Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

Vibe coding and agentic engineering are getting closer than I'd like

simonwillison.net

Post reply on HN
Print view View original

671–680 of 958 posts

Re: Vibe coding and agentic engineering are getting closer than I'd like

#671

Post by cess11 »

"But I’m not reviewing that code. And now I’ve got that feeling of guilt: if I haven’t reviewed the code, is it really responsible for me to use this in production?"

"I know full well that if you ask Claude Code to build a JSON API endpoint that runs a SQL query and outputs the results as JSON, it’s just going to do it right. It’s not going to mess that up. You have it add automated tests, you have it add documentation, you know it’s going to be good."

This really is Wordpress and early PHP all over again, but it's the seasoned folks rather than the amateurs that buy into it.

I believe these tools will be refined and locked down and eventually turn into RAD stuff used by certified enterprise consultants, much like SAP and Salesforce and IBM solutions and so on. From this I come to the conclusion that it is not a good idea to become dependent on them at this stage, which is corroborated by the pecuniary expense as well as excruciatingly fast change in available products.

Re: Vibe coding and agentic engineering are getting closer than I'd like

#672

Post by Finbel »

californical wrote: post #629
Finbel wrote: post #533

Earlier quoted context omitted.

>Also, when did we stop liking to learn? Why is it a bad thing to know all the ins and outs of a programming language? I do not know the inns and out of the assembly layer my high level code end up as. It's not because I don't like to learn, it's because I genuinely don't need to. At a certain level of AI performance, how will this be any different?

Because you may not know the specifics of the assembly being generated, but you’ve likely learned a language built on top of assembly. And the compilers do some great tricks behind the scenes to generate efficient assembly, but those tricks are specifically coupled to semantics of the source language. An LLM is not coupled to anything and can generate output that simply does not relate to the input. This doesn’t happ…

To me this is semantics as far as it's related to "why don't you want to learn?"

I have a co-worker in another team that write java endpoins we consume. I can tell him what I need and I trust the output. I don't need to know java to trust him, it doesn't mean I don't want to learn.

There are thousand examples like this across every stack and abstraction level. From ssh-handshakes to gps.

Sure my co-worker is fundamentally different from a compiler which is fundamentally different from an LLM.

My argument is that the chain-of-trust where you offload knowledge to an external source is identical. We do it all the time but somehow doing it with an LLM means we no longer want to learn?

Re: Vibe coding and agentic engineering are getting closer than I'd like

#673

Post by cultofmetatron »

bamboozled wrote: post #636
cultofmetatron wrote: post #633

2 days ago, we updated a stripe library which broke everything. With AI, I was able to one shot wrapping all of the calls into a shared service, patched the broken api contract across the entire app and got our signup and payment flows working again. solid day and a half of work. this would have taken a days of back and forth debugging previously. AI is not a panacea for everything but its doign valuable work right n…

What does this have to do with the article? I'd say if you're a semi-competent developer, as probably many people reading the article and commenting already are, this comment adds nothing new to the discussion and would already be a very vanilla usage example of "AI". I think the point is that while you can "do things" like extracting the stripe integrations out into their own service in ten minutes, you're not stepp…

> basically you're speed running yourself into harder decisions when using AI.

on the contrary, I freed myself from the burden of having to find all the places in the code base where we used stripe and patched them in one go along with the tests to prevent regressions. That represents DAYS of work that I condensed into a few hours.

who cares if it can't know good structure and how to handle failures? I know how to do that. I have a skills file I created that tells stripe our policy for handling error failures, defaults for structures as well as guidelines for how we should deal with communications between different systems. Before i spent hours building this stuff out. now I just spend 20-30 min reviewing a pr to make sure it follows my directives and move onto other problems.

Thats said, i agree with you on principle. I hand coded an app from a solo dev to now managing a team and gettin ready for an imminent series A. AI doesn't save you from scaling issues, you still need to have a clear idea of what you want from the ai and build processes that give it the context to do its job.

I call that job security :)

Re: Vibe coding and agentic engineering are getting closer than I'd like

#674

Post by skinfaxi »

stephenr wrote: post #616
ditchfieldcaleb wrote: post #495

Earlier quoted context omitted.

I agree with you on everything you said here except: > when you know how the thing works and have that mental context, you will always be faster than an AI That's just plain false, honestly. No one can type at the speed AI can code, even factoring in the time you need to spend to properly write out the spec & design rules the AI needs to follow when implementing your app/feature/whatever. And that gap will only incre…

> LLMs get more intelligent The Spicy Autocomplete koolaid club is out in force today I see. We clearly have different ideas of what the word "intelligent" means.

Explaining your idea of intelligent would have been a better comment than name calling and shallow dismissal.

Re: Vibe coding and agentic engineering are getting closer than I'd like

#675

Post by draxil »

ditchfieldcaleb wrote: post #495
u8 wrote: post #440

The disconnect for AI is that it is a jagged frontier and it only really shines when one of its jagged frontiers extends counter to one of your valleys. If you've been writing Perl for 30 years, you might not want to learn JavaScript just to make a little fun idea in your head to show your wife. Vibe code that shit man. Who cares? Your wife does not care about LOC or those internal design decisions you made. If you'r…

I agree with you on everything you said here except: > when you know how the thing works and have that mental context, you will always be faster than an AI That's just plain false, honestly. No one can type at the speed AI can code, even factoring in the time you need to spend to properly write out the spec & design rules the AI needs to follow when implementing your app/feature/whatever. And that gap will only incre…

Except it's often faster to make the change yourself than explain it to an AI.

Re: Vibe coding and agentic engineering are getting closer than I'd like

#676

Post by fransje26 »

randyrand wrote: post #587
dataviz1000 wrote: post #52

Have you noticed that the coding agents get really close to the solution on the first one shot and then require tons of work to get that last 10% or 5%? If we shift the paradigm of how we approach a coding problem, the coding agents can close that gap. Ten years ago every 10 or 15 minutes I would stop coding and start refactoring, testing, and analyzing making sure everything is perfect before proceeding because a bu…

I can go long session with it making great code. But the first time I say “No, it should be …” it’s nearly game over. If you say it 3+ times in a row, you’re basically doomed. Sure, you can get it to fix the bug, but it comes at the cost of future prompts often barely working.

I second that experience.

The moment I hit the "no, it should be.." point, I know it's the end of it.

Sometimes I can salvage something by asking for a summary of the work and reasoning done, and doing a fresh restart. But often times, it's manual corrections and full restart from there.

Re: Vibe coding and agentic engineering are getting closer than I'd like

#677

Post by parasti »

As a web developer, I feel like this take is wildly optimistic. My remaining qualifications that still provide some sort value are providing historical/business/architectural context to the agent and testing the agent's output. And that's only because 1) it's not all written down in Markdown and 2) the agent is massively nerfed by costs and Anthropic. The thing in the middle where I get a coffee and write code in a variety of languages, then pop open a debugger has been fully obsoleted.

Re: Vibe coding and agentic engineering are getting closer than I'd like

#678

Post by heavyset_go »

user34283 wrote: post #257
xantronix wrote: post #117

Earlier quoted context omitted.

I know there are good uses of LLMs out there. I do. But. The current fever pitch mandates from above seem to want it applied liberally, and pushing back against that is so discouraging and often career-limiting as to wear the fabric of one's psyche threadbare. With all the obvious problems being pointed out to people, there are just as many workarounds; and these workarounds, as is often revealed shortly thereafter,…

In my opinion you are just wrong. It’s an absolute game changer, and it can now multiply your productivity fivefold if it’s a solo greenfield project. Maybe half a year ago it was as you said. You had to wait for the agent to finish, you had to review carefully, and often the result was not that great. You did not save a lot of time. Now I can spin up 3+ parallel conversations in Codex, each in a git worktree. My wor…

All of the "solo green field projects" I let LLMs mostly write, despite supplying the scaffolding, structure and specific implementation details as code, prompts or context, I can't tell you much about 6+ months later, except for the parts I did write.

It's like I never wrote them, because I didn't. I've got the gist of them, but it's the same way I get the gist of something like Numpy: I know how it works theoretically, but certainly not specifically enough to jump in and write some working Fortran that fixes bugs or adds features.

I now have a bunch of stalled projects I'm not very familiar with. I no longer do solo green field projects that way.

Re: Vibe coding and agentic engineering are getting closer than I'd like

#679

Post by simonw »

timr wrote: post #647
simonw wrote: post #623

Earlier quoted context omitted.

41,964 commits is a lot more than "a month of greenfield work". https://tools.simonwillison.net/github-repo-stats?repo=OpenC...

Seriously? Commit count is right up there with lines of code as a classically dumb measurement of productivity.

Sure, but it's still a good counter to "a month of work".

Re: Vibe coding and agentic engineering are getting closer than I'd like

#680

Post by fransje26 »

dataviz1000 wrote: post #52

Have you noticed that the coding agents get really close to the solution on the first one shot and then require tons of work to get that last 10% or 5%? If we shift the paradigm of how we approach a coding problem, the coding agents can close that gap. Ten years ago every 10 or 15 minutes I would stop coding and start refactoring, testing, and analyzing making sure everything is perfect before proceeding because a bu…

> Ah, but you say, that is a lot of work done by a human! That is the whole point. The humans are still needed. The process using the tool like this yields 10x speed at writing code.

Shame that what is left for the humans is the shitty, tedious part of the work.. It reminds me of the quote:

   I want AI to do my laundry and dishes so that I can do art and writing, not for AI to do my art and writing so that I can do laundry and dishes..
Post reply on HN