a lot of prose-like material about {insert thing}. Marketing, consultants, PMs, or even domain-specific analysts, .. ya know, the types of office jobs where people basically write emails, attend meetings, discuss reports, and produce mostly text-or-data artefacts all day long.
That sounds like fraud. You claimed you have knowledge of the job you got hired for, but you don't actually have.
If the company want's hire someone who doesn't have a clue and only uses SO, that's of course fine, but I doubt, that this is the case.
People gotta eat.
All people do, only few resort to fraud. If you start with that argument, you can rationalize any crime.
“Fake it ‘till you make it” is not a venerable mafia saying, if you catch my meaning.
But if the AI has a blindspot then they will fail hard since they rely on the LLM for everything.
aka there are a lot more bullshit artists around these days
i know of several engineers who produce absolute slop and who probably would have produced nothing at all in pre AI times (which would have been preferable) and probably let go or never hired (even better).
they impose such an enormous drag on productivity that they more than wipe out any gains from people using the tools responsibly.
All those programmers who could not write fizzbuzz can ship something now.
I started manual coding again at least 1 day a week.
Its good for the brain, but man it's really slow.
And the AI results prove it: Over the past couple years I've been able to bring several work projects to completion single handedly.
Our R&D company is now handling more clients and we are returning better outcomes/products faster than before.
So...what to do? Results don't lie.
However I will say LLM's can't write multi threaded double buffers by them selves. Probably because they are so application specific.
But I think the reason I can go do fast is because I went so slow before. I have an intuition I can transfer into an LLM. Question is if that intuition will become sloppy as time goes on...
“I wanted to see if I can get in on this 10x magic. I decided to put my money where my mouth was and I bought $10 worth of DeepSeek credits to use with a project I was working on.“
Wow. I guess that’s the punchline!
I'm sure DeepSeek isn't the point here. You can change the name to whatever you prefer and the article still holds.
Actually, I think the author put DeepSeek on purpose to avoid the obvious ChatGPT/Claude comparison — because whatever he chose, there would be a question of why model A and not B, while the point of the article isn't about models comparison at all.
It is the point. Also open source model enthusiast tell you otherwise, there is a coding quality gap between these models. If I use DeepSeek, I do so knowing that I have to limit to simpler tasks on smaller, well specified prompts. What the author did, letting the model do the planning, is not something DeepSeek will excel at. I'm using GPT (Terra, Sol, Luna), Claude (Opus 5, Fable), Qwen 3.8 and GLM 5.3 Flash daily and have to vary which model I use where because there's a huge intelligence step function difference here. That's why this article is so useless:
Imagine someone trying to make the case that riding bicycles is a terrible experience and their whole argument is that they took a random cheapo bike with flat tires and rode it for 3min and that wasn't fun. Sure, but if you buy a 25k carbon bike you will have a different experience. I'd not trust that person. If someone told me they have 10 bikes they ride daily and can explain the differences, in detail, between their bikes, and what they excel at. I'd trust that person's opinion.
There’s a night-and-day difference between frontier and budget models, no question. But the issue isn't the tooling at all : if you put someone who doesn't know the rules of the road on a $15k carbon road bike, they're just gonna slam into a telephone pole at 30 mph instead of 6 mph
Excellent analogy. This paragraph invalidates the entire post and honestly just looks lazy. The author may be right anyway, but with that level of experience with these tools, he is really just guessing.
Maybe the failure is in trying only one model / one prompt.
Right. Better throw money at 5 different ones and then people come and tell you that you just need MORE agents and throw MORE money at it or you're not doing it right.
I am convinced coding agent makes me extra productive.
I'm also convinced the effect would not be there If I had a 10$ budget.
> I'm also convinced the effect would not be there If I had a 10$ budget.
Sure it would be. I pay $10/month to OpenCode for a Go subscription, it's fine for day-to-day coding tasks. I wouldn't necessarily try and one-shot a production app on that budget, but with decent planning and test-driven-development, it gets the job done
The cheapest coding subscription is $20 and you can get a refund if you cancel within 7 days.
Actually, I lied, Codex is free and I developed my first AI written application using it and the free tier limits were generous enough to work on it for several months.
Hey! I'm building www.freepi.ai which is free! So you can also try that!
I think the point is that $10 isn’t exactly a lot of money to put where your mouth is, nor a serious effort to see if it works.
I got here and... I agreed with everything up to this point.
The author makes a good point. If you don't know what you're doing, AI accelerates that. No question.
But they put a whopping... ten bucks into using DeepSeek and weren't impressed with the initial results.
I know they try to cover this with "you just aren't prompting correctly!" but if, in 2026, you aren't able to have an LLM generate decent quality code... IDK what to tell you. Good luck I guess?
> in 2026, you aren't able to have an LLM generate decent quality code
Nobody is. The only difference is that some of us are willing to say the emperor has no clothes.
The workflow isnt a single prompt then you have working code from the LLM, its 30 prompts of discussion and then reading each proposed edit and having tangent discussions until you're somewhat ok with it. Then reviewing the changeset as a whole and refactoring a few times
I am.
> but if, in 2026, you aren't able to have an LLM generate decent quality code... IDK what to tell you.
I think the author's point still stands. At some point you need to determine whether the output is of decent quality. A lot of people don't have the ability and experience to do that.
In general, the frontier models are not capable of reliably authoring non-trivial code without careful oversight yet. They are great at producing code that can pass tests, but not neccessarily a code review. This means if you care about code quality you still need a human in a loop understanding what has been done, and that becomes the bottleneck. And less disciplined folks will indeed become increasingly dependent.
However, over time the complexity of problems where you can get away with less/no oversight is increasing. And the models are already great at solving certain classes of problems where one doesn't really care that much about code quality, that wouldn't have even been attempted in a pre-LLM world. Over the weekend I was using Claude to add features to the compiled (no source available) firmware of one of my audio devices, adding workflow features by patching assembly and custom DSP code.
In coding, as with other areas, what's emerging is jagged intelligence.
> Uses only DeepSeek and comes to the conclusion that LLM's are bad at coding?
Why not use actual frontier models, and you know do some real research, before writing a blog post?
DeepSeek has been the most used model on open router by far, only temporarily overtaken by ox alpha when it was free, so it's the most representative experience
Ryanair is Europe's most popular budget airline. If I fly with them once can I write a blog post now calling all air travel miserable?
My point is, there is a monumental difference between flying Ryanair and Emirates First Class for example.
Caviar is better in the sky, it turns out.
[dead]
> Why not use actual frontier models, and you know do some real research, before writing a blog post?
Seems like they've already done more research than you. Their results not matching your expectations does not indicate a lack of research. If you feel they missed something, why don't you go spend your own money on whatever frontier model you want and go publish a post with your own conclusions.
What expectations do I have? They used one AI model and then wrote a blog post as if they have an informed opinion. It has no credibility whatsoever, why would I listen to anyone's opinion who hasn't done proper research.
It's like playing one video game and then claiming all video games are boring and have bad gameplay and stories.
> do some real research
It is funny that only real research on productivity gains from AI shows at best very minimal gains, but AI bros will always tell you "no no no, you have used wrong model, try a different one, there are more of them, you have to try, trust me" and call that a "research".
Show me the research you are talking about.
Ah yes I'm an AI bro because I said the author must do thorough testing before coming to a conclusion about LLM's.
Looking forward to the authors next blog post on how after driving one car they find that all cars are slow and uncomfortable. Then the follow up after trying one phone and that all phones have bad cameras and battery life.
It doesn't fit the narrative that they need to adhere to. I'm a skeptic through and through, but yeah... This article reads like propaganda.
Pre-AI, the limiter on my bad code was how much code I could type in a day.....
Now with AI I can 10x my output and 10x my bad code!
Whoever is giving a bad coder resources and access will eventually stop
Not at any decently sized company. Delivering features is all that matters. By the time you're cleaning up the buggy mess, they were promoted two times and work somewhere else now. If anything, AI makes it much easier to suck and still get things done to an acceptable degree.
Nope. That would require the people paying for the code (those requesting the projects) to see something wrong with the code. As a coder, you would look in an see a mess. As a user they'll probably see something relatively nice, with some maybe strange behavior and non ideal, but what software sin't non ideal? It does most of what we wanted, and you finished it how fast!?
I think you underestimate the ... lack of caring by most Director/VP/CxO folks
'Or else admit this is a dopamine game that makes you feel like The Universe's Most Special Programmer™ when it's really just gamified mass-scale intellectual dependency.'
Disesdi Shoshana Cox
Alternatively, AI teaches you lessons about technical debt, code maintainability, and architecture faster too. Traditionally it took at least one or two years to really bump into those problems. Now any student can get into these problems within one or two weeks.
> “after four years of open source LLMs, we should have three AirBnBs, two Stripes, and three Dropboxes thanks to the power of AI.”
Airbnb, Stripe and Dropbox were created in a different time when the market was much less competitive.
Saturation of software development velocity doesn’t increase large scale product opportunities in the market. It can also mean that opportunities get filled even more quickly by niche players, and nobody gets to grow to Airbnb scale.
IMO the latter is what’s currently happening. AI-powered companies are like little mammals scurrying around between the feet of the dinosaurs, and commentators like the OP look at the evolution of the brontosaurus as evidence that the mammals don’t seem to be growing as they should.
It’s laughable to compare fundraising today to 15 years ago.
You won’t get capital let alone VC if you’re not AI.
It’s infected everything much like crypto did just 3-4 years ago.
There are some good points, and I ask the question of where is the ground breaking stuff myself, but severely weakened by
* stretching the timeline: the actual real programming ability appeared in LLMs in the last 6-8 months, not 3-4 years,
* using the weakest possible tool: and I bought $10 worth of DeepSeek credits that is a far cry from Claude with Fable.
Also, I know nothing about marathons but for most uses putting the app, database, and background processes on the same server is very much the right starting point. With the next steps being employing Cloudflare or similar solutions long before managing a fleet of servers.
I remember being told in early 2025 that only now has "real programming ability" appeared in LLMs.
And since then there's been 3 more "now real programming ability has been made available and previous stuff was just toy examples" cycles (summer 2025, winter 2025 and spring 2026)
Looking forward to the next "everything before this was trivial and bad, here's the good stuff" moment
If you haven't been using the latest NanoQwen27B-5.2.10-Turbo-KimiGLM-Pro-10.4 in the past 13 hours, your information on what LLMs can do is outdated and you're falling behind. Check your priors.
It's a bit Goomba Fallacy. For different people the threshold was crossed at different times. That you're getting told different things by different people doesn't make those statements self-contradictory.
Yeah, it's interesting how the current thing is always the end of programming, but then when a new thing comes it turns into absolute trash that can't be used for anything.
Even as someone using AI on the regular I'm starting to hate the "You didn't actually use this exact most expensive model so your point is invalid" argument.
This is fair to say if someones last experience with AI was copy-pasting code into GPT3 chat windows years ago, but Deepseek is a more than capabale model and enough for someone to get an informed opinion about the technology.
If people have actual counter argument, use those. And if some of those counter argument are "What you say isn't possible, the neweste model can do and here are examples of that", that is fine.
But a blanket "Nuh-uh, it wasn't Model X" is not only a poor argument but also automatically invalidates any criticism when a new, better model comes out - and that can't be the basis of a good argument.
In this case, the argument is valid though. Only recently have some of the models become powerful enough (for some) to actually be useful in day to day programming, without too much hand-holding. For most, this change occurred with the introduction of Claude Opus, and OpenAI and Google have caught up. Unfortunately, none of the "open" models is at this level yet.
The field is moving fast, and asking for scientific arguments is not realistic. It takes an extreme amount of effort to show what exactly is different.
We were in a similar position with static vs dynamic typing for decades. There is still no scientific proof that one is better than the other, but it is quite obvious to professionals which flavor works better in a given situation.
So, even though the argument might be sloppy, I subscribe to it. Using DeepSeek to dismiss better models is the bad argument here.
Edit: added "(for some)" as a disclaimer that you still need to be a fairly decent programmer to actually benefit.
> Only recently have some of the models become powerful enough (for some) to actually be useful in day to day programming, without too much hand-holding.
That still is very much not the case. Every model, yes, including $most_recent_model, needs supervision or you will get burned.
It’s always static just FYI.
The models got good starting with 2026 and that isn't some attempt at excusing it. Companies like OpenAI started building dedicated models around their coding harness called Codex, there was gpt-5.3-codex and it was both cheaper and better at using the harness than the regular models. Then they started merging the two model types into their main release models. All of this happened like 6 months ago.
You don't have to pay money to use Codex, there is a very generous free tier that costs you nothing, you just have to accept being told you're out of tokens every day. Because your token limits are low, you need to make sure that you accept or reject everything manually and when it tells you that it wants to run a command you have to paste in the command into your terminal and only paste the relevant output back otherwise it floods the context window.
> The models got good starting with 2026 and that isn't some attempt at excusing it.
No, they still aren't good.
> Even as someone using AI on the regular I'm starting to hate the "You didn't actually use this exact most expensive model so your point is invalid" argument.
What the author of the article is doing is dismissing a technology so disruptive that it's basically all everyone's talking about in the "tech space" at the moment (I mean look at HN frontpage for the past few months), by trying a relatively mediocre (but still quite good) model for about 10 seconds.
The reality is that frontier models suddenly got very good in the past 3-6 months. It has it's problems, and you need to learn how to use this new tool (as with any tool).
But models can and do generate good code. They also can and do generate absolute garbage (even Fable).
You need a good harness, tools to help the model check it's own output, good context, and a good idea of what you actually want. If you have those 4 things, the chances of generating absolute garbage are pretty slim (but yeah, still there).
Is the issue that nobody has come up with the idea for the great ground breaking ideas?
Or B that the ideas are there but are not get released as the code is ai slop?
I wish I could quote so much from this. But it encapsulates my issue with AI. Yes it can help. But it's not the end all be all. We still need smart people to make decisions.
Also, AI was build on what people figured out before. What happens when people stop figuring things out? Or stop making it available to AI? We hit a wall.
AI is like a tool used by mega-corporations to change the world according to what they want. We see this with the increase in RAM prices. I don't want to pay the overprice here - AI companies owe us money. People seem still mostly in the AI hype phase, but a lot more criticism has amplified in the last months. It is only a matter of time until the hype phase is over.
His AI experience is $10 of DeepSeek - but I think he is right anyway in the main points.
You can most definitely ship crap much faster than you used to. It's obvious to anyone when you're shipping crap. And it seems everywhere I look people are shipping crap. Both software and writing.
The hardest part of shipping quality software is not and has never been "writing the code". The hard parts are product taste, architecture and ensuring your product actually solves the problems it should in an efficient and secure manner.
If you don't have an intuition of those, you will likely still be shipping crap. Just faster.
Yep, using AI to remove your thinking process and judgement, and just let it generate stuff as fast as possible, will definitely make you ship crap faster! If you did the same thing with a junior developer you would get the same result.
Where AI shines for me is accelerating the learning and exploration process. I can get up to speed with new tech fast. It is good at spotting issues in designs and code. It can knock out quick tests or benchmarks to support me. The quality of what I can produce with AI support is much higher than I could without it.
So it really just depends on what you use it for. If the goal is "replace humans and ship fast" that's one thing. If the goal is "explore the problem space in greater depth", it's another.
> Most apps these days gather credit card details
That's one of the few consumer-facing areas where there are still standards in place, namely PCI-DSS. As far as I know the audits require the name of a human who is responsible for payment security. Card companies can one-hit kill your startup if you're breaking those rules (maybe purely blockchain startups are exempt).
Yes, you can offload this to stripe, but then your app should never see the card number and certainly not the CVV. You end up storing these, even by accident, both stripe and the card companies will hate you.
Speaking anecdotally, at least: I don't think the lack of publicly available software is necessarily evidence that AI tools are ineffective.
From the limited perspective I have as a senior developer working in industry, I think that:
1. It has been a relatively short amount of time since the November 2025 quality increase in frontier LLMs. 2. It takes time for developers to learn how to use agents effectively, once they begin to trust them. 3. There is a significant difference in the output quality dependent on how you guide the agent. 4. Experienced developers are creating lots of quality software that is used internally (or only personally by them) but which they don't have the time or confidence to release for others to use.
I largely agree with this piece but the author isn't really giving the AI development process a fair shot. It's not a matter of "prompt better" but of establishing processes that account for and correct agent shortcomings.
For me, it's taken thousands of dollars and months of experiments to get a process that produces working and maintainable code. You can't just go in with $10 and be like "Wow it doesn't work" after playing around with it for 30 seconds, which is how far $10 takes you on models that are doing work. All that does is expose the shortcomings that need correction.
You can get around these not by just "prompting better", but by establishing rules and pipelines that are enforced with standard software engineering tools like compilers, linters, type checkers, borrow checkers, profilers, benchmarking, smoke tests, proof assistants, and so forth. Those things we know work really well, so AI coding shouldn't be viewed as AI doing all the work, but AI orchestrating all of the tools that we know work so well in software development.
The magic of agentic engineering is only possible because our software tools are so good not because agents are.
It almost feels like ages ago when the promise of AI was to be a sponge of knowledge, to help us dig in quicker and speed up the search process of looking up implementations on Google / Stackoverflow because the expectation was that we authored code and demonstrate it to our peers it was correct as the role of an earnest code contributor.
You cannot overstate how disruptive agentic powered development has had on the startup scene. There is clearly an embrace or loosening of that expectation which is upending the profession and also paradoxically giving superpowers to us who have been in the business a long time.
It’s hard to know where this is all going as we are being challenged by this transformative, powerful, yet problematic technology every day at work.
This made me think of "1,000 True Fans". I'm sure there will be more than a few business school case studies written about competition in this era of software development. Even things like niche apps singularly tailored for small skilled trades are seeing competition. AI tooling will definitely allow new/smaller/solo players to clone apps and undercut the legacy players on price more rapidly than before. Some of these very narrow apps are fairly expensive ($100s/month), if replacements of acceptable quality can be produced they will take some market share. This may be death (shrinkage) by a 1000 cuts for some established players.
> Borrowing Disesdi's logic in their arithmetic (which is hard to argue against), after four years of open source LLMs, we should have three AirBnBs,1 two Stripes,2 and three Dropboxes3 thanks to the power of AI. > > So. Where the fuck are they?
I think this is a great question, and I have a completely unqualified theory.
I work in a fairly specialised field, I've notice when I use AI to try to generate code in fields where I'm very experienced, it produces a worse output at a slower rate than I can produce manually[0].
When I use it in fields I'm not an expert in by any stretch, like webdev, genAI massively improves my output by a huge factor.
I think this makes the productivity gains a little fuzzy. Claude code makes me 10x faster at webdev, but I'd be a very slow webdev. Ultimately, the next innovation in a field, will probably be from an expert in that field, and they therefore won't see the kind of "10x productivity boost" that non-experts do.
[0] I still find AI helpful for exploring code etc, just not so much producing output.
> we should have at least a handful of AirBnBs or Dropboxes
Neither was ever about some bottleneck on pumping out code and everything about marketing and network effects
Previously, the limit on bad code entering any serious product was gated by having devs that at least knew how to change what they copied from SO.
Seriously, that was the lowest level of skill I saw in 30 years of development.
Now we have people who can't even do that proudly showing off PRs to widely used products.
"B-b-but I do the systems design and hard thinking".
Sure, buddy.
At least once a week I have to drop what I'm doing and race in to stop the non dev manager from merging a fully vibe coded patch which will blow up production. Claude Code has made this guy a negative 10x developer.
Just let him take down prod on a friday night once and leave him to deal with the PagerDuty alerts solo. Cures the urge to vibe-code infrastructure real fast
People like that are usually pretty good at deflecting onto other people. Likely you will end up being blamed in some way.
how does this non dev manager have perms to do this
They don't have permissions but if I don't rush to block the PR with a bunch of review comments, one of the other AI brained devs will just hit approve on it without reading or understanding anything.
Yes it is hell here. But it's a new kind of problem that just showed up.
I'm experiencing similar rubber stamping around my company.
CEO hired an AI forward CTO / Head of Engineering. He said at his last company his directive was NOT to look at code anymore.
I've watched thousands of lines of misguided, verbose, brittle code (but there's 24 tests that nobody reads) go essentially straight into production. Nobody understands it, we're adding so much trash documentation that is impossible to keep up with.
It's an absolute train wreck, I feel farther away from the system than I did when I started. Commit messages / PR descriptions are to be generated by AI (and pushed).
It's only broken production code a couple of times so far, but my guess is as this continues we will be unable to maintain the codebase without AI involved.
AI can be a force multiplier, we are not using it that way.
My main issue with it is that when a human developer broke the production codebase there is generally an "A-Ha" moment, like, "oh no... I know exactly what that error means, let me run a hotfix and we can change things so that doesn't happen again."
Now we're at the full behest of Claude "do you see what happened to our codebase, robot? Can you help fix it?". Not a fun place to be...
This is so maddening. I'm tech lead and I run a tight ship on the repo I'm in charge of, all PRs unofficially wait for my approval (it's mostly juniors), but if I ever go on PTO, people just approve absolute shit. It is a common occurrence that I come back from vacation and have to get people to revert merges.
He's the manager - that's like corporate root privileges
Manager has rights to approve themselves... Access control working as designed.
At worst, needs another manager to approve. Now they can both saddle up on the velocity/time to market wave... together. The racket begins in earnest.
How does a manager do this, they ask? I answer: the same way they do everything. Lunch. If the effort fails, at least they tried! Deck chairs, Titanic, etc.
I think my 10x comes elsewhere than writing code. For ex llms are much better at bash than me. So good I don't have to leave my shell. Jira? ServiceNow? Ha! Claude you go update that ticket. Go check my history and git commits and write me my standup. I'm not leaving the terminal.
I don't disagree with the title of this weblog post, but its premise is wrong: the existence of AI has an impact on the total number of tech companies, but not on the number of companies that make it real big (the AirBnBs, Stripes and Dropboxes the author is looking for).
I tried really hard to get senior management to understand this where I work. Yes, AI is an accelerator, but that doesn't necessarily mean it's going to accelerate you in the "right" direction.
I have a coffee cup with the writing "Do stupid thing faster with cafe". That's how I feel about myself when I use AI carelessly... The speed with which I can make a mess is astronomical!
When I do use it, I try to only use it to do things I can already do.
I don’t want to solve a new problem with an LLM programming assistant anyways - that’s the fun part of my work, why would I get rid of it?
AI does make me suck more, which is a good thing. It sped me up and has me iterating on things in 3 days that normally would have taken me more than a month to do. Surprise, surprise: I'm discovering that I have a lot of blind spots and false presumptions about what good programming is. My understanding of the craft is patchy, but that wasn't as noticable when going at the speed of human-programming.
It's like cars. If you only ever drive at 30 km/h, a 30 yo car, that had its last oil change in 2008, might be great. But, if, in the same car, you start driving at 200 km/h, you might notice that there's a lot of room for improvement. My programmer-self is the 30 yo car in this metaphor.
> 4. There is, of course, the alternative that leaders are the real bottlenecks. But that can't be right...
Sshh, don't say that part to loud, especially on HN.
How you use AI as a person who can actually code is up to you.
You can use AI in two ways, either…
1. set the plane on autopilot and arrive at the destination having no idea how you got there - like any old idiot who can type a prompt. And also pay a very expensive air fare for this super power auto pilot.
2. You can pilot the plane yourself. Plan the route in advance. Set waypoints. Make decisions along the way. And land the plane manually. And the reward is you know the route and you pay 1/10 the cost.
If one truly believes it's not in a bubble, then borrow and leverage an unlimited amount and bet the house & your family on creating a business with infinite growth and value. Don't know how to do your own prompts? Borrow and then hire others to do it for you. If those who you hire don't know, they can hire others too.*
[* This is not financial advice. Please don't actually do this.]
You could make the converse argument, too: why don't you borrow and leverage and bet and short these AI companies, if you so vehemently think it's a bubble?
The bottom line is that markets are complicated, useful technologies are often accompanied by bubbles, investors are not always rational, and people generally try to make the best decision with the information they have available. The answer is likely somewhere in the middle, but that's a lot more boring and a lot less inflammatory.
> You could make the converse argument, too: why don't you borrow and leverage and bet and short these AI companies, if you so vehemently think it's a bubble?
Because going long and going short are very different. For successfully shorting something you need to have a pretty precise estimate of when the crash will happen. Being a comparatively short time off can cost you everything.
Also, the volatility of leveraging positions often costs more than any gain over the long term.
Most look forward to picking up the assets at a heavy discount. =3
https://www.youtube.com/watch?v=rE75WvOtcu8
Yeah, but that applies to the argument in the original comment too. It's supposedly directed at someone who denies we're in an AI bubble, telling them to put their money where their mouth is. But then suddenly they're supposed to risk literally everything they have on "creating a business with infinite growth and value" before their "unlimited" leveraged loans are called in.
Obviously there's some hyperbole there, and I'm not really sure what serious point was being made, but it does seem to depend on the anti-bubble guy having some very confident and specific beliefs...
My point was rather that it is an asymmetric situation from an investment perspective. I am not taking any sides regarding bubble or not.
You mean for companies that make -$2.50 for every $1 of revenue, and bury their groups $2Tn debt in a footnote?
It is not a question of if the bubble will go, but when... but you are right that Bears or Bulls always get it wrong predicting the future (if they are a legal investor.)
https://www.youtube.com/watch?v=wTiYaWFP59Q
Personally, Shrek movie release correlation with market corrections is funny, and a new film is due June 2027. Please hedge your bets with a diversified portfolio. =3
Good point. There is too much inflammatory rhetoric on here about AI, too many opinions masquerading as facts, too much black-or-white thinking.
While I do believe its a bubble, that seems like a silly viewpoint that presumes infinite risk tolerance.
Going long is not infinite, you only risk what you bet.
He said borrow with infinite leverage.
The Fermis Paradox of AI, 'where are all those companies'
> AI Can Make You Suck Faster Too
Inflatable dolls with AI ? Or vacuum cleaners ?
Why does it have to lead to a 10x increase in revolutionary companies?
Next to maintaining and expanding my own network of websites I 10xd writing boring CRUD applications for companies that were otherwise unable to afford it, making all their employees more productive. There's true economic value in that.
This is fine, but software also needs maintenance, especially security updates. If we have an explosion of new software without the corresponding increase in maintenance work - very likely from clients that pay peanuts for software - the world will be taken over by botnets and malware.
It's going to be a worldwide software experiment.
I just allow access to their application only for their IP address(es) and my deployment server checks if I need to update anything. That's not expensive.
> three AirBnBs, two Stripes, and three Dropboxes
you want more SaaS?
> The chatbot recommended some of the dumbest shit you could possibly do
The "I tried it and it sucked" is borderline conspiratorial at this point. There are enough talented, thoughtful developers saying there's something real here, and it is worth believing them and investing some time to understand it, even if you come out the other side and decide you don't want to use LLMs.
Use a very good model. Set up a good harness. Spend some time on your system prompts and skills. Develop your intuitions about how the model works, what it's good at, how to scope the work, and how to steer it. Recognize when it's alleviating menial work and recognize when it's making choices you really need to understand yourself. Be patient and accept that the failures are going to be very painful for a while.
Don't write it off until you've genuinely seen the upsides.
You describe my experience with Claude (except my gut feeling is that system prompts and especially skills are useless, and I've even done limited testing that agrees with that).
However, your discourse is not the usual borderline religious discourse that's used by LLM advocates. You're not extreme enough either pro or against LLMs :)
Ha, well that's good because I would hate to be labeled an "LLM advocate". I've been on the negative side of the conversation many times, and I think the downsides are very real and often ignored. But calling them useless is just not reality.
As far as system prompts and skills, my approach has been to start with nothing and add very specific instructions as I notice issues or feel like I need them. Mostly around trying to keep it terse, kill the obnoxious rhetoric ("It's not X. It's Y"), and give specific guidance on how to write code. Skills are things like "use edge-tts to generate spoken audio for this answer and send it to me" kind of thing.
> except my gut feeling is that system prompts and especially skills are useless
I tend to agree. Not with the extreme version of this statement: there are some genuinely useful instructions I give in my Claude.md and prompts. Comment style, how much to push back, which subagents to orchestrate for tests, reviews, etc.
But I've tried the single-sentence versions of those and versions with multiple files of long prompts for the orchestrator and its subagents, and I can't tell the difference in output quality. If anything, the shorter version is better
> Comment style, how much to push back, which subagents to orchestrate for tests, reviews, etc.
Are those "skills" or just instructions?
My definition for "skills" is the "you are the greatest software architect ever born" type bullshit.
> Don't write it off until you've genuinely seen the upsides.
This is ridiculous. Basically you're saying "if you try it and find it sucks, you're wrong, keep at it until you change your mind". That isn't a tool at that point, it's a religion.
Of course you don't need to keep trying until you think it's good. But if you haven't found anything it's good at, and meanwhile lots of smart and seemingly-honest people are saying they've found it to be very good at some things, then it's pretty likely that you really are holding it wrong (and perhaps not even trying to hold it right in the first place).
How do you account for the lots of smart and seemingly-honest people who are saying they've earnestly tried it and find it not useful? These are all just personal accounts. The only empirical evidence we have is the output, but there is conspicuously no big successful AI project taking over anything, which is some minor evidence against.
No, they're saying there are upsides, even if you think they are small, or think they don't outweigh the downsides, you at least have to be honest about where the utility is.
No, it just means you have to learn to use the tool well.
[dead]
- its just a next token predictor with a high probability bro https://blog.florianherrengt.com/how-llms-work.html
- when are all the "I am running agents 24x7 a day on max plan" people on HN gonna learn this?
- best case scenario bro: LLMs get infinitely better and nobody needs to code anymore
"software requirement prompting is the new skill" I can write those pretty well bro
- worst case scenario: AI market crash, cognitive debt spikes across every major organization filled with vibe coding juniors that have never spent a single day in their life debugging a production setup without AI.
- Now the whole world is filled with 90% programmers that cannot add 2 numbers in c++ without using GPT
- Guess what bro? I am now one of the most valuable programmers there is :)
If you don’t understand how the sausage was made you have made legacy sausage.
If you don’t understand how the sausage was made, do you even know if it's sausage?
when I tell codex to build me a chat app and it uses gpt-4o
> and ruining the quality of internet search engines.
Tbf, I'm not sure we can really lay this one at the feet of GenAI. The SEO bros had pretty thoroughly ruined search before LLMs took off - the process just accelerated a little at that point.
The code smell in my repos are at an all time high and I'm a senior dev, can't image how worse vibe coders have it.
it hasn't even been a year since Opus 4.5 and my work codebase is thoroughly ensloppified. From the top, the assumption is you don't edit it by hand. It was fine at first but hand editing is becoming more and more painful. LLM coding is a cancer in this way.
I think it depends on what area of the world you work. We recenty had one of our plant managers build a web portal to keep track of some of our operational tech, with features you wouldn't find in standard products. I've been turning it into a container app that can actually deploy safely into our cloud infrastrcture. It's quite frankly better quality than what most external software companies have provided us with in the past.
Don't get me wrong. It's not great. It would never pass any of our policies for things that actually operate stuff on the power grid, but as an administrative tool that can live in total isolation from the vital networks. It's perfectly fine. It's also not like we would have hired the best software companies to build it otherwise. We'd hire some low-level cheap consultant house who would then likely get cheap student labour to build it. With that in mind though, the AI is much better than what the realistic alternative would be.
Money wise it's also cheaper. It's been roughly €1000 + the time it's taken us both. If I had known they were doing it, I would have rolled out the developer cowork app/skills/whateveryoucallconfigurationsthesedays to them. This would have avoided their AI building it to be depoyed on a VM rather than in our managed k8s in our Azure. It would also have written the code a little different, used UV and maybe django rather than flask. But hey. For what it is, it's like a 90% cost saving compared to buying what would've been a less maintainable and lower quality system.
I think perhaps the greater issue will be finding people who want to extract the gold from the heap of shit and getting it to run in production. I don't personally mind, but it's not like any of my colleagues would've wanted the task.
By coincidence, we've been building software for five years that other plant managers are now considering using to keep track of some operational metrics. And I don't think we should put LLMs into our development cycle, simply because the data our software is built for can have critical meaning in some situations.
The insecurity in a vibe-coded web portal isn't that someone hacks it with XSS, it's that after the next vibe-coded release, some X quietly becomes −Y somewhere no one expects.
From this perspective, having no software at all might be better, or as in your case, safer.
I don't think you should keep AI away from software development for security reasons. We're in the European economic area and are completely NIS2 compliant and we use LLM's to aid in our software development for high risk systems. None of that is vibe coded though, so it's not like I disagree with you either.
As you point out this portal isn't that, but what protects us is the processes around compliance. This can't grow from X to Y because not even the CEO has the authority to overwrite our compliance gates. The EU is a tremendous help in this area since personal liability changed things completely.
> Quothling 6 months ago [1] I work in a NIS2 regulated sector and I'm not sure we can ever let any AI agent run in anything we do.
> We're in the European economic area and are completely NIS2 compliant and we use LLM's to aid in our software development for high risk systems.
Please pick one. Either way this is a nightmare level of threat to sovereignty.
1. https://news.ycombinator.com/item?id=47063153
I don't think they are opposite. We still don't let AI run anything we do. That being said, things have changed. 4 months ago I wouldn't touch the M365 Copilot thing if you paid me to, now it's basically the only AI I use professionally. Cowork changed the way we work, and I've previously spoken about how I think Microsoft sort of won the AI enterprise "war" by selling the tools and not the mode. If you'd asked me a year ago I would've been very anti AI in general. I got proven wrong. I do think it's going to be interesting to see where the EU goes with things though. Currently you can't use all features of Cowork as an example, because they are illegal in the EU. So you can't let it run your browser session for you as an example. I wonder how long that will last.
I still doubt we will ever give an AI access to run code on our systems directly though. In isolation, sure, but other than that.
This is a side note, but my personal favorite part of Cowork is that I can roll out our compliance policy to every developer as a Microsoft Teams app (no, that makes no sense to me either). So when they try to install some package that isn't pre-approved their Cowork agent won't let them and will instead explain how they might get approval. If they then continue to reference it, Cowork will even alert us.
Yes cowork is the first part of copilot 365 that's actually useful, to me too. It is however paid separately per token which means you need the expensive 30$ subscription and pay tokens on top of that.
If you just get cowork directly through claude you do get generous usage within their 20$ subscription.
I have to use it through Microsoft too because of their lobbying our company but I don't think they won any war. They're just reselling other people's stuff. The integration with office is alright but I don't really rely on that. In my personal life I avoid them.
What you're telling me now is that in the short period of 4 months, an American corporation took an employee of a company the EU relies on for critical matters from "wouldn't touch it if you paid me" to outsourcing thinking.
Well, I'm speechless.
Are you purposely twisting things to fit a certain narrative?
If you really want to go into it, I've previously talked about how we used AI tools provided to us by one of our major investors who do so for all the companies they are invested into. These were also Anthropic and OpenAI models. The main difference is that Cowork has access to files on a users one drive (and that we get a lot more control over what goes into it).
This is the funny part, because I have nothing to twist.
You work in the extreme opposite setup to mine. VC vs bootstrapped. LLMs vs hand-coded. High margins vs open-source. If you ever look at our codebase, you will see that there are very few code dependencies, and of course bringing in another as large as LLMs is not in scope.
If you wish, we can always continue our exchange by email (mine is in hn profile).
> What you're telling me now is that in the short period of 4 months
Not OP, but I opened the link to the comment and it was six months. I get why you're skeptical, and I think it's fair to point out. But that's sort of glaring when I opened the link to contextualize your comment.
> Either way this is a nightmare level of threat to sovereignty.
I agree, the EU can't become sovereign or have privacy, if the citizens constantly work against it.
Denial of Service is still an issue whether its from an external attack or spaghetti soup code bug. Outcome for your users is the same.
From my perspective it looks like were just allowing hostile developers within our environments now lol.
Your perspective is correct. We call them shadow developers, but it's a massive issue and security threat. The more tools Microsoft adds to their AI admin center (and put behind that ridicilous Agent365 DLC license) the more we see just how big of an issue it has become. People are using their personal credit cards to buy AI tools and use them wildly irresponsible.
In the big threat picture a vibe coded web tool that's not on the internet and runs in total isolation on it's own management group on the "this might get hacked" tenant in Azure is nothing though. I'd worry more about all that OT which was compromised from the factory which sits around in the energy sector. Especially because it's very easy to draw you a risk analysis that will tell you that unless you're running a Nuclear Powerplant then it makes no financial sense to secure your stuff. The audits are so rare and the consequences so low that it's cheaper to just pay the fine (if you ever get one).
Most actual security happens when someone on the ground decides that it's just too stupid that something clearly labeled "DO NOT PUT ON THE INTERNET" was put directly on the internet.
If you aren't using AI to write your code you should definitely be using it to find bugs in the code you write by hand.
I wonder how much of this "simple but bespoke" stuff would be better served with some customized off the shelf software rather than a vibe coded tool.
There's not a lot of software where users dont really care if it goes wrong.
I doubt it would be better served as a customized off the shelf software. They certainly could have done it with our existing systems with SmartSheet, Microsoft Fabric and/or Power Apps with SharePoint for document storage.
From an enterprise perspective this becomes complicated for various reasons. RBAC is one area. In the perfect world you have a system to handle roles and rights to every system, something that you can give managers access to so they can maintain the access available to their employees, something linked with HR. In reality you have EntraID with a hieracy which is sort of automated by HR data, but not really, because sometimes HR puts everyone on the CEO level by mistake, and, if you trusted HR as authoritative that would've just broken all the EU laws. So you have all those Entra groups and you need IT Operations to maintain them and since you want to build it on job roles and not people you'll typically not be able to maintain them in the off the shelf system. Which means that you would have had to build a web portal for the plant managers manager where they could maintain a couple of Entra groups in a web interface. That or you'll have to setup an IT support flow where you add yet another system that IT has to maintain access for.
Then we get to the actual customization. Maybe you buy a custom API on top of your BC365 platform. Maybe your C-levels deciced that paying €50k a year to avoid outages on major updates isn't worth the cost. Then when things predictably and completely avoidable fail you're going to hav to deal with the literal shitstorm. You'd think that all the people being locked out of their jobs and the €150k cost of getting an immediate and prioritised update to the system would mean you'd start paying for that $50k service after this. You'd be wrong. Ok, to be fair, in this particular example it would be a different scenario. For a small system like this you'd find a cheap consultant house in your area and get them to build the customization for you. Only they would outsource it to some solo developer who will build it in a way that basically requires that specific person to alter it. Then when it breaks or needs to be customized futher a year down the line, that person is no longer a solo developer. So you reach out to another cheap consultant house and do it all over again, from scratch.
This doesn't even mention how poorly all those 300 off the shelf systems work together. I mean, I don't maintain a SDK delivering a way to use Apache-Arrow to write and read parquet files from our datalake in the same manner for fun. I do it because those 600 container apps which basically simply translate data from one system to another need it to be as slim as possible.
Am I jaded? Sure. But who isn't in enterprise IT?
for an administrative tool the criticality is not determined by the network boundary but by which decisions within the plant are based on these figures
With the models we got at out disposal, you can easily deploy a kickass internal tool, it's fine even for mvp's unless you're not handling sensitive user data, shooting yourself in the foot is easier than ever, and talking about code smell for most application (especially frontend) does it even matter?
[flagged]
[flagged]
[dead]