(author here) in a way, yes.
you can see how I start with a god-mode euphoria, and slowly realise the vibe is mostly for boilerplates, and very (very) basic tasks.
"I freestyled with absolutely zero planification, and I'm paying the cost now as the product inches toward a production-ready form."
I'd say it helped me transition back into building, and even tho it can be (extremely) painful at times, I still enjoy ‘coding’ this way. These days I probably decide on and closely monitor 95% of the backend (and I'd probably be much faster coding some parts myself), but I still let Claude handle about 80% of the React bullshit I hate.
tbh, it could be a breakthrough in model design, an smart optimization (like the recent DeepConf paper), a more brute force (like the recent CodeMonkeys paper), or a completely new paradigm (at which point we won't even call it an LLM anymore). either way, I believe it's hard to claim this will never happen.
It's pretty easy to understand why it will never happen. AI isn't alive. It's more intellectually akin to a sword or a gun than it is to even the simplest living thing. Nobody has any intention of changing this because there's money to be had selling swords and guns and no money to be had in selling living entities that seek self-presentation and soul
Funny how some people get triggered by this. If you read the whole thing, I actually explain how fucking dumb and clueless LLMs still are once you go past boilerplate. You basically need to keep them in check constantly, even for a pet project like mine. I'm not celebrating anything here, just sharing my journey, the fun and frustrations I've had, going through the stages of vibe-coding: from god-mode euphoria to realizing how deceiving the first rush is.
But I do believe we'll get to the point they will replace more and more engineers. yes, I don't know how fast, I don't know if LLM will be able to reach that point. But eventually, all that money in research will get somewhere I believe.
> You basically need to keep them in check constantly, even for a pet project like mine.
Once you do that, then it is not "vibe coding".
> But I do believe we'll get to the point they will replace more and more engineers.
Well some software engineers will get replaced. However, your claim was "all engineers", which isn't realistic.
Given the amount of safety mission critical software that runs the internet, air traffic control for planes, cars and embedded devices, etc they will always need human software engineers to review, test and maintain all of that, including the Linux kernel itself which runs almost everywhere.
Fully replacing all of them with LLMs would be outright irresponsible.
> But eventually, all that money in research will get somewhere I believe.
Of course, LLM security researchers and consultants breaking vibe-coded apps.
(Author here) Maybe you missed the point of what I wrote. I thought the disclaimer made it clear this is just a tiny project for 3 users only and not something meant to scale :) Is my product inferior to Notion, Slack, etc.? OF COURSE. Do I use Notion extensively? Fuck no. I'm more of a Bear (now Craft!) user, but I needed Notion for a handful of tiny features that Tiptap now gives me. So should I pay $60 per seat for the little I need, and miss out on the fun of building my own tool? I think not. But hey, that's just me :)
it's in Hongdae ("Hongdae T Stay" on Booking). I paid ₩39,000 a night (~23 euros) with a 25 day stay.
I don't think people stay long there, it was mostly foreigners. The bed is like a plank, there is no window. But it was cool, I enjoyed it :)
hey (author here).
obviously, the rest of the content adds a lot of nuance to this statement. it's a bit provocative on purpose.
But in practice, now that I am working with it, what I needed from those tools already works, with no major bugs so far. I haven’t recreated the tools! just the parts I need to able to plug in and plug out features. Also, many of those features are usually available in great libraries (like Tiptap).
No major bugs maybe for you as a user, but I would bet there are some very serious security issues in multiple components. So I hope none of that code is reachable from the outside. Which ofc is already not true as you are processing data from the outside with llms.
I've been in the game for over 25 years and built plenty of architectures with over millions of MAU, so I am not super anxious about what I've built here. But any API I use can be breached. I would say the risk I take is about the same as using Airtable. And sadly, anything can happen. Also, I am not using it for financial transactions, mostly just deal flow material.
Feels like what OpenAI should've shipped in the first place.