Agreed. I am impressed by both the satire of this, and the very high-quality implementation. It is so well executed, that it is hard to laugh at the absurdity of the lemmings-like patterns modern AI start-ups have fallen into.
At first I had a good chuckle of “this really encapsulates the tropes, even down to being React” and the more I scrolled through the components, it looks like a very serious library - lots of knobs to turn and consideration for various implementations!
I was going to say that too. Some of these I definitely am guilty of. I have a few dozen that aren't on the list but it's a breath of fresh air to see it so well organized even though, we all know what it is :D fantastic job to the author(s).
This will be really powerful for voice. Being able to reason makes LLM so much smarter but with voice your latency budget is so tight that you can't spare the time typically.
I really love using AI to code but more and more I wonder ... Are things really that different? So I guess I'm the business as usual type.
I think on the frontend side we're going to see a lot more scope for teams.
On a backend infra side it seems as hard as ever. Still have to think really hard problems, think deeply about data structure and flow, and deal with second- and third-order effects. Or even harder because the models like to confidently lie.
The harder question is how we train people but that doesn't seem insurmountable either. Most of us cut our teeth as junior engineers somewhere, implementing tasks that Claude can now do without breaking a sweat but was that really the most efficient way to train and learn?
I work on backend infra. I touch a lot of things and have multiple instances of claude code running at once. I do feel like I'm doing more simultaneously. but that's still more that I have to keep track of and make sense of in my head. claude is making me way more productive, for sure, but at the same time I feel way more overwhelmed than I've ever been.
idk how anyone else is doing it and managing all of this. supposedly there's people with large teams of agents that they can just trust to do everything end-to-end.
I’ve been finding I can mitigate the sense of being overwhelmed by trying to keep my tasks within the same domain or part of the code base. I know it’s not always an option, though.
I guess the goal is to stack context rather than spread it.
I find it incredibly exhausting otherwise. It’s a skill I don’t have in spades. I’ll have to develop it, though.
I don’t think the people trusting agents end to end are getting the quality they think they are. I’m bullish on AI as a permanent and genuinely useful tool in our kit, but I’m not seeing signs that you can actually let them loose at all. Looping on a very well defined problem, sure. But open-ended tasks across large infrastructures and complex domains, no, it’s going to be duct tape and sprawl as far as the IDE can see. It’s going to sprawl faster than context limits can grow, and the mess will only get worse.
From what I can tell, the solution is an ever-expanding roster of agents to make piece work of these immense tasks. The token expense for this approach is insane, though. I used deep-research in Claude Code today and it dispatched 103 agents and consumed something like 3.5MM tokens in ten minutes. That can’t be the future, can it?
Right there with you - this has been my experience as well, to a tee
I can do more now…so I do - it’s really that simple
And it’s way more exhausting because there’s no room to breathe - fresh code to work with every few minutes with prepping for the next set of tasks in between
On the one hand the dopamine’s got me hooked on this like a video game
On the other…I’m as overwhelmed as ever line you said, even if it’s my own doing
It is definitely a different type of stress. I used to lock in on one problem in deep work. Now it’s a constant juggling of 7 problems. It feels good to close one out, but your other 6 keep you perpetually locked in the stress loop.
idk if anyone here is a gamer, but it feels like I'm playing Dynasty Warriors. endless wave after wave after wave of things to do, notifications about fires to put out all over the map, all at once.
Yeah I mean, as far as I can tell the result of the agent mania is the same amount of software, but an acceleration in the decline of performance and quality. I'm also increasingly seeing early adopters going back to a more "traditional" approach to development. So idk, maybe the result will be more jobs fixing up all the vibe code while we transition to a more mature implementation of language models into our workflows.
Reversing to which direction? Because what I've always seen here is a pretty good mix of positive and negative sentiments. Usually we get a lot of AI related submissions, but with skeptics/opposers in the comments.
I’m not sure. I’ve been reading death-of-the-software engineer for years, but recently the -vibe- feels different. I don’t have anything anecdotal to back it up so take it with a grain of salt. I might be reading what I want to see
I'm assuming it's a turn to the negative and not more positivity you're seeing? Geohot's article and Hasimoto's tweet about AI psychosis kind of made me pay attention.
In my opinion, as AI was oversold for too long, it was was easy to dismiss it. Classical image processing was marketed as “AI”. Doomsday predictions about AI seemed laughable, just as SkyNet in the Terminator seemed unrealistic.
The early ChatGPT versions were also pretty silly and equally oversold.
At this point, the popular messaging of AI is still 90% fiction but the remaining real 10% is now a force to be reckoned with.
Companies laying off Indian call center employees to replace with AI is something I never would have dreamed of.
My experience of using AI as a search engine has surprised me. I never expected an overgrown pile of matrices to work that well.
> My experience of using AI as a search engine has surprised me. I never expected an overgrown pile of matrices to work that well.
The first version of Google was also surprising. Mind blowing use of linear algebra (also a pile of matrices, but this time sparse matrices mostly) to rank websites
So maybe the search business was always meant to use pile of matrices
LLMs seem to be best at writing web apps intended to be internal tools. I don’t need to ever really read the code because the functionality is relatively simple- but still valuable. I had Claude Code build me a CLI tool for running common kubectl commands against our EKS cluster. With EKS (AWS) you generally use the AWS CLI to authenticate and then choose the correct EKS cluster and then you can run your kubectl commands. So this CLI tool remembers all the AWS accounts and EKS cluster names and namespaces and pods that I care about. It’s been a huge time saver, and I never would have had the time to build it as part of my day to day. And because it’s an internal tool, I don’t have to worry about things like security/authentication, and any bugs aren’t as big a deal as if they were customer-facing. I wouldn’t want to use it for generating an app that is going on the public internet and doing anything sensitive/important/valuable with my or other people’s data (I’m sure that doesn’t stop others from vibe coding commercial products).
My feelings right now can be described as follows:
If AI progress stalls now or grinds to a halt, we get to keep a lot of new jobs that are going to show up (it opens a lot of doors), while maintaining software devs as an interesting career.
Egoistically, I would love that.
If it keeps going and software devs get replaced, so many jobs are going to disappear.
Even though the synthetic benchmarks paint a picture of LLMs coming a long way since 2022, my practical experience has been that they aren’t tangibly better. No doubt someone reading this will chime in and say LLMs are way better at writing code or whatever, and maybe that’s true, but there’s no difference between ChatGPT 3.5 and Claude Opus 4.8 as far as my trusting the output. Opus 4.8 still messes up plenty. It’s particularly bad with identifying and fixing CI yaml, but it struggles in the usual areas too.
So I’m thinking we’ve just about reached apex with LLMs, and they have failed at replacing software engineers (companies can freeze hiring juniors at their own, future peril using any excuse they like).
Yep, that has been my experience as well. There hasn't been any meaningful improvement in LLMs since ChatGPT first launched. They still fall over, in the same ways, and with more or less the same high rates.
The difference for me has been that you can use llm as your main typing interface, you couldn't do that (without being annoying) before I think opus 4.5
In my experience I'd have to agree. I'm shipping more and I'm onboarding onto domains faster but the same bottlenecks exist, the same complexities creep up and the same talents that help individuals push through are still relevant.
What I like is it's still hard, but you can do things like "prove this using game theory" or "find the optimal value for this and a proof" without having to know game theory or deep math really
How do you ensure that the output isn't BS if you don't know those areas? What value does it add to have a mathematical proof of questionable validity?
if you have an intuition something is true, you can verify the proof. even if you couldn't construct the proof. allowing skepticism to terminate thought is a dead end.
I honestly don't get it - how isn't everyone having a blast with AI? Every one of those side projects you never had time for you can build in a weekend. You can explore five ideas at once. You can do big refactors/cleanups you'd never be able to dream of in the past. As a software engineer it's been fantastic.
I've been using Claude for a side project I've been trying to write for some time. Seeing it generate features at a much faster clip than I can (that actually compiles) has definitely been addictive. However, the joy I used to get from seeing something that I wrote work is all but gone, and it sucks big time to lose that.
The back and forth is more annoying now as well. Instead of trying different approaches to code to produce an outcome, I'm prompting "this is broken; can you fix it?" or "this almost works! can you do $x instead" and doing stuff in another window while Claude churns. This isn't fun or stimulating at all for me. It's like carrying a dog to a ball instead of making them run for it.
Using it also reminds me that a big part of the experience I've spent years accumulating is (or feels) no longer useful. That everyone can "write" (produce) code now is a good thing, but SO MUCH TIME AND ENERGY was spent on putting writing and understanding code on a pedestal (especially on this site), and seeing that get torn down while gaslighting us into thinking that this never happened has been affecting my psyche for sure.
There's also the dirty feeling I get from abdicating more and more of my skills to a big company that is probably salivating at the idea of developers not being able to write code without Claude Code anymore.
I have a handful of projects I'd like to work on, but I'd rather leave them on the shelf until I can hand-write them than use agents to finish them and, in doing so, create a codebase that I'm less likely to be able to maintain without
agents.
I guess I can turn the question back at you: how aren't you not losing your mind at becoming a glorified spec writer?
I get to spend more time working on the things I enjoy. For example, data modeling and workflow orchestration, and building product to solve customer problems. And really, that comes down to spending a lot of my time just thinking really hard - because once I have a clear plan, it's actually not that hard to build it. Not only that, but I can build something, react to it, rebuild it, react to it, and come up with something I think is much better than I would have been able to build myself. Not to mention much faster.
Can't disagree that Claude ships much much faster. It is also nice to be able to type up a plan of what I'd like to build and have it go after verifying the plan it generates.
I guess it's iterating in plain English instead of iterating in code that I'm having trouble latching onto. It's a "left brain vs right brain" kind of situation for me (not sure how valid that trope is, but it best expresses what I'm feeling).
I'm pretty good at outlining what I want to accomplish and how to get there, but cracking open vim and getting into it is super duper exciting for me, and I'm pretty good at communicating what is and isn't working with the agent when I use it. Claude (or any agent, I suppose) takes that away and doesn't fill that gap.
It's also been harder for me to stay focused with Claude since I'm basically printing code and reading the results. Claude's changesets err towards being pretty large (though I have rules [^1] to try and keep that at bay). When I use it, my mind shifts from "writing code that others might (probably won't) maintain someday)" to "writing code that builds the thing". Given those heavy diffs, I'm less inclined to pore through them than I would be if I were actively writing the code. I fear that this will lead to having less understanding of what my code is doing over time, which is sweeping the bugs and tech debt under the rug (especially since Claude, atm, likes to solve bugs with adding more code). This is probably one of those things that will just take some adjustment.
Then again, maybe part of this has to do with me being a long-time a Vim user and not having used a proper IDE for everyday development in a long time...
Okay, I think I finally came up with a good anecdote for how I'm feeling. I wear Mexican huarache sandals [^0] pretty much exclusively when the weather's warm (which is almost all year round in Houston). They are so unique, and getting them to work perfectly with my feet has been a Linux-like experience (i.e. lots and lots and LOTS of experimentation).
I used to buy them from Luna, Xero and others at, like, $150/pair. However, they are INSANELY EASY to make at home for a fraction of the price. I made four pairs at home. They didn't fit quite right, so I'll need to make them again, but I take MASSIVE satisfaction in wearing what I made.
The journey is as exciting as the destination for me; more so, even!
(Selfishly, I took a lot of pride in being able to show others how to do damn nearly everything they do in VSCode in Vim, but that's another victim of the Era of AI.)
I'm also a lifelong vim user. I guess what I'm getting at is that I find the journey even more engaging now. I think I'm a good programmer. I've worked at great companies. I was competitive in ACM contests and Top Coder. But I find the journey even more engaging now because I can focus on really, really deeply understanding something, and less time on glue. Writing code by hand is still fun, don't get me wrong, but I'm also enjoying the step change in scale.
I don't think IDEs as they exist today are necessarily the right abstraction anymore anyway, to the extent they ever were. At least any more than a C++ IDE that was centered on assembly language as the main thing would be. I want data models, API contracts, and data flows. I don't know what the right answer is, but I think there's something coming.
That has been my feeling too. I have completed soo many personal projects (or improvements) that were collecting dust on my 'mental shelf'.
Without AI I would probably never get to them because realistically, I do not have dozens, or hundreds of personal hours to devote to fun, but unnecessary projects.
Same reason I've passed along offers for management roles and continue to ignore "fiverr.com". It's less having done "The Thing" and more asking for it/a substitute. With an LLM/freelancers, the muscle memory, skills, or whatever I might develop (along with the thing) have been, well, outsourced.
What's more, I can already explore five ideas at once. There is no backlog formed by incapability, lol.
I don't understand the eagerness for "productivity as a service", celebrating quotas, and lot of other conjecture I could get twisted up with... but I'll skip it this time. Rarely pays off :)
Because some of us actually enjoy programming. For some people those side projects aren't about the destination, but the journey of learning how something works by making it with your bare hands.
but that is still possible, so what makes you sad? is it that others can build theirs without bare hands? or you are no longer rewarded for bare hand programming? if the latter was it really the "journey"?
You do know that there's many different personalities that people can have right? A lot of people love writing code and don't care about those things at all.
AstraZeneca is doing some really interesting research in this area - cell therapies that reset the immune system to eliminate the dysfunctional cells driving autoimmune disease, and then allow a healthy immune system to rebuild (for diseases like lupus, rheumatoid arthritis, and multiple sclerosis).
There are already clinics where they basically remove your immune system and give you a new one. If you don’t die in the process, you are likely to be cured of MS.
(Any existing damage will remain.)
Currently this is reserved for the most quickly progressing cases but if we can make this safer and cheaper, it might in future be applied as an early stage cure, so people can go on to live healthy lives.
That being said, Astra Zenecas approach does seem much safer, if it’s proven to be effective!
Yeah AHSCT is no joke. I mentioned in another comment my wife has MS - diagnosed last year in her mid 40s with thankfully no severe impairment. They discussed AHSCT with us but didn’t recommend it unless another disease modifying treatment didn’t work. Thankfully, Tysabri seems to be working well for her.
My mom passed from leukemia years ago. Or rather, from an infection as she was starting HSCT. I’m sure it’s safer than it was 30 years ago, but being without an immune system for a period of time really is still a last resort.
It would be amazing if this type of treatment worked out. MS in particular seems to be a race between technology and your immune system. You hope the next cutting edge treatment is ready by the time the current state of the art stops working for you.
Anokion (now bankrupt) also seemed to have some progress along these lines (link below).
A close family member suffers from MS and is on the more effective but less safe drugs available. They haven’t suffered a relapse since starting them four years ago, but they have been hospitalised twice as a result of side effects.
As we learn more about the relationship between the immune system and various seemingly unrelated diseases the research and understanding has massively increased over the last few years. I’m cautiously optimistic that better treatments aren’t far away. An ancestor was lobotomised for hysteria in the 1960s, before being diagnosed with MS.
not my primary field, but I will say, though, from the folks we see in our clinic, nowadays w/ ocrevus/kesimpta/rituxan, patients are way more stable and dealing with way less side effects than 10-15 years ago.
I like your idea on the concrete examples and deep dives. We started out with the approach of including tools that we hear or are contributed by people explain how they use, so we had some validation. Perhaps if we can find a way to channel some community examples, that would be a good way to grow this weekly. It also makes me wonder if we need a way to prune things from here!
I feel so many of these. LOL @ GitHub endorse-ish, more -ish every day now. Overall though seems like a pretty good hit rate.
Surprised to see datadog as a regret - it is expensive but it's been enormously useful for us. Though we don't run kubernetes, so perhaps my baseline of expensive is wrong.
As I've been looking at this problem from a different angle, I wonder how much execution and planning should be coupled. Once we have specs in GitHub, for example, it feels like we can use whatever tool to execute on them.
Thanks!! Good question...and yea the way I have it set up right now (for my personal workflow) is that Mimir sends the specs to Github as an Issue, and then Claude Code picks it up and starts building. But any agent could do that, i think.
https://github.com/purplefish-ai/factory-factory - creating an IDE for managing a swarm of claude agents, and more importantly, increasingly baking workflows into them (e.g. design -> build -> review -> push -> address comments).
And I'm gonna be honest, I kind of want to use a few of these components for real (the ASCII art is fantastic).
reply