[babblingfish]

My experience hasn't been LLMs automate coding, just speeds it up. It's like I know what I want the solution to be and I'll describe it to the LLM, usually for specific code blocks at a time, and then build it up block-by-block. When I read hacker news people are talking like it's doing much more than that. It doesn't feel like an automation tool to me at all. It just helps me do what I was gonna do anyways, but without having to look up library function calls and language specific syntax

[Aurornis]

> My experience hasn't been LLMs automate coding, just speeds it up.

This is how basically everyone I know actually uses LLMs.

The whole story about vibecoding and LLMs replacing engineers has become a huge distraction from the really useful discussions to be had. It’s almost impossible to discuss LLMs on HN because everyone is busy attacking the vibecoding strawman all the time.

[miki123211]

As a professional programmer, I think both are useful in different scenarios.

You're maintaining a large, professional codebase? You definitely shouldn't be vibe coding. The fact that some people are is a genuine problem. You want a simple app that you and your friends will use for a few weeks and throw away? Sure, you can probably vibe code something in 2 hours instead of paying for a SaaS. Both have their place.

[falloutx]

> The whole story about vibecoding and LLMs replacing engineers has become a huge distraction

Because the first thing that comes from individual speed up is not engineers making more money but there being less engineers, How much less is the question? Would they be satisfied with 10%, 50% or may be 99%?

[throwaway6734]

I perform software engineering at a research oriented institution and there are some projects I can now prototype without writing a line of code. The productivity benefits are massive

[latexr]

> It’s almost impossible to discuss LLMs on HN because everyone is busy attacking the vibecoding strawman all the time.

What’s “the vibecoding strawman”? There are plenty of people on HN (and elsewhere) repeatedly saying they use LLMs by asking them to “produce full apps in hours instead of weeks” and confirming they don’t read the code.

Just because everyone you personally know does it one way, it doesn’t mean everyone else does it like that.

[kylecazar]

Half strawman -- a mudman, perhaps. Because we're seeing proper experts with credentials jump on the 'shit, AI can do all of this for me' realization blog post train.

[tech_tuna]

There's a crazy amount of hype, fear and blatant lies in the mix. And the pace is absolutely bonkers. The pace of announcements is even more bonkers. Maybe things will settle down to a new normal at some point.

You might think that everyone has FOMO or is an anti-AI Luddite when of course there are a LOT of us somewhere in the middle, just trying to get our work done and trying to figure out what our careers will look like in 5-10 years.

One big thing that no one seems to talk about - GenAI is unlocking many new (and oftentimes "small") business ideas that were not practical just a few years ago. I have witnessed this firsthand. . . however, it will also take away jobs. How many, who knows?

tl;dr everyone is full of shit or selling something or terrified to the point where they can't think straight. And no one has a crystal ball.

[sanderjd]

Yeah I also sense this disconnect between the reality and hype.

In part, I think what people are responding to is the trajectory of the tools. I would agree that they seem to be on an asymptote toward being able to do a lot more things on their own, with a lot less direction. But I also feel like the improvements in that direction are incremental at this point, and it's hard to predict when or if there will be a step change.

But yeah, I'm really not sure I buy this whole thing about orchestrating a symphony of agents or whatever. That isn't what my usage of AI is like, and I'm struggling to see how it would become like that.

But what I am starting to see, is "non-programmers" beginning to realize that they can use these tools to do things for their own work and interests, which they would have previously hired a programmer to do for them, or more likely, just decided it wasn't worth the effort. I think for those people, it does feel like a novel automation tool. It's just that we all already knew how to do this, by writing code. But most people didn't know how to do that. And now they can do a lot more.

And I think this is a genuine step change that will have a big effect on our industry. Personally, I think this is ultimately a very good thing! This is how computers should work, that anybody can use them to automate stuff they want to do. It is not a given that "automating tasks" is something that must be its own distinct (and high paying) career. But like any disruption, it is very reasonable to feel concerned and uncertain about the future when you're right in the thick of it.

[conartist6]

The best advice to juniors is "do not use AI!"

Dunno why the author thinks an AI-enhanced junior can match the "output"of a whole team unless he means in generating lines of code, which is to say tech debt.

Being able to put a lot of words on screen is not the accomplishment in programming. It usually means you've gone completely out of your depth.

[iLoveOncall]

> Dunno why the author thinks an AI-enhanced junior can match the "output"of a whole team

Because the author has a vested interest in peddling this bullshit given he works on Gemini at Google.

[palata]

I think it does both: you can have an LLM automate bad coding (that's the vibe coding part), and you can have an LLM speed up good coding.

Many times, bad code is sufficient. Actually too many times: IMHO that is the reason why the software industry produces lower quality software every year. Bad products are often more profitable than good products. But it's not always for making bad products: sometimes it's totally fine to vibe code a proof or concept or prototype, I would say.

Other times, we really need stable and maintainable code. I don't think we can or want to vibe code that.

LLMs make low-quality coding more accessible, but I don't think they remove the need for high-quality coding. Before LLMs, the fraction of low-quality code was growing already, just because it was already profitable.

An analogy could be buildings: everybody can build a bench that "does the job". Maybe that bench will be broken in 2 months, but right now it works; people can sit on it. But not everybody can build a dam. And if you risk going to jail if your dam collapses, that's a good incentive for not vibe coding it.

[jvans]

i notice a huge difference between working on large systems with lots of microservices and building small apps or tools for myself. The large system work is what you describe, but small apps or tools I resonate with the automate coding crowd.

I've built a few things end to end where I can verify the tool or app does what I want and I haven't seen a single line of the code the LLM wrote. It was a creepy feeling the first time it happened but it's not a workflow I can really use in a lot of my day to day work.

[lovich]

All I know is that firing half my employees and never hiring entry level people again nets me a bonus next quarter.

Not really sure why this article is talking about what happens 2 years from now since that’s 8 times longer than anything anyone with money or power cares about.

[Esophagus4]

Hmmm I know this it’s true because if management only thought quarterly, no one would ever hire anyone. Hiring someone takes 6+ months to pay off as they get up to productivity.

[falloutx]

What a benevolent bossman here, keeping 50% of the jockeys around this quarter. He is probably sacrificing one of his yachts for this.

[petesergeant]

I’m doing both. For production code that I care about, I’m reading every line the LLM writes, correcting it a lot, chatting with an observer LLM who’s checking the work the first LLM and I are writing. It’s speeding stuff up, it also reduces the friction on starting on things. Definitely a time saver.

Then I have some non-trivial side projects where I don’t really care about the code quality, and I’m just letting it run. If I dare look at the code, there’s a bunch of repetition. It rarely gets stuff right the first time, but that’s fine, because it’ll correct it when I tell it it doesn’t work right. Probably full of security holes, code is nasty, but it doesn’t matter for the use-cases I want. I have produced pieces of software here that are actively making my life better, and it’s been mostly unsupervised.

[noufalibrahim]

I'm somewhere in between myself. Before LLMs, I used to block a few sites that distracted me by adding entries in /etc/hosts file to mapping them to 127.0.0.1 on my work machine. I also made the file immutable so that it would take a few steps for me to unblock the sites.

The next step was for me to write a cron job that would reapply the chattr +1 and rewrite the file once in 5 minutes. Sort of an enforcer. I used Claude (web) to write this and cut/pasted it just because I didn't want to bother with bash syntax that I learned and forgot several times.

I then wanted something stronger and looked at publicly available things like pluckeye but they didn't really work the way I wanted. So I tried to write a quick version using Claude (web) and started running it (October 2025). It solved my problem for me.

I wanted a program to use aider on and I started with this. Every time, I needed a feature (e.g. temporary unblocks, prevent tampering and uninstalling, blocking in the browser, violation tracking etc.), I wrote out what I wanted and had the agent do it. OVer the months, it grew to around 4k lines (single file).

Around December, I moved to Claude code from aider and continued doing this. The big task I gave it was to refactor the code into smaller files so that I could manage context better. IT did this well and added tests too. (late December 2025).

I added a helper script to update URLs to block from various sources. Vibe-coded too. Worked fine.

Then, I found it hogging memory because of some crude mistakes I vibe-coded early on fixed that. Cost me around $2 to do so. (Jan 2026).

Then I added support to lock the screen when I crossed a violation threshold. This required some Xlib code to be written. I'm sure I could have written it but it's not really worth it. I know what to do and doing it by hand wouldn't really teach me anything except the innards of a few libraries. I added that.

So, in short, this is something that's 98% AI coded but it genuinely solves a problem for me and has helped me change my behaviour in front of a computer. There are no companies that my research revealed that offer this as a service for Linux. I know what to do but don't have the time write and debug it. With AI, my problem was solved and I have something which is quite valuable to me.

So, while I agree with you that it isn't an "automation tool", the speed and depth which it brings to the environment has opened up possibilities that didn't previously exist. That's the real value and the window through which I'm exploring the whole thing.

[falloutx]

It seems alright, but I wonder if it crashes the economy for vast majority of internet businesses. I personally run some tool websites like ones to convert images, cut videos but the traffic for now seems stable, but my tools don't target devs. Most likely you didnt actually need it, but who am i to judge, I just find myself doing random projects because it "takes less time".

[lightning19]

Interesting, I did something similar but with browser policies on my PC although I just got the LLM to write the json

[antonymoose]

It’s a better Google for me. Instead of searching AWS or StackOverflow it hallucinates a good enough output that I can refactor into an output.

[bryanrasmussen]

The reason why it is better is that with search you have to narrow your search down to a specific part of what you are trying to do, for example if you need a unique id generating function as part of what you are trying to do you first search for that, then if you need to make sure that whatever gets output is responsive 3 columns then you might search for that, and then do code to glue the things together to what you need, with AI you can ask for all of this together, get something that is about what the searched for results would have been, do your glue code and fixes you would normally have done.

It trims the time requirement of a bit of functionality that you might have searched for 4 examples down by the time requirement of 3 of those searches.

It does however remove the benefit of having done the search which might be that you see the various results, and find that a secondary result is better. You no longer get that benefit. Tradeoffs.

[trueismywork]

You can think of LLMs as a higher level language for whatever programming language you are using, but informal with ambiguous grammar.

[noufalibrahim]

I don't think that works. The fact that it can produce different output for the same input, usage of tools etc. don't really fit into the analogy or mental model.

What has worked for me is treating it like an enthusiastic intern with his foot always on the accelerator pedal. I need to steer and manage the brakes otherwise, it'll code itself off a cliff and take my software with it. The most workable thing is a pair programmer. For trivial changes and repeatedly "trying stuff out", you don't need to babysit. For larger pieces, it's good to make each change small and review what it's trying.

[334f905d22bc19]

On the foolishness of "natural language programming". - prof.dr.Edsger W.Dijkstra

https://www.cs.utexas.edu/~EWD/transcriptions/EWD06xx/EWD667...

[christophilus]

And a buggy (non-deterministic) compiler that will occasionally do random things that break your security model, leak sessions, and the like.

[michalsustr]

My experience (with minfx.ai) has been that it is very important to build a system which imposes lots of constraints on the code. The more constrained you can make it, the better. Rust helps a lot in this. Thanks to this, for the first time in my career, I feel like the bigger the system gets, /the easier/ it is to develop, because AI can discover and reuse common components. While human would struggle searching for these and how to use them in a large codebase. Very counter-intuitive!

[barrkel]

If by block by block you mean you stop using an IDE and spend most of your time looking at diffs, sure. Because in a well structured project, that's all you need to do now: maintain a quality bar and ensure Claude doesn't drop the ball.

[Valord]

This is how I use it for work-production code.