@nateberkopec They have opensource repos but they are so full of pure slop covered in AI hype bullshit stars that you couldn't possible figure out if they built anything meaningful.
English
CLAM
2.7K posts
If you don’t have:
a) an open source repo to back up your claims
b) an 8+figure company under your belt
… I do not want to hear your opinion about this. Literally just internet rando noise.
David Cramer@zeeg
Everyone is slowly coming to this realization, and I assure you, no one is running multitudes of agents overnight. No one that is doing anything of substance at least. There _are_ people pretending to be scientists, or fully caught up in their drug infused AI overdose, that think their slop machines are changing the world. They're not tho, and they're just wasting a bunch of money and compute to create a lot of LoC that will just get thrown away. The state of the art is still "can we even one shot a production quality patch that we wont regret later", and its rarer than you'd expect based on discourse.
English
@Star_Knight12 This like literally got disproven and study after study shows that more effort in prompting and more tools actual made the models code worse.
The difference between a good dev and a bad dev is their ability to plan and rreview code architecture. Just like it was before.
English
@doodlestein
The amount of commits across your repos is insane how can you possible be aware of what's actually going on?
Outside of building this flywheel what have you built with the flywheel and have you successfully used it to build or maintain something in production serving real users?
Not hating just curious.
English
Funny how confident people are about things that are readily disproven.
"I can't do it, so no one can."
I'm doing this sort of thing across literally dozens of projects. Over 1,000 GitHub commits every day this week. Yes, I spend ~$12k/month on subs and work a lot. But I do it
David Cramer@zeeg
Everyone is slowly coming to this realization, and I assure you, no one is running multitudes of agents overnight. No one that is doing anything of substance at least. There _are_ people pretending to be scientists, or fully caught up in their drug infused AI overdose, that think their slop machines are changing the world. They're not tho, and they're just wasting a bunch of money and compute to create a lot of LoC that will just get thrown away. The state of the art is still "can we even one shot a production quality patch that we wont regret later", and its rarer than you'd expect based on discourse.
English
GoodAlexander says he pays $160 an hour to run Claude Opus while open source models cost $2 an hour
"Claude Opus is $80 per hour per terminal if you're using it with proper zero data retention through the API, I run two terminals so that's $160 an hour"
"To put that in perspective, the new models on Qwen 3.6 are $2 an hour, you're looking at a 40 to 80x delta and a lot of these models are state of art"
English
CLAM retweetou
Talking to smarter folks than me, I'm convinced many of the AI folks in my timeline are full of shit.
Nobody is "running 20 agents over night" and building stuff for actual users. Maybe some are building internal tools or disposable software. Maybe.
But building software people like using? That doesn't get hacked on day one or blow up after the 3rd user? Nope.
I don't even understand what that's supposed to look like. Do you work out a 57 pages document that perfectly describes what you want to build and then summon 14 agents and have them run wild for 6 hours? And what comes out on the other end isn't a broken pile of shit?
Nope. Not buying it.
PS: it may also be that I have an IQ of 82 and can't figure it out.
English
National Treasury has proposed giving itself the power to expropriate people's crypto assets should they hold more than a certain amount. mybroadband.co.za/news/cryptocur…
English
"His own spend is somewhere around $200-300 a day, and he's still nowhere close to generating the ROI to justify that kind of usage."
I can't fathom two things about this:
1. Why is he willing to spend $9k-$15k per month per developer on a tool that doesn't provide an ROI?
2. How are he and his team managing to spend that much?
Ivan Burazin@ivanburazin
The co-founder of a $3B+ application monitoring platform says they budgeted $15k/month for each developer's Claude Code usage, and it's still not enough. @zeeg revealed that they allocated more money for devtools this year than they ever have in the history of their company. His own spend is somewhere around $200-300 a day, and he's still nowhere close to generating the ROI to justify that kind of usage. But his broader point is that we're still very early. Using the technology, learning it, and understanding the negatives and positives is what's really important at this moment.
English
CLAM retweetou
Cualquiera que trabaje en proyectos reales sufre con esto a diario.
El software que generan los LLMs es una basura, y a más grande el proyecto, peor. Tienes que llevarlos muy de la mano y revisar bien su output.
Pero aquí X está lleno de gente que asegura que ya "cualquiera" puede construir software. Y, curiosamente, el 99% de ellos no trabaja en proyectos mínimamente grandes.
David Fowler@davidfowl
There's a really high cost of duplicating code and we're all relearning that everyday with these coding agents. Until they get better at generating reusable code, bugs will run rampant. You can't "abstract it away" with more agents. You just get crappy software....
Español
@JoeSoapSA @StandardBankZA Do not admit to anything, they are legally obligated to pay you back regardless.
English
I never believed @StandardBankZA was capable of unethical behaviour. A fraudster got past their security and stole R4599 from my card. Nearly got another R14000. They agreed to refund me if I admitted in writing that I received and authorised a OTP, which I did not receive or do. But they will pay me back if I admit I am a thief. Sick!! No integrity!!
English
Couldn’t have said it better myself, if you are the proposed administrator of UBI FUCK YOU. If you have such low self esteem that this is what you feel acceptable for your future and children’s futures FUCK YOU too.
yung macro 宏观年少传奇@apralky
“UBI” is obviously nowhere near the panacea many of you seem to think it is. The median left-leaning Westerner isn’t angry at Elon Musk because he can buy a million times more groceries than them. They aren’t upset with Palantir because Peter Thiel can afford to eat a thousand burgers to their one. This whole thing is in large part post-material. It’s the hierarchy & subordination they’re uncomfortable with. They feel their dignity is being trampled and their autonomy progressively diminished – rightly or wrongly they feel politically disenfranchised and stripped of a say over the future. Offering a guaranteed food budget and a pod to spend the night in return for further disempowerment is incredibly tone-deaf and should be expected to provoke more, not less, outrage.
English
This is a stupid discussion we don’t even know where things are going.
UBI administered by inherently corrupt central power is a dumb fucken idea even if this conversation was relevant.
Value does not come from thin air, you create now value then your issued UBI is of no value.
English
CLAM retweetou
CLAM retweetou
my entire career strategy hangs on a strong belief that we are not going to see fully automated software generation in my lifetime.
but we are going to see an end to the ralph-loop, spec-driven-one-shot-dream, and the "end of white collar work" hype.
we're already seeing some high profile players in software development start to set their coarse along the same path i've been following these past few years. it starts with "wow" then "i can use my skills to fully automate this" and then it proceeds to "fast but no cognitive ownership" then disappointment, confusion, frustration, and ends with "hey, this isn't going to work guys, we need to be more disciplined and look at the code, keep our cognitive ownership, and just use the tools to improve our outcomes. these are not our replacement, these are our accelerators. it's the same story, different tooling".
i'm already there. if a company were to come to me today and say "we tried all the trendy stuff but it just made everything worse, we're losing control of our code base, we need to either ditch these things or make them a power-up" then i am ready for that.
if i am wrong, then so be it. my career is over anyway because i have zero interest in giving up cognitive ownership and responsibility (the ability to respond), while remaining accountable.
i'd rather wash dishes or stack shelves than submit myself to the horrors of remaining accountable without cognitive ownership, agency, and responsibility.
English
@alexwd @ZssBecker @hyros_official Anyone building anything of consequence knows you cannot leave everything an agent does unattended. If your opinion is not this you are clearly not doing that.
Please present your system of consequence that is completely unattended by humans. Otherwise sit down.
English
I’m not sure if you’re an engineer yourself (I’m assuming no), so I think this might explain some of our gap here.
I’m skeptical of your claims here that you have millions of lines of code and someone on your team “knows it like the back of their hand”. Even before AI generated code it’s never really worked this way.
Anyway, I’ll take that as hyperbole (or maybe engineering not being your speciality).
My point here is that this is ~no longer necessary with the right agents and setups.
In the same way that Jeff Bezos doesn’t need to know every line of code in AWS.
At a minimum, you need to consider that as agents/LLMs are getting better at a rapid pace there will at some point be a point where humans reviewing code is unnecessary. And you should be planning and working towards that point. If you’re not, you will be left behind. Full stop. Because humans reviewing code is ~10,000 slower than agents reviewing code. You might not think we’re there yet (I would disagree because I am doing it, but it’s not “easy” yet), but you have to at least accept that we will be there soon.
In that case if you’re not planning to move to a workflow like this you will be outcompeted by a team pumping out quality code at a 10,000x pace compared to your team.
You know, for calculators (or “computers” as they were called there was also a long period where mechanical calculators made mistakes and human calculators checked their work. But eventually calculator machines were reliable enough that humans no longer needed to check their work.
Can you imagine if someone ran a company today where a human, by hand, checked every calculation that their excel spreadsheet did? They would die so fast. Outcompeted by companies who didn’t do such foolish things.
We’re rapidly entering into a world like that with code, don’t get left behind.
English
CLAM retweetou
For all the "SaaS is cooked" 30k lines of code a day vibe coders.
In a picture. This is what's going goin on in your code base every day.
But don't worry, that code it just told you "Is exactly right! Smart move!" is going to work at scale.
English