Zech

Building AI agent orchestration systems. Orchestration is a systems problem, not a prompting problem. A decade of Python. Smarter Dev (13k devs).

[Zech @ZechCodes]

@AlexReibman I only use it for one-off research or answers, stuff I’d rather not create a dedicated thread for. Sometimes I’ll have it monitor/schedule one of those tasks. Not super useful otherwise

[Zech @ZechCodes]

Funny but not everyone wants to (or even can) spend the tokens cloning everything they’d like a better version of.

It’s a weird balance determining what’s worth sharing and not. Always free to throw code on gh with a note that it’s an unsupported, unserious, vibe project intended to be forked.

[Andrej Karpathy @karpathy]

@davegoldblatt "Hi Claude good morning how are you I'm great thx build me a YouTube-style site but restricted to videos that have been repeatedly recommended on HN over the years pls. Don't make any mistakes."

[Andrej Karpathy @karpathy]

@davegoldblatt "Hi Claude good morning how are you I'm great thx build me a YouTube-style site but restricted to videos that have been repeatedly recommended on HN over the years pls. Don't make any mistakes."

[Zech @ZechCodes]

@Hekmatyarkaspe @karpathy @davegoldblatt Practically nothing. Only a few thousand YouTube embed links with comments and metadata fits in a small DB and you only need a few API endpoints and a SPA to serve it. Probably could put it on CloudFlare on the free tier and share it with friends and family.

[Rikiya Takehi @rikiyatakehi]

We asked GPT-6 Astra to edit Qwen3.5-0.8B weights to improve math.
It directly changed only 0.016% of weights, fixing 0.8B's reasoning loops & improving math up to 6x and even instruction following 3x, without gradients or training.

We call this Weight-Space Programming: frontier LLMs can now natively read/write neural network weights! 🧵
Blog: https://t.co/bOWCcFhCVq

[dax @thdxr]

where do i buy exabyte drives i didn't realize these were required when doing rust

[Theo - t3.gg @theo]

The ability to build anything has come with a critical lesson:

Engineers have really, really bad ideas.

[Theo - t3.gg @theo]

Two things can be true at once, and I think the argument is because of people confusing the two.

1. AI has no creative taste and makes slop when asked
2. AI can do incredibly tedious work to make awesome things possible

For a really simple example: having AI write a novel would be terrible. AI translating a novel to a different language with human oversight is much less terrible.

Wish people would stop confusing the two

[Mario Zechner @badlogicgames]

it's beaver o'clock again

[Santiago @svpino]

This is exactly what I see everywhere I go and what I hear from everyone I talk to: they need more people, not fewer.

In fact, many companies are telling me they are stuck because they can't find enough talent in the market.

They are already using the latest-and-greatest AI-thingies out there, but they don't have enough warm bodies to operate them.

These agents aren't magic. They, for the most part, aren't autonomous. They are cool and all, but they need people who know their stuff.

This is even more acute for building software. Yes, anyone can vibe-code simple apps here and there today, but building working software for thousands of user is still very demanding, and agentic coding helps but doesn't work without experts driving it.

[dax @thdxr]

the slop conversation is so one sided, it's not just that you can produce more code

you can also observe, debug, and refactor more code

even if the code is slop you have such good tools to deal with it. even if you write code manually, it's still 100x easier to maintain it

[Theo - t3.gg @theo]

Already see a lot of people discounting Muse Wonder because it wasn’t RL’d in 3d stuff like Astra and Opus.

3d game demos are cool! They don’t represent real code work at all.

Wonder is an incredible code model. Don’t sleep on it.

[Andrej Karpathy @karpathy]

Dusting off this tweet from April because this gap in shared understanding of LLM capability is *widening*. It's now less "two groups of people speaking past each other" and more a sharp funnel.

- Napkin math somewhere around 6B people (~75% of the population) have barely come in contact with LLMs at all.
- Around 1-2B (~20%) are casual and infrequent users of free-tier ChatGPT-like products. This group sees derpy chatbots and treats them a bit like a better Google search, a writing aid, or etc. Maybe an agent tries to book you a flight. I have non-tech friends who (reasonably, imo) say they have not much use for it at all. Even many professionals outside of math&code are in this tier. For example, execs and many other functions spend a lot of their time talking to other people, so while they understand what is happening intellectually, it is still second-hand and a bit abstract.
- Now we get to professional use of frontier-grade LLMs in math&code. Somewhere around 20M people (0.2%) see first-hand that large, complex projects that used to take them weeks/months can now be completed by agents with a prompt. Building apps, copying apps, translating apps, decompiling apps from binaries... This has all happened very quickly and recently - less than 1 year ago, I was writing code manually by hand, typing memorized computer code commands into a code editor character by character, occasionally pressing Tab to autocomplete a little chunk of code.
- And finally we get to the ~5,000 people (~0.00006%) with access to frontier-grade systems internally. The external world has seen the preview. It looks like swarms of thousands of agents collaborating over weeks on software mega projects: minting zero days, running cyber attacks and defenses at machine speeds, discovering new science, advancing the frontier of mathematics. Things that would have taken top professionals in the industry years of work. Meanwhile, human review and comprehension are starting to fall behind. For example, people are still involved in the "archeology" of the OpenAI-HF incident from many months ago. Mathematicians may be poring over the 722 manuscripts on frontier mathematics for a while.

The funnel is driven by a combination of factors:
1. The impact scales with ambition, problem size, and horizon. A question with a paragraph answer barely stresses the system. You need a reservoir of big, difficult problems that you really care about. This aspect drives the consumer / professional dimension of the funnel.
2. The jaggedness of the system (which I have written about a lot separately). Capability peaks in domains that are digital, verifiable and economically valuable. This is because LLM capability emerges from reinforcement learning on verifiable rewards on a curated environment mixture driven by revenue potential. This aspect primarily drives the area (e.g. math&code) dimension of the funnel.
3. Access. Free-tier, paid-tier, internal.

So this is the weirdness of the moment. The general public has mostly not interacted with these systems. When they have, it looks like a derpy chatbot. The majority of professionals still see only a modest uplift. And a small sliver of professionals are experiencing the vertigo of the curve going vertical. And it is all happening at the same time.

[lauren @poteto]

codebases have always contained slop. we have always been okay with this because code rarely needs to be perfect, outside of mission critical "hard" (eg embedded) software.

swe is a unique form of engineering because you're often building something that constantly evolves and is almost organic - user tastes change, your mental models evolve, and features come and go.

admittedly, i and many others who've been coding for a long time were initially (like back in 2024) skeptical that agents would write production code. "but what about all the slop that would enter our beautiful codebase?".

well, even before, we had human slop, and there was a lot of it! if we're being truthful, most people call "code i didn't write" slop. the reality is that 99% of people (i include myself in that number) don't even know what Good Code looks like, let alone be capable of writing it. i would propose that Good Code is:

• correct
• bug free
• fast and cost efficient to run

it has always been exceedingly difficult and expensive to write truly Good Code, because for the longest time (maybe @bendlang will change that), code was not something you could cheaply and formally guarantee to have all 3 of these properties. this is why writing hard software is much more difficult and expensive, because depending on the use case, you may only get one chance and so you have to be right from the start.

but the vast majority of software we use does not need to be Good Code to provide value to users. in fact, it is very likely that most software we use is not formally verifiable! we accepted some bugs in exchange for velocity and progress, so we rely on tests and o11y instead and have adopted a more incremental approach. this is a good thing!

agents help us move faster, and if you're open to the idea, are capable of helping us write better code. it won't be Good Code (that is still expensive), but it is likely to be much better on average at scale.

the new craft of swe is to figure out how to do this. it's your job now! if your productivity with and without agents is not significantly different, you're doing something wrong. but it's not too late to embrace it and use it as an opportunity to learn a whole new skill tree

[Brad Zacher @bradzacher]

Speaking as someone whose career mission has been linting, guardrails, and code quality at massive scales -- I can whole-heartedly vouch for the fact that most of humanity is, in fact, a slop factory and that all codebases have always contained slop.

By-and-large agents aren't much sloppier than the average human now-a-days.

Sure they do generate plenty of garbage -- but you only notice it because you're pair-programming with them. If you pair-programmed with some of the people you work with you'd probably see rounds of slop many of those people also go through before eventually refining and producing a reviewable PR.

I could fill many volumes about the slop I've seen over the last decade. I have seen some of the "war crimes" committed to a codebase by humans. I've also seen a lot of complete and utter filth produced by agents and committed to a codebase!

But the common theme you'd find across all the volumes is this: as disgusting as the code may have been, the code did, in fact, work and it did what it needed to.

[dax @thdxr]

i think something people underestimate is you can get better at vibecoding

like a lot better

there's probably as much a range as between day 1 of programming to day 10,000

[Zech @ZechCodes]

@avrldotdev Nah I just worry others see someone saying “get into infra” as “infra keeps you relevant”

[Zech @ZechCodes]

@jlydev @dadiomov Yep. They feel like they know the material because the homework got done, they don’t realize they’re not building the “muscle.” More feedback = faster iteration on each student’s learning approach. They might even show up to office hours if they know they’re struggling