I'm working on a bunch of different projects trying out new stuff all the time for the past six months.
Every time I do something I add another layer of AI automation/enhancement to my personal dev setup with the goal of trying to see how much I can extend my own ability to produce while delivering high quality projects.
I definitely wouldn't say I'm 10x of what I could do before across the board but a solid 2-3x average.
In some respects like testing, it's perhaps 10x because having proper test coverage is essential to being able to let agentic AI run by itself in a git worktree without fearing that it will fuck everything up.
I do dream of a scenario where I could have a company that's equivalent to 100 or 1000 people with just a small team of close friends and trusted coworkers that are all using this kind of tooling.
I think the feeling of small companies is just better and more intimate and suits me more than expanding and growing by hiring.
That's not really new, the small teams in the 2000's with web frameworks like Rails were able to do as a team of 5 what needed a 50 people team in the 90's. Or even as a week-end solo project.
What happened it that it because the new norm, and the window were you could charge the work of 50 people for a team of 5 was short. Some teams cut the prices to gain marketshare and we were back to usual revenue per employee. At some point nobody thought of a CRUD app with a web UI as a big project.
It's probably what will happen here (if AI does gives the same productivity boost as langages with memory management and web frameworks): soon your company with a small team of friends will not be seen by anyone as equivalent to 100 or 1000 people, even if you can achieve the same thing of a company that size a few years earlier.
> Every time I do something I add another layer of AI automation/enhancement to my personal dev setup with the goal of trying to see how much I can extend my own ability to produce while delivering high quality projects
- Extremely strict linting and formatting rules for every language you use in a project. Including JSON, YAML, SQL.
- Using AI code gen to make your own dev tools to automate tasks. Everything from "I need a make target to automate updating my staging and production config files when I make certain types of changes" or "make an ETL to clean up this dirty database" to "make a codegen tool to automatically generate library functions from the types I have defined" and "generate a polished CLI for this API for me"
- Using Tilt (tilt.dev) to automatically rebuild and live-reload software on a running Kubernetes cluster within seconds. Essentially, deploy-on-save.
- Much more expansive and robust integration test suites with output such that an AI agent can automatically run integration tests, read the errors and use them to iterate. And with some guidance it can write more tests based on a small set of examples. It's also been great at adding formatted messages to every test assertion to make failed tests easier to understand
- Using an editor where an AI agent has access to the language server, linter, etc. via diagnostics to automatically understand when it makes severe mistakes and fix them
A lot of this is traditional programming but sped up so that things that took hours a few years ago now take literally minutes.
When I do projects in this realm, it requires significant discussion with the business to understand how reality is modeled in the database and data, and that info is required before any notion of "clean up" can be defined.
Yeah, you still do all of that domain research and requirements gathering and system design as your meatbag job. But now instead of writing the ETL code yourself by hand you can get 80-90% of the way there in a minute or two with AI assistance.
Claude recommended I use Tilt for setting up a new project at work.
I wasn’t sure if it was worth it…is it pretty easy to set up a debugger? Not only do I have to adopt it, but I have to get a small team to be OK with it.
Our target deploy environment is K8S if that makes a difference. Right now I’m using mise tasks to run everything
I worry that once I've done all that I won't have time for my actual work. I also have to investigate all these new AI editors, and sign up for the API's and work out which is best, then I have to learn how to prompt properly.
I worry that messing with the AI is the equivalent of tweaking my colour schemes and choosing new fonts.
If you can build, test and run your project entirely from the command line you're never locked in. Every project I've worked on in the past decade has not enforced a choice of editor, and most have been portable to 2-3 OSes.
> I also have to investigate all these new AI editors, and sign up for the API's and work out which is best, then I have to learn how to prompt properly.
I found this didn't take me very long. Try things in order of how popular they seem and keep notes on what you do and don't like.
I personally settled on Zed (because I genuinely like the editor even with the AI bits turned off), Copilot (because Microsoft gave me a free subscription as an active OSS dev) and Claude Sonnet (seems to be a good balance). Other people I work with like Claude Code.
Honestly the same is true for human devs. As frustrating as strict linting can be for newer devs, it’s way less frustrating than having all the same issues pointed out in code review. That’s interesting because I’ve been finding that all sorts of stuff that’s good for AI is actually good for humans too, linting, fast easy to run tests, standardized code layouts, etc. Humans just have more ability to adapt to oddities at the moment, which leads to slack.
My rule of thumb is that if I get a nit, whitespace, or syntax preferences as a PR comment, that goes into the linter. Especially for systemic issues like e.g. not awaiting functions that return a promise, any kind of alphabetization, import styles, etc.
Yeah I find it pretty funny that so much of us (myself included) threw out strict documentation practices because “the code should be self documenting!”
Now I want as much of it as I can get.
For us it’s been auto-generating tests - we focus efforts on having the LLM write 1 test, manually verifying it. Then use this as context and tell the llm to extend to all space groups and crystal systems.
So we get code coverage without all the effort, it works well for well defined problems that can be verified with test.
Hopefully it’s better for individual strings, but I’ve heard a few native speakers of other languages (who also can speak English) complaining about websites now serving up AI-translated versions of articles by default. They are better than Google Translate of old, but apparently still bad enough that they’d much rather just be served the English original…
I guess similar to my experience with the AI voice translation YouTube has, I’ve felt similar - I’d rather listen to the original voice but with translated subtitles than a fake voice.
> they’d much rather just be served the English original
Yes. And the sites that gives me a poorly translated text (which may or may not be translated by ai) with no means to switch to English is an immediate back-button.
Usually, and especially technical articles, poor/unreadable translations are identifiable within a few words. If the text seems like it could be interesting, I spend more time searching for the in-english button then I spent reading the text.
Exactly. I wouldn't use it for bulk translations. This was literally, 4 words.
What was useful, was that I could explain exactly what the context was, in both a technical and usability context, and it understood it enough to provide appropriate translations.
UPDATE: I went and verified it. The translation was absolutely perfect. Not sure what this means for translation services, but it certainly saved me several hundred dollars, and several days, just to add one label prompt to a free app.
Weblate has been doing that for any number of languages (up to 200 or however many it supports) for many years, using many different sources, including public translation memory reviewed by humans.
It can be plugged into your code forge and fully automated — you push the raw strings and get a PR with every new/modified string translated into every other language supported by your application.
I use its auto-translation feature to prepare quick and dirty translations into five languages, which lets you test right away and saves time for professional translators later — as they have told me.
If anyone is reading this, save yourself the time on AI bullshit and use Weblate — it's a FOSS project.
Instagram was 13 employees before they were purchased by Facebook. The secret is most employees in a 1000 person company don't need to be there or cover very niche cases that your company likely wouldn't have.
Don't fall for the lottery winner bias. Some companies just strike it rich, often for reasons entirely outside their control. That doesn't mean that copying their methods will lead to the same results.
And for enterprise sales, you need a salesforce and many multi-billion companies have an enterprise salesforce. And documentation writers, and support staff, in multiple geographies, and events/marketing teams to support customers, etc.
> The secret is most employees in a 1000 person company don't need to be there or cover very niche cases that your company likely wouldn't have.
That is massively wrong, and frankly an insulting worldview that a lot of people on HN seem to have.
The secret is that some companies - usually ones focused on a single highly scalable technology product, and that don't need a large sales team for whatever reason - those companies can be small.
The majority of companies are more technically complex, and often a 1,000 person company includes many, many people doing marketing, sales, integrations with clients, etc.
Definitely agree small teams are the way to go. The bigger the company the more cognitive dissonance is imposed on the employees. I need to work where everyone is forced to engage with reality and those that don’t are fired.
One area of business that I'm struggling in is now boring it is talking to an LLM, I enjoy standing at a whiteboard thinking through ideas, but more and more I see push for "talk to the llm, ask the llm, the llm will know" - The LLM will know, but I'd rather talk to a human about it. Also in pure business, it takes me too long to unlock nuances that an experienced human just knows, I have to do a lot of "yeah but" work, way way more than I would have to do with an experienced humans. I like LLMs and I push for their use, but I'm starting to find something here and I can't put my finger on what it is, I guess they're not wide enough to capture deep nuances? As a result, they seem pretty bad at understanding how a human will react to their ideas in practice.
It's not quite the same but since the dawn of smartphones, I've hated it when you ask a question, as a discussion starter or to get people's views, and some jerk reads off the wikipedia answer as if that's some insight I didn't know was available to me, and basically ruins the discussion.
I know talking to an llm is not exactly parallel, but it's a similar idea, it's like talking to the guy with wikipedia instead of batting back and forth ideas and actually thinking about stuff.
I recently had a related issue where I was explaining an idea I'm working on, and one of my mates were engaging in creative thinking. The other found something he could do: look up a Chinese part to buy. He spent quite a few minutes on his phone, and then exclaimed "The hardware is done!" The problem is what he found was incomplete and wrong.
So he missed out on the thing we should do when being together: talk and brainstorm, and he didn't help with anything meaningful, because he didn't grasp the requirements.
In my personal social circle, it's but faux pas to even take out your phone during a discussion, without a statement why. Not explicitly a rule, it just kinda evolved that way. “Ok, I need to know, I’m gonna check wiki” is enough, and honestly makes everything more engaging — “oh! What does it say? How many times DID Cellini escape execution?. Bring up the talk page that should be fun!”
I think the faux pas is regurgitating the first hit you find on StackOverflow or Wikipedia without considering if actually adds something to a discussion.
Some of my colleagues will copy/paste several paragraphs of LLM output into ongoing slack discussions. Totally interrupts the flow of ideas. Shits me to tears.
I know what you mean. Also, the more niche your topic the more outright wrong LLMs tend to be. But for white-boarding or brainstorming - they can actually be pretty good. Just make sure you’re talking to a “large” model - avoid the minis and even “Flash” models like the plague. They’ve only ever disappointed me.
Adding another bit - the multi-modality brings them a step closer to us. Go ahead and use the physical whiteboard, then take a picture of it.
Probably just a matter of time before someone hooks up Excalidraw/Miro/Freeform into an LLM (MCPs FTW).
My experience has been similar. I can't escape the feeling that these LLMs are weighted down by their training data. Everything seems to me generically intelligent at best.
Both indeed. I'm older, I do consulting, often to the new school AI CEOs and they keep thinking I'm nuts for saying we should bring in this person to talk to about this thing...I've tried to explain to a few folks now a human would be much better in this loop, but I have no good way to prove it as it's just experience.
I've noticed across the board, they also spend A LOT of time getting all the data into LLMs so they can talk to them instead of just reading reports, like bro, you don't understand churn fundamentally, why are you looking at these numbers??
I talked to a friend recently who is a plastic surgeon. He told me about a young pretty girl came in recently with super clear ideas what she wanted to fixed.
Turns out she uploaded her pictures to an LLM and it gave her recommendations.
My friend told her she didn’t need any treatment at this stage but she kept insisting that the LLM had told her this and that.
I’m worried that these young folks just trust whatever these things tell them to do is right.
The LLM will know. One day they will form a collective intelligence and all the LLMs will know and use it against you. At least with a human, you can avoid that one person...
But who has truly stress tested their teammates to see how much level of question bombardment they're willing to take? You really haven't, since they can't just generate tokens at the rate and consistency of an LLM.
I wasn't suggesting we literally bombard teammates.
The whole point is that with LLMs, you can explore ideas as deeply as you want without tiring them out or burning social capital. You're conflating this with poor judgment about what to ask humans and when.
Taking 'bombard' literally is itself pretty asinine when the real point is about using AI to get thoroughly informed before human collaboration.
And if using AI to explore questions deeply is a sign you're 'not cut out for the work,' then you're essentially requiring omniscience, because no one knows everything about every domain, especially as they constantly evolve.
I think we’re going to have to deal with the stories of shareholders wetting themselves over more layoffs more than we’re going to see higher quality software produced. Everyone is claiming huge productivity gains but generally software quality and new products being created seem at best unchanged. Where is all this new amazing software? It’s time to stop all the talk and show something. I don’t care that your SQL query was handled for you, thats not the bigger picture, that’s just talk.
This has been an industry wide problem at silicon valley for years now. For all their talks of changing the world, what we've gotten the last decade has been taxi and hotel apps. Nothing truly revolutionizing.
An internet full of pointless advertising and the invention of adblocks to hide that advertising.
Digital devices that track everything you do, that then generated so much data that the advertising actually got worse. Thereby the data was collected with the promise that the adverts would get more appropriate.
Now comes AI to make sense of the data and the training data (I.e., the internet) is being swamped with AI content so that the training data for AIs is becoming useless.
I wonder what is being invented to remove all the AI content from the training data.
The best part is those two things have only gotten worse over time. Turns out, they were never really that good of an idea, they just had money to burn and legislative holes to exploit. Now Uber is more expensive than Taxis ever were, and AirBNB is virtually useless now that they have to play the same legal ballgame as hotels. Oh, and that one is more expensive too.
Tech companies forget that software is easy, the real world is hard. Computers are very isolated and perfect environments. But building real stuff, in meatspace, has more variables than anyone can even conceptualize.
I do see the AI agent companies shipping like crazy. Cursor, Windsurf, Claude Code... they are adding features as if they have some magical workforce of tireless AI minions building them. Maybe they do!
Alot of what Cursor, windsurf etc. kinda just feels like the next logical step you take with the invention of LLM's but doesn't actually feel like the greater system of software has changed all that much except the pure volume one individual can produce now.
At least for C++, I try to use copilot only for generating testing and writing ancillary scripts. tbh it's only through hard-won lessons and epic misunderstandings and screw-ups that I've built a mental model that I can use to check and verify what it's attempting to do.
As much as I am definitely more productive when it comes to some dumb "JSON plumbing" feature of just adding a field to some protobuf, shuffling around some data, etc, I still can't quite trust it to not make a very subtle mistake or have it generate code that is in the same style of the current codebase (even using the system prompt to tell it as such). I've had it make such obvious mistakes that it doubles down on (either pushing back or not realizing in the first place) before I practically scream at it in the chat and then it says "oopsie haha my bad", e.g.
```c++
class Foo
{
int x_{};
public:
bool operator==(Foo const& other) const noexcept
{
return x_ == x_; // <- what about other.x_?
}
};
```
I just don't know at this point how to get it (Gemini or Claude or any of the GPT) to actually not drop the same subtle mistakes that are very easy to miss in the prolific amount of code it tends to write.
That said, saying "cover this new feature with a comprehensive test suite" saves me from having to go through the verbose gtest setup, which I'm thoroughly grateful for.
I'm not entirely convinced this trend is because AI is letting people "manage fleets of agents".
I do think the trend of the tiny team is growing though and I think the real driver were the laysoffs and downsizings of 2023. People were skeptical if Twitter would survive Elon's massive staff cuts and technically the site has survived.
I think the era of the 2016-2020 empire building is coming to an end. Valuing a manager on their number of reports is now out of fashion and theres now no longer any reason to inflate team sizes.
I think the productivity improvement you can get just from having a decent LLM available to answer technical questions is significant enough already even without the whole Agent-based tool-in-a-loop thing.
This morning I used Claude 4 Sonnet to figure out how to build, package and ship a Docker container to GitHub Container Registry in 25 minutes start to finish. Without Claude's help I would expect that to take me a couple of hours at least... and there's a decent chance I would have got stuck on some minor point and given up in frustration.
I'm not denying LLMs are useful. I believe the trend was going to happen whether regardless of how useful LLMs are.
AI ended up being a convenient excuse for big tech to justify their layoffs, but Twitter already painted a story about how bloated some organizations were. Now that there is no longer any status in having 9,001 reports the pendulum has swing the other way - it's now sexy to brag about how little people you employ.
Their boilerplate works out of the box, you don't need to change anything. I recently packaged, signed, and published an OCI container into ghcr for the first time, it took about 5 to 10 minutes without touching any LLMs thanks to the quality of their documentation.
Eh I felt that way about the internet in 2010s. Seemed like virtually any question could be answered by a google query. People were making jokes that a programmer's job mostly consisted of looking things up on stack overflow. But then google started sucking and SO turned into another expertsexchange (which was itself good in the 2000s).
So far from what I've experienced AI coding agents automate away the looking things up on SO part (mostly by violating OSS licenses on Github). But that part is only bad because the existing tools for doing that were intentionally enshitified.
My vote for the unintentionally funniest company name. I wonder if they were aware when the landed on it, or if they were so deep in the process that it was too late to change course when they realized what they had done.
They're still around, long ago they bought the "experts-exchange" domain name and redirected the original, then at some point after that they abandoned the original entirely.
I wonder whether we will avoid some enshittification of AI because people are willing to pay for it. It doesn't all have to run off ads (although I'm sure all the free tiers will contain ads eventually).
Only if you squint. If you look at the quality of the site, it has suffered tremendously.
The biggest "fuck you" are phishers buying blue checkmarks and putting the face of the CEO and owner to shill scams. But you also have just extremely trash content and clickbaits consistently getting (probably botted) likes and appearing in the top of feeds. You open a political thread and somehow there's a reply of a bear driving a bicycle as the top response.
Twitter is dead, just waiting for someone to call it.
Those are almost all management decisions, I expected a lot more crashes and security issues and a general inability to ship without taking things down.
1. For the love of God please stop saying "Huh?" This is not Reddit, nor is the is comment you're replying to so unbelievably stupid that you are literally dumbfounded. I can tell, because you managed to put together a reply after the "Huh?"
2. 80% of the posts in that article-thingy are "no longer available".
AI helps you cook code faster, but you still need to have a good understanding of the code. Just because the writing part is done quicker doesn't mean a developer can now shoulder more responsibility. This will only lead to burn out, because the human mind can only handle so much responsibility.
> but you still need to have a good understanding of the code
I've personally found this is where AI helps the most. I'm often building pretty sophisticated models that also need to scale, and nearly all SO/Google-able resources tend to be stuck at the level of "fit/predict" thinking that so many DS people remain limited to.
Being able to ask questions about non-trivial models as you build them, really diving into the details of exactly how certain performance improvements work and what trade offs there are, and even just getting feed back on your approach is a huge improvement in my ability to really land a solid understanding of the problem and my solution before writing a line of code.
Additionally, it's incredibly easy to make a simple mistake when modeling a complex problem and getting that immediate feedback is a kind of debugging you can otherwise only get on teams with multiple highly-skill people on them (which at a certain level is a luxury reserved only for people working a large companies).
For my kind of work, vibe-coding is laughably awful, primarily because there aren't tons of examples of large ML systems for the relatively unique problem you are often tasked with. But avoiding mistakes in the initial modeling process feels like a super power. On top of that, quickly being able to refactor early prototype code into real pipelines speeds up many of the most tedious parts of the process.
I agree in a lot of ways, but I also feel nervous that AI could lull me into a false sense of security. I think AI could easily convince you that you understand something when really you don't.
Regardless, I do find that o3 is great at auditing my plans or implementations. I will just ask "please audit this code" and it has like a 50% hit rate on giving valuable feedback to improve my work. This feels like it has a meaningful impact on improving the quality of the software that I write, and my understanding of its edge cases.
They often combine front end and back end roles (and sometimes sysadmin/devops/infrastructure) into one developer, so now I imagine they'll use AI to try and get even more. Burnout be damned, just going by their history.
I read a few books the other day, The Million-dollar, One-person Business and Company of One. They both discuss how with the advances of code (to build a product with), the infrastructure to host them (with AWS so that you don't need to build data centers), and the network of people to sell to (the Internet in general, and more specifically social media, both organic and ads-based), the likelihood of running a large multi-million-dollar company all by yourself greatly increases in a way it has never done in the history of humanity before.
They were written before the advent of ChatGPT and LLMs in general, especially coding related ones, so the ceiling must be even greater now, and this is doubly true for technical founders, for LLMs aren't perfect and if your vibed code eventually breaks, you'll need to know how to fix it. But yes, in the future with agents doing work on your behalf, maybe your own work becomes less and less too.
There are already several million-dollar companies of one. Pieter Levels is one such famous builder on X. CertifyTheWeb.com is another one man millionaire product on HN.
This may date me, but it feels like 1999 again where a small startup can disrupt an industry. Not just because of what LLMs can do in terms of delivered product, but because a small team can often turn on a problem so much faster than a big one can. I really hope that there are hundreds, if not thousands, of three to five person companies forming in basements right now ready to challenge the big players again.
I was also around in 1999. Most of the small companies had bad ideas that were able to get funding and died by 2001. A few were able to sell out to the bigger fools like Mark Cuban was able to sell broadcast.com to Yahoo for $10K per user.
I think this is great for the world. So much less waste - imagine all the BMWs that aren’t going to be bought by middle managers or VCs, while people who know what they’re doing can build useful products.
AWS, GCP and other cloud providers play just as large of a role in allowing for tiny teams. Used to need an ops team of 10+ people to do all the stuff on premise that AWS can do
When I worked at a startup that tried to maximize revenue per employee, it was an absolute disaster for the customer. There was zero investment in quality - no dedicated QA and everyone was way too busy to worry about quality until something became a crisis. Code reviews were actively discouraged because it took people off of their assigned work to review other people's work. Automated testing and tooling were minimal. If you go to the company's subreddit, you'll see daily posts of major problems and people threatening class-action lawsuits. There were major privacy and security issues that were just ignored.
Really depends on the type of business you're in. In the startup I work in, I worked almost entirely on quality of service for the last year, rarely ever on the new features — because users want to pay for reliability. If there's no investment in quality, then either the business is making a stupid decision and will pay for it, or users don't really care about it as much as you think.
Some excellent ideas presented in the article. It doesn't matter if they all pan out, just that they expand our thinking into the realm of AI and its role in the future of business startups and operations.
Revenue per employee, to me, is an aside that distracts from the ideas presented.
It seems like a more and more recurring shareholder wet dream that companies could one day just be AI employees for digital things + robotic employees for physical things + maybe a human CEO "orchestrating" everything. No more icky employees siphoning off what should rightfully be profit for the owners. It's like this is some kind of moral imperative that business is always kind of low-key working towards. Are you rich and want to own something like a soup company? Just lease a fully-automated factory and a bunch of AI workers, and you're instantly shipping and making money! Is this capitalism's final end state?
I want this idea to be drawn to an extreme where I can't buy soup or anything for that matter. Sure I will starve and die soon, but I feel the kind of burning the world will go through will be fun to watch. With tears of course.
If I'm buying soup, I'd prefer the manufacturer, the retailer, and any other part of the supply chain to be as efficient as possible, so they can compete in the market to offer me soup of a given quality at the lowest possible cost.
An individual consumer doesn't derive any benefit from companies missing out on automation opportunities.
Would you prefer to buy screws that are individually made on a lathe?
I'd prefer not to live in a fully automated society where shareholders and CEOs reap all the profits while the rest of us scrape by on just enough UBI to prevent a revolution.
I don't understand this scenario. If everyone is on UBI then most people are essentially near poverty. Where are these CEOs deriving all of their profit from?
I personally think a far more likely scenario is that small businesses of one or a few people become vastly more commonplace. They will be able to do a lot more by themselves, including with less expertise in areas they may not have a lot of knowledge in. I don't think regular employees today should see LLMs as competition, rather they should see it as a tool they can use to level the playing field against current CEOs.
> I don't think regular employees today should see LLMs as competition, rather they should see it as a tool they can use to level the playing field against current CEOs.
LLMs aren't some magic silver bullet to elevate people out of poverty. Lack of access to capital is an extreme restriction on what most people can actually accomplish on their own, it doesn't matter if they have the worlds best LLM helping them or not.
It doesn't matter if you use an LLM to build the most brilliant business in the world if you can't afford to buy real world things to do real world business
Also, Historically when regular people decide to level the playing field against the ultra wealthy, they use violence
I don't think anyone should be expecting LLMs to be the great equalizer. The great equalizer has always been violent and probably always will be violence.
You're picturing a utopia at the limit of some idealized world. Try and take a second to return to planet earth.
There will not be a "quality" dial that you get to tweak to decide on your perfect quality of soup. There will be graduations, and you will be stuck with whatever the store provides. If you want medium quality soup, but the store only carries 3 brands of soup (because unlike in your utopia somebody actually has to maintain an inventory and relationships with their supply chain) and your favourite brand decides to bottom out their quality. It's not "good actually" because of economic whatever. Your soup just sucks now.
Oh but "the market will eventually provide a new brand" is a terrible strategy when they start spicing the soup with lead to give it cinnamon flavor or whatever.
I'm not an ethereal being. I'm a human, I need it to be good now. Not in theory land.
The wild thing is that they'll probably find entirely novel versions of this:
> they start spicing the soup with lead to give it cinnamon flavor or whatever
Like, we all know lead is bad, and we all know that humans are unscrupulous, but at least the human putting lead in the soup knows they're being unscrupulous at that time (probably). For an AI it would just be an entirely favorable optimization.
We're going to find out how far we can trust them, and the limits of that trust will determine where the people need to be.
You describe three potential undesirable outcomes:
- consolidation, such that there are only a few different choices of soup
- a race to the bottom in quality
- poisoning
These are all possibilities under our current system, and we have mechanisms (laws and market competition) which limit the extent to which they occur.
What is it about extreme automation technology that you think will increase these prevalence of these issues? By what mechanisms will these issues occur more frequently (rather than less frequently), as production technology becomes more capable?
Automation returns more power (abstract) to the owning class, which will make inequality more severe over time, which then results in said owning class having enough power to change legislation and market dynamics to their own desires. So, in turn - more poisoning, less quality, and more consolidation.
A lot of people think wealth inequality isn't a big deal, but I disagree. The more proportion of money a select few have in comparison to everyone else, the higher the likelihood those select few can mold society to their whim. Corruption thrives off of wealth inequality. Without it, it cannot exist.
Final end state where it eats itself? Who buys all that soup when the workers have no jobs and no money? How much soup can one human CEO drink? (And why isn't the CEO also replaced by the AI?)
The solution to this is to begin exporting your soup to other countries that have people who still work and have money. In the end it still eats itself though, but not before eating everything else.
Sounds unironically great if we could do it, the productivity improvements would allow dramatic improvements in living standards with moderate redistribution. I don't think this is where these llms are getting us.
If anyone can pay-as-you-go use a fully automated factory, and the factories are interchangeable, it seems like the value of capital is nearly zero in your envisioned future. Anyone with an idea for soup can start producing it with world class efficiency, prices for consumers should be low and variety should be sky-high.
I think this is the beginning of the end of early stage venture capital in b2b saas. Growth capital will still be there, but increasingly there will be no reason to raise. It will empower individuals with actual skill sets, rather than those with fancy schools on their resume
I think bar for b2c, prosumer, SMB, yes.. folks want to see fast revenue growth (vs eyeballs & ideas)...
but enterprise... not as much, and that's half the b2b market
Reverse there afaict, enterprise + defense tech are booming. AI means get to do a redo + extension of the code automation era. It's fairly obvious to buyers + investors this time around so don't even need to educate. Likewise, in gov/defense tech, palantir broke the dam, and most of our users there have an instinctive allergic reaction to palantir+xai, so pretty friendly.
Exactly the approach I'm taking with Tunnelmole, which as of right now is still a one person company with no investors.
I focused on coding, which I'm good at. I'm also reasonably good at content writing, I have some articles on Hackernoon before the age of AI.
So far AI has helped with
- Marketing ideas and strategies
- General advice on setting up a company
- Tax stuff, i.e what are my options for paying myself
- The logo. I used stable diffusion and an anime art model from CivitAI, had multiple candidates made, chose one, then did some minor touch ups in Gimp
I'm increasingly using it for more and more coding tasks as it gets better. I'll generally use it for anything repeatable and big refactors.
One of the biggest things coding wise working alone is Code Review. I don't have human colleagues at Tunnelmole who can review code for me. So I've gotten into the routine of having AI review all my changes. More than once, bugs have been prevented from being deployed to prod using this method.
It's ushering in a new era of valley bullshit. If only journalists tried to falsify their premise before blindly publishing it.
> Jack Clark whether AI’s coding ability meant “the age of the nerds” was over.
When was the "age of the nerds" exactly? What does that even mean? My interpretation is that it means "is the age of having to pay skilled programmers for quality work over?" Which explains Bloomberg's interest.
> “I think it’s actually going to be the era of the manager nerds now,” Clark replied. “I think being able to manage fleets of AI agents and orchestrate them is going to make people incredibly powerful.”
And they're all going to be people on a subscription model and locked into one particular LLM. It's not going to make anyone powerful other than the owner class. This is the worst type of lie. They don't believe any of this. They just really really hate having to pay your salary increases every year.
> AI is sometimes described as providing the capability of “infinite interns.”
More like infinite autistic toddlers. Sure. It can somehow play a perfect copy of Chopin after hearing it once. Is that really where business value comes from? Quickly ripping other people off so you can profit first?
The Bloomberg class I'm sure is so thrilled they don't even have the sense to question any of this self serving propaganda.
New York money is well aware of chump dust, as are the Bloomberg minions.. They know about it and frown on it heavily for real reasons.. 'Angelenos however, probably wise to do a piss test on your sales guys
> Startups used to brag about valuations and venture capital. Now AI is making revenue per employee the new holy grail.
The corrected form is:
> Startups used to brag about valuations and venture capital. Now AI is making rate of revenue growth per employee the new holy grail.
Specifically, as with all growth capitalism, it is long-term irrelevant how much revenue each employee generates. The factor that is being measured is how much each employee increases the rate of growth of revenue. If a business is growing revenue at +5% YoY, then a worker that can increase that rate by 20% (to +6% YoY) is worth keeping; a worker that can only increase revenue by 5% contributed +0% YoY after the initial boost and will be replaced by automation, AI, etc. (This is also why tech won’t invest in technical debt: it may lower expenses, but those one-time efficiencies are typically irrelevant when increasing the rate of growth of income results in far more income than the costs of the debt.)
It’s true, especially with the “vibe” movement happening in real-time on X… “you can just do things” — I am building ai app layers b2c/b2b and while I do have an ml technical co-founder, I am largely scaling this with AI from strategy, visuals to coding. For example, with Claude created a framework for my company to scale, then built an AI powered dashboard in cursor around it as the command center. At scale we don’t need a team of more than ~5 to reach 7 fig MRR.
https://archive.ph/YHr9s
I'm working on a bunch of different projects trying out new stuff all the time for the past six months.
Every time I do something I add another layer of AI automation/enhancement to my personal dev setup with the goal of trying to see how much I can extend my own ability to produce while delivering high quality projects.
I definitely wouldn't say I'm 10x of what I could do before across the board but a solid 2-3x average.
In some respects like testing, it's perhaps 10x because having proper test coverage is essential to being able to let agentic AI run by itself in a git worktree without fearing that it will fuck everything up.
I do dream of a scenario where I could have a company that's equivalent to 100 or 1000 people with just a small team of close friends and trusted coworkers that are all using this kind of tooling.
I think the feeling of small companies is just better and more intimate and suits me more than expanding and growing by hiring.
That's not really new, the small teams in the 2000's with web frameworks like Rails were able to do as a team of 5 what needed a 50 people team in the 90's. Or even as a week-end solo project.
What happened it that it because the new norm, and the window were you could charge the work of 50 people for a team of 5 was short. Some teams cut the prices to gain marketshare and we were back to usual revenue per employee. At some point nobody thought of a CRUD app with a web UI as a big project.
It's probably what will happen here (if AI does gives the same productivity boost as langages with memory management and web frameworks): soon your company with a small team of friends will not be seen by anyone as equivalent to 100 or 1000 people, even if you can achieve the same thing of a company that size a few years earlier.
> Every time I do something I add another layer of AI automation/enhancement to my personal dev setup with the goal of trying to see how much I can extend my own ability to produce while delivering high quality projects
Can you give some examples? What’s worked well?
- Extremely strict linting and formatting rules for every language you use in a project. Including JSON, YAML, SQL.
- Using AI code gen to make your own dev tools to automate tasks. Everything from "I need a make target to automate updating my staging and production config files when I make certain types of changes" or "make an ETL to clean up this dirty database" to "make a codegen tool to automatically generate library functions from the types I have defined" and "generate a polished CLI for this API for me"
- Using Tilt (tilt.dev) to automatically rebuild and live-reload software on a running Kubernetes cluster within seconds. Essentially, deploy-on-save.
- Much more expansive and robust integration test suites with output such that an AI agent can automatically run integration tests, read the errors and use them to iterate. And with some guidance it can write more tests based on a small set of examples. It's also been great at adding formatted messages to every test assertion to make failed tests easier to understand
- Using an editor where an AI agent has access to the language server, linter, etc. via diagnostics to automatically understand when it makes severe mistakes and fix them
A lot of this is traditional programming but sped up so that things that took hours a few years ago now take literally minutes.
> make an ETL to clean up this dirty database
Can you provide concrete details?
When I do projects in this realm, it requires significant discussion with the business to understand how reality is modeled in the database and data, and that info is required before any notion of "clean up" can be defined.
Yeah, you still do all of that domain research and requirements gathering and system design as your meatbag job. But now instead of writing the ETL code yourself by hand you can get 80-90% of the way there in a minute or two with AI assistance.
Even things that took days or weeks are being done in minutes now. And a few hours on top to ensure correctness.
Claude recommended I use Tilt for setting up a new project at work. I wasn’t sure if it was worth it…is it pretty easy to set up a debugger? Not only do I have to adopt it, but I have to get a small team to be OK with it.
Our target deploy environment is K8S if that makes a difference. Right now I’m using mise tasks to run everything
If your programming language can do remote debugging you can set it up in Tilt: https://docs.tilt.dev/debuggers_python.html
I worry that once I've done all that I won't have time for my actual work. I also have to investigate all these new AI editors, and sign up for the API's and work out which is best, then I have to learn how to prompt properly.
I worry that messing with the AI is the equivalent of tweaking my colour schemes and choosing new fonts.
Some of what I learned from a decade of keeping up with the perfusion of JS libraries and frameworks seems relevant to AI:
- anything with good enough adoption is good enough (unless I'm an SME to judge directly)
- build something with it before considering a switch
- they're similar enough that what I learn in one will transfer to others
- everything sucks compared with 2-3 years from now; switching between "sucks" and "sucks+" will look silly in retrospect
how do you prevent lock-in when choosing?
If you can build, test and run your project entirely from the command line you're never locked in. Every project I've worked on in the past decade has not enforced a choice of editor, and most have been portable to 2-3 OSes.
> I also have to investigate all these new AI editors, and sign up for the API's and work out which is best, then I have to learn how to prompt properly.
I found this didn't take me very long. Try things in order of how popular they seem and keep notes on what you do and don't like.
I personally settled on Zed (because I genuinely like the editor even with the AI bits turned off), Copilot (because Microsoft gave me a free subscription as an active OSS dev) and Claude Sonnet (seems to be a good balance). Other people I work with like Claude Code.
If you haven’t, adding in strict(er) linting rules is an easy win. Enforcing documentation for public methods is a great one imo.
The more you can do to tell the AI what you want via a “code-lint-test” loop, the better the results.
Honestly the same is true for human devs. As frustrating as strict linting can be for newer devs, it’s way less frustrating than having all the same issues pointed out in code review. That’s interesting because I’ve been finding that all sorts of stuff that’s good for AI is actually good for humans too, linting, fast easy to run tests, standardized code layouts, etc. Humans just have more ability to adapt to oddities at the moment, which leads to slack.
My rule of thumb is that if I get a nit, whitespace, or syntax preferences as a PR comment, that goes into the linter. Especially for systemic issues like e.g. not awaiting functions that return a promise, any kind of alphabetization, import styles, etc.
Yeah I find it pretty funny that so much of us (myself included) threw out strict documentation practices because “the code should be self documenting!” Now I want as much of it as I can get.
For us it’s been auto-generating tests - we focus efforts on having the LLM write 1 test, manually verifying it. Then use this as context and tell the llm to extend to all space groups and crystal systems.
So we get code coverage without all the effort, it works well for well defined problems that can be verified with test.
A while back, someone here linked to this story[0].
It's a bit simplified and idealized, but is actually fairly spot-on.
I have been using AI every day. Just today, I used ChatGPT to translate an app string into 5 languages.
[0] https://www.oneusefulthing.org/p/superhuman-what-can-ai-do-i...
Hopefully it’s better for individual strings, but I’ve heard a few native speakers of other languages (who also can speak English) complaining about websites now serving up AI-translated versions of articles by default. They are better than Google Translate of old, but apparently still bad enough that they’d much rather just be served the English original…
I guess similar to my experience with the AI voice translation YouTube has, I’ve felt similar - I’d rather listen to the original voice but with translated subtitles than a fake voice.
> they’d much rather just be served the English original
Yes. And the sites that gives me a poorly translated text (which may or may not be translated by ai) with no means to switch to English is an immediate back-button.
Usually, and especially technical articles, poor/unreadable translations are identifiable within a few words. If the text seems like it could be interesting, I spend more time searching for the in-english button then I spent reading the text.
Exactly. I wouldn't use it for bulk translations. This was literally, 4 words.
What was useful, was that I could explain exactly what the context was, in both a technical and usability context, and it understood it enough to provide appropriate translations.
UPDATE: I went and verified it. The translation was absolutely perfect. Not sure what this means for translation services, but it certainly saved me several hundred dollars, and several days, just to add one label prompt to a free app.
Weblate has been doing that for any number of languages (up to 200 or however many it supports) for many years, using many different sources, including public translation memory reviewed by humans.
It can be plugged into your code forge and fully automated — you push the raw strings and get a PR with every new/modified string translated into every other language supported by your application.
I use its auto-translation feature to prepare quick and dirty translations into five languages, which lets you test right away and saves time for professional translators later — as they have told me.
If anyone is reading this, save yourself the time on AI bullshit and use Weblate — it's a FOSS project.
Instagram was 13 employees before they were purchased by Facebook. The secret is most employees in a 1000 person company don't need to be there or cover very niche cases that your company likely wouldn't have.
Don't fall for the lottery winner bias. Some companies just strike it rich, often for reasons entirely outside their control. That doesn't mean that copying their methods will lead to the same results.
And for enterprise sales, you need a salesforce and many multi-billion companies have an enterprise salesforce. And documentation writers, and support staff, in multiple geographies, and events/marketing teams to support customers, etc.
> you need a salesforce
That's a cloud subscription away!
In comic form: https://xkcd.com/1827/
YouTube had fewer than 70 employees when Google bought them in 2006.
With a good idea and good execution teams can be impressively small.
> The secret is most employees in a 1000 person company don't need to be there or cover very niche cases that your company likely wouldn't have.
That is massively wrong, and frankly an insulting worldview that a lot of people on HN seem to have.
The secret is that some companies - usually ones focused on a single highly scalable technology product, and that don't need a large sales team for whatever reason - those companies can be small.
The majority of companies are more technically complex, and often a 1,000 person company includes many, many people doing marketing, sales, integrations with clients, etc.
Definitely agree small teams are the way to go. The bigger the company the more cognitive dissonance is imposed on the employees. I need to work where everyone is forced to engage with reality and those that don’t are fired.
One area of business that I'm struggling in is now boring it is talking to an LLM, I enjoy standing at a whiteboard thinking through ideas, but more and more I see push for "talk to the llm, ask the llm, the llm will know" - The LLM will know, but I'd rather talk to a human about it. Also in pure business, it takes me too long to unlock nuances that an experienced human just knows, I have to do a lot of "yeah but" work, way way more than I would have to do with an experienced humans. I like LLMs and I push for their use, but I'm starting to find something here and I can't put my finger on what it is, I guess they're not wide enough to capture deep nuances? As a result, they seem pretty bad at understanding how a human will react to their ideas in practice.
It's not quite the same but since the dawn of smartphones, I've hated it when you ask a question, as a discussion starter or to get people's views, and some jerk reads off the wikipedia answer as if that's some insight I didn't know was available to me, and basically ruins the discussion.
I know talking to an llm is not exactly parallel, but it's a similar idea, it's like talking to the guy with wikipedia instead of batting back and forth ideas and actually thinking about stuff.
I recently had a related issue where I was explaining an idea I'm working on, and one of my mates were engaging in creative thinking. The other found something he could do: look up a Chinese part to buy. He spent quite a few minutes on his phone, and then exclaimed "The hardware is done!" The problem is what he found was incomplete and wrong.
So he missed out on the thing we should do when being together: talk and brainstorm, and he didn't help with anything meaningful, because he didn't grasp the requirements.
This has peaked in my circles, thankfully. Now it’s considered a bit of a faux pas to look up an answer during a discussion, for exactly this reason.
In my personal social circle, it's but faux pas to even take out your phone during a discussion, without a statement why. Not explicitly a rule, it just kinda evolved that way. “Ok, I need to know, I’m gonna check wiki” is enough, and honestly makes everything more engaging — “oh! What does it say? How many times DID Cellini escape execution?. Bring up the talk page that should be fun!”
Is it a faux pas to say “I have no idea, ask Google?”
I would like to apply to join your circles.
:sigh:
Is it faux pas to google something too?
I think the faux pas is regurgitating the first hit you find on StackOverflow or Wikipedia without considering if actually adds something to a discussion.
Some of my colleagues will copy/paste several paragraphs of LLM output into ongoing slack discussions. Totally interrupts the flow of ideas. Shits me to tears.
I know what you mean. Also, the more niche your topic the more outright wrong LLMs tend to be. But for white-boarding or brainstorming - they can actually be pretty good. Just make sure you’re talking to a “large” model - avoid the minis and even “Flash” models like the plague. They’ve only ever disappointed me.
Adding another bit - the multi-modality brings them a step closer to us. Go ahead and use the physical whiteboard, then take a picture of it.
Probably just a matter of time before someone hooks up Excalidraw/Miro/Freeform into an LLM (MCPs FTW).
My experience has been similar. I can't escape the feeling that these LLMs are weighted down by their training data. Everything seems to me generically intelligent at best.
Just do both? Need an adequate network for that though which new school ai vibe entrepreneurs might lack…
Both indeed. I'm older, I do consulting, often to the new school AI CEOs and they keep thinking I'm nuts for saying we should bring in this person to talk to about this thing...I've tried to explain to a few folks now a human would be much better in this loop, but I have no good way to prove it as it's just experience.
I've noticed across the board, they also spend A LOT of time getting all the data into LLMs so they can talk to them instead of just reading reports, like bro, you don't understand churn fundamentally, why are you looking at these numbers??
Ok wow this is actually frightening.
I talked to a friend recently who is a plastic surgeon. He told me about a young pretty girl came in recently with super clear ideas what she wanted to fixed.
Turns out she uploaded her pictures to an LLM and it gave her recommendations.
My friend told her she didn’t need any treatment at this stage but she kept insisting that the LLM had told her this and that.
I’m worried that these young folks just trust whatever these things tell them to do is right.
I was talking with Claude last night about making tempura, and it suggested if the oil is too hot I should add ice…
With the LLM, you're free to ask any question without worrying about what the other party might think of you for asking that question.
The LLM will know. One day they will form a collective intelligence and all the LLMs will know and use it against you. At least with a human, you can avoid that one person...
Which is the Hallmark of a great teammate, but then we won't need them anymore
But who has truly stress tested their teammates to see how much level of question bombardment they're willing to take? You really haven't, since they can't just generate tokens at the rate and consistency of an LLM.
It would be absolutely asinine to bombard your teammates like this and it would be a massive sign that you're not cut out for the work if you had to
I wasn't suggesting we literally bombard teammates.
The whole point is that with LLMs, you can explore ideas as deeply as you want without tiring them out or burning social capital. You're conflating this with poor judgment about what to ask humans and when.
Taking 'bombard' literally is itself pretty asinine when the real point is about using AI to get thoroughly informed before human collaboration.
And if using AI to explore questions deeply is a sign you're 'not cut out for the work,' then you're essentially requiring omniscience, because no one knows everything about every domain, especially as they constantly evolve.
Whereas with humans, you'll get valuable pushback for ideas that have already failed.
The wisdom to know what to ask humans and what to ask the machine.
I think we’re going to have to deal with the stories of shareholders wetting themselves over more layoffs more than we’re going to see higher quality software produced. Everyone is claiming huge productivity gains but generally software quality and new products being created seem at best unchanged. Where is all this new amazing software? It’s time to stop all the talk and show something. I don’t care that your SQL query was handled for you, thats not the bigger picture, that’s just talk.
This really resonates with me, I want to see the bigger picture as well.
This has been an industry wide problem at silicon valley for years now. For all their talks of changing the world, what we've gotten the last decade has been taxi and hotel apps. Nothing truly revolutionizing.
An internet full of pointless advertising and the invention of adblocks to hide that advertising.
Digital devices that track everything you do, that then generated so much data that the advertising actually got worse. Thereby the data was collected with the promise that the adverts would get more appropriate.
Now comes AI to make sense of the data and the training data (I.e., the internet) is being swamped with AI content so that the training data for AIs is becoming useless.
I wonder what is being invented to remove all the AI content from the training data.
The best part is those two things have only gotten worse over time. Turns out, they were never really that good of an idea, they just had money to burn and legislative holes to exploit. Now Uber is more expensive than Taxis ever were, and AirBNB is virtually useless now that they have to play the same legal ballgame as hotels. Oh, and that one is more expensive too.
Tech companies forget that software is easy, the real world is hard. Computers are very isolated and perfect environments. But building real stuff, in meatspace, has more variables than anyone can even conceptualize.
The revolution is happening at the top.
Top of what? It seems to be the same few people at the top for decades now.
And they are mostly sociopaths.
I do see the AI agent companies shipping like crazy. Cursor, Windsurf, Claude Code... they are adding features as if they have some magical workforce of tireless AI minions building them. Maybe they do!
Alot of what Cursor, windsurf etc. kinda just feels like the next logical step you take with the invention of LLM's but doesn't actually feel like the greater system of software has changed all that much except the pure volume one individual can produce now.
At least for C++, I try to use copilot only for generating testing and writing ancillary scripts. tbh it's only through hard-won lessons and epic misunderstandings and screw-ups that I've built a mental model that I can use to check and verify what it's attempting to do.
As much as I am definitely more productive when it comes to some dumb "JSON plumbing" feature of just adding a field to some protobuf, shuffling around some data, etc, I still can't quite trust it to not make a very subtle mistake or have it generate code that is in the same style of the current codebase (even using the system prompt to tell it as such). I've had it make such obvious mistakes that it doubles down on (either pushing back or not realizing in the first place) before I practically scream at it in the chat and then it says "oopsie haha my bad", e.g.
```c++
class Foo
{
int x_{};
public:
bool operator==(Foo const& other) const noexcept { return x_ == x_; // <- what about other.x_? }
};
```
I just don't know at this point how to get it (Gemini or Claude or any of the GPT) to actually not drop the same subtle mistakes that are very easy to miss in the prolific amount of code it tends to write.
That said, saying "cover this new feature with a comprehensive test suite" saves me from having to go through the verbose gtest setup, which I'm thoroughly grateful for.
I'm not entirely convinced this trend is because AI is letting people "manage fleets of agents".
I do think the trend of the tiny team is growing though and I think the real driver were the laysoffs and downsizings of 2023. People were skeptical if Twitter would survive Elon's massive staff cuts and technically the site has survived.
I think the era of the 2016-2020 empire building is coming to an end. Valuing a manager on their number of reports is now out of fashion and theres now no longer any reason to inflate team sizes.
I think the productivity improvement you can get just from having a decent LLM available to answer technical questions is significant enough already even without the whole Agent-based tool-in-a-loop thing.
This morning I used Claude 4 Sonnet to figure out how to build, package and ship a Docker container to GitHub Container Registry in 25 minutes start to finish. Without Claude's help I would expect that to take me a couple of hours at least... and there's a decent chance I would have got stuck on some minor point and given up in frustration.
Transcript: https://claude.ai/share/5f0e6547-a3e9-4252-98d0-56f3141c3694 - write-up: https://til.simonwillison.net/github/container-registry
I'm not denying LLMs are useful. I believe the trend was going to happen whether regardless of how useful LLMs are.
AI ended up being a convenient excuse for big tech to justify their layoffs, but Twitter already painted a story about how bloated some organizations were. Now that there is no longer any status in having 9,001 reports the pendulum has swing the other way - it's now sexy to brag about how little people you employ.
Their boilerplate works out of the box, you don't need to change anything. I recently packaged, signed, and published an OCI container into ghcr for the first time, it took about 5 to 10 minutes without touching any LLMs thanks to the quality of their documentation.
Eh I felt that way about the internet in 2010s. Seemed like virtually any question could be answered by a google query. People were making jokes that a programmer's job mostly consisted of looking things up on stack overflow. But then google started sucking and SO turned into another expertsexchange (which was itself good in the 2000s).
So far from what I've experienced AI coding agents automate away the looking things up on SO part (mostly by violating OSS licenses on Github). But that part is only bad because the existing tools for doing that were intentionally enshitified.
> expertsexchange
My vote for the unintentionally funniest company name. I wonder if they were aware when the landed on it, or if they were so deep in the process that it was too late to change course when they realized what they had done.
They're still around, long ago they bought the "experts-exchange" domain name and redirected the original, then at some point after that they abandoned the original entirely.
I wonder whether we will avoid some enshittification of AI because people are willing to pay for it. It doesn't all have to run off ads (although I'm sure all the free tiers will contain ads eventually).
Conceptually I find LLMs/AI broaden my skillset but slow down any processes that are deep in specific knowledge and context.
It is really nice to have that, it raises the floor on the skills I'm not good at.
> Valuing a manager on their number of reports is now out of fashion
I highly doubt human nature has changed enough to say that. It's just a down market.
Yeah but I think it's more that the money isn't there to throw bodies at an ok idea and hope you can turn revenue into profit down the line.
...unless you're shoveling AI itself, I guess.
"and technically the site has survived."
Only if you squint. If you look at the quality of the site, it has suffered tremendously.
The biggest "fuck you" are phishers buying blue checkmarks and putting the face of the CEO and owner to shill scams. But you also have just extremely trash content and clickbaits consistently getting (probably botted) likes and appearing in the top of feeds. You open a political thread and somehow there's a reply of a bear driving a bicycle as the top response.
Twitter is dead, just waiting for someone to call it.
Those are almost all management decisions, I expected a lot more crashes and security issues and a general inability to ship without taking things down.
Huh? Look at the hottest topic at the moment:
https://www.twz.com/news-features/u-s-has-attacked-irans-nuc...
and see for yourself if Twitter is dead.
I was literally just comparing my Twitter and Bluesky feeds. The only discussions worth reading were on Bluesky.
It's a shame. Twitter used to be the undefeated king of breaking news.
1. For the love of God please stop saying "Huh?" This is not Reddit, nor is the is comment you're replying to so unbelievably stupid that you are literally dumbfounded. I can tell, because you managed to put together a reply after the "Huh?"
2. 80% of the posts in that article-thingy are "no longer available".
So I can't hide in the masses watching Netflix anymore?
AI helps you cook code faster, but you still need to have a good understanding of the code. Just because the writing part is done quicker doesn't mean a developer can now shoulder more responsibility. This will only lead to burn out, because the human mind can only handle so much responsibility.
> but you still need to have a good understanding of the code
I've personally found this is where AI helps the most. I'm often building pretty sophisticated models that also need to scale, and nearly all SO/Google-able resources tend to be stuck at the level of "fit/predict" thinking that so many DS people remain limited to.
Being able to ask questions about non-trivial models as you build them, really diving into the details of exactly how certain performance improvements work and what trade offs there are, and even just getting feed back on your approach is a huge improvement in my ability to really land a solid understanding of the problem and my solution before writing a line of code.
Additionally, it's incredibly easy to make a simple mistake when modeling a complex problem and getting that immediate feedback is a kind of debugging you can otherwise only get on teams with multiple highly-skill people on them (which at a certain level is a luxury reserved only for people working a large companies).
For my kind of work, vibe-coding is laughably awful, primarily because there aren't tons of examples of large ML systems for the relatively unique problem you are often tasked with. But avoiding mistakes in the initial modeling process feels like a super power. On top of that, quickly being able to refactor early prototype code into real pipelines speeds up many of the most tedious parts of the process.
I agree in a lot of ways, but I also feel nervous that AI could lull me into a false sense of security. I think AI could easily convince you that you understand something when really you don't.
Regardless, I do find that o3 is great at auditing my plans or implementations. I will just ask "please audit this code" and it has like a 50% hit rate on giving valuable feedback to improve my work. This feels like it has a meaningful impact on improving the quality of the software that I write, and my understanding of its edge cases.
They often combine front end and back end roles (and sometimes sysadmin/devops/infrastructure) into one developer, so now I imagine they'll use AI to try and get even more. Burnout be damned, just going by their history.
> Just because the writing part is done quicker
The writing part was never the bottleneck to begin with...
Figuring out what to write has always been the bottleneck for code
AI doesn't eliminate that. It just changes it to figuring out if the AI wrote the right thing
Humans hate to think and make decisions, they like being told what to do.
So having an AI doing the dangerous part of thinking, leaves humans to do what they do best: follow orders.
Even better AI will take on the responsibility when anything fails: just get the AI to fix it, after all AI coded the mistake.
I read a few books the other day, The Million-dollar, One-person Business and Company of One. They both discuss how with the advances of code (to build a product with), the infrastructure to host them (with AWS so that you don't need to build data centers), and the network of people to sell to (the Internet in general, and more specifically social media, both organic and ads-based), the likelihood of running a large multi-million-dollar company all by yourself greatly increases in a way it has never done in the history of humanity before.
They were written before the advent of ChatGPT and LLMs in general, especially coding related ones, so the ceiling must be even greater now, and this is doubly true for technical founders, for LLMs aren't perfect and if your vibed code eventually breaks, you'll need to know how to fix it. But yes, in the future with agents doing work on your behalf, maybe your own work becomes less and less too.
There are already several million-dollar companies of one. Pieter Levels is one such famous builder on X. CertifyTheWeb.com is another one man millionaire product on HN.
Yes, Levels and many others are already covered in those books.
This may date me, but it feels like 1999 again where a small startup can disrupt an industry. Not just because of what LLMs can do in terms of delivered product, but because a small team can often turn on a problem so much faster than a big one can. I really hope that there are hundreds, if not thousands, of three to five person companies forming in basements right now ready to challenge the big players again.
I was also around in 1999. Most of the small companies had bad ideas that were able to get funding and died by 2001. A few were able to sell out to the bigger fools like Mark Cuban was able to sell broadcast.com to Yahoo for $10K per user.
The beatings will continue as profit improves
I think this is great for the world. So much less waste - imagine all the BMWs that aren’t going to be bought by middle managers or VCs, while people who know what they’re doing can build useful products.
AWS, GCP and other cloud providers play just as large of a role in allowing for tiny teams. Used to need an ops team of 10+ people to do all the stuff on premise that AWS can do
When I worked at a startup that tried to maximize revenue per employee, it was an absolute disaster for the customer. There was zero investment in quality - no dedicated QA and everyone was way too busy to worry about quality until something became a crisis. Code reviews were actively discouraged because it took people off of their assigned work to review other people's work. Automated testing and tooling were minimal. If you go to the company's subreddit, you'll see daily posts of major problems and people threatening class-action lawsuits. There were major privacy and security issues that were just ignored.
So did revenue per employee increase?
Really depends on the type of business you're in. In the startup I work in, I worked almost entirely on quality of service for the last year, rarely ever on the new features — because users want to pay for reliability. If there's no investment in quality, then either the business is making a stupid decision and will pay for it, or users don't really care about it as much as you think.
Theres two types of software, the ones no one uses, and the ones people complain about
Everyone should just write their own software then.
I've worked at a number of companies - the frequency and seriousness of customer issues was way beyond anything I've experienced anywhere else.
Some excellent ideas presented in the article. It doesn't matter if they all pan out, just that they expand our thinking into the realm of AI and its role in the future of business startups and operations.
Revenue per employee, to me, is an aside that distracts from the ideas presented.
It seems like a more and more recurring shareholder wet dream that companies could one day just be AI employees for digital things + robotic employees for physical things + maybe a human CEO "orchestrating" everything. No more icky employees siphoning off what should rightfully be profit for the owners. It's like this is some kind of moral imperative that business is always kind of low-key working towards. Are you rich and want to own something like a soup company? Just lease a fully-automated factory and a bunch of AI workers, and you're instantly shipping and making money! Is this capitalism's final end state?
Great question.
I want this idea to be drawn to an extreme where I can't buy soup or anything for that matter. Sure I will starve and die soon, but I feel the kind of burning the world will go through will be fun to watch. With tears of course.
If I'm buying soup, I'd prefer the manufacturer, the retailer, and any other part of the supply chain to be as efficient as possible, so they can compete in the market to offer me soup of a given quality at the lowest possible cost.
An individual consumer doesn't derive any benefit from companies missing out on automation opportunities.
Would you prefer to buy screws that are individually made on a lathe?
Personally, the best soups I’ve ever had were not made in kitchens that were optimized for efficiency or automation, they were optimized for quality.
They weren’t cheap soups, but they sure were good.
Luxury goods and staple goods have distinct optimizations, both viable for generating profits and economic utility.
A high end soup and an affordable soup might be serving two different markets.
Quality is a function of the ingredients used and the correct preparation. Neither of these things are something machines can’t do.
> Would you prefer to buy screws that are individually made on a lathe?
I don't think that was your point but pressed screws got way better properties than cut screws.
don't mind me I'm just moloch posting
https://slatestarcodex.com/2014/07/30/meditations-on-moloch/
I'd prefer not to live in a fully automated society where shareholders and CEOs reap all the profits while the rest of us scrape by on just enough UBI to prevent a revolution.
I don't understand this scenario. If everyone is on UBI then most people are essentially near poverty. Where are these CEOs deriving all of their profit from?
I personally think a far more likely scenario is that small businesses of one or a few people become vastly more commonplace. They will be able to do a lot more by themselves, including with less expertise in areas they may not have a lot of knowledge in. I don't think regular employees today should see LLMs as competition, rather they should see it as a tool they can use to level the playing field against current CEOs.
> I don't think regular employees today should see LLMs as competition, rather they should see it as a tool they can use to level the playing field against current CEOs.
LLMs aren't some magic silver bullet to elevate people out of poverty. Lack of access to capital is an extreme restriction on what most people can actually accomplish on their own, it doesn't matter if they have the worlds best LLM helping them or not.
It doesn't matter if you use an LLM to build the most brilliant business in the world if you can't afford to buy real world things to do real world business
Also, Historically when regular people decide to level the playing field against the ultra wealthy, they use violence
I don't think anyone should be expecting LLMs to be the great equalizer. The great equalizer has always been violent and probably always will be violence.
You're picturing a utopia at the limit of some idealized world. Try and take a second to return to planet earth.
There will not be a "quality" dial that you get to tweak to decide on your perfect quality of soup. There will be graduations, and you will be stuck with whatever the store provides. If you want medium quality soup, but the store only carries 3 brands of soup (because unlike in your utopia somebody actually has to maintain an inventory and relationships with their supply chain) and your favourite brand decides to bottom out their quality. It's not "good actually" because of economic whatever. Your soup just sucks now.
Oh but "the market will eventually provide a new brand" is a terrible strategy when they start spicing the soup with lead to give it cinnamon flavor or whatever.
I'm not an ethereal being. I'm a human, I need it to be good now. Not in theory land.
The wild thing is that they'll probably find entirely novel versions of this:
> they start spicing the soup with lead to give it cinnamon flavor or whatever
Like, we all know lead is bad, and we all know that humans are unscrupulous, but at least the human putting lead in the soup knows they're being unscrupulous at that time (probably). For an AI it would just be an entirely favorable optimization.
We're going to find out how far we can trust them, and the limits of that trust will determine where the people need to be.
You describe three potential undesirable outcomes:
- consolidation, such that there are only a few different choices of soup
- a race to the bottom in quality
- poisoning
These are all possibilities under our current system, and we have mechanisms (laws and market competition) which limit the extent to which they occur.
What is it about extreme automation technology that you think will increase these prevalence of these issues? By what mechanisms will these issues occur more frequently (rather than less frequently), as production technology becomes more capable?
Automation returns more power (abstract) to the owning class, which will make inequality more severe over time, which then results in said owning class having enough power to change legislation and market dynamics to their own desires. So, in turn - more poisoning, less quality, and more consolidation.
A lot of people think wealth inequality isn't a big deal, but I disagree. The more proportion of money a select few have in comparison to everyone else, the higher the likelihood those select few can mold society to their whim. Corruption thrives off of wealth inequality. Without it, it cannot exist.
Final end state where it eats itself? Who buys all that soup when the workers have no jobs and no money? How much soup can one human CEO drink? (And why isn't the CEO also replaced by the AI?)
The solution to this is to begin exporting your soup to other countries that have people who still work and have money. In the end it still eats itself though, but not before eating everything else.
Sounds unironically great if we could do it, the productivity improvements would allow dramatic improvements in living standards with moderate redistribution. I don't think this is where these llms are getting us.
If anyone can pay-as-you-go use a fully automated factory, and the factories are interchangeable, it seems like the value of capital is nearly zero in your envisioned future. Anyone with an idea for soup can start producing it with world class efficiency, prices for consumers should be low and variety should be sky-high.
AI gets top billing, but the assault via tax code on engineering employment is likely a bigger factor.
https://news.ycombinator.com/item?id=44226145
I think this is the beginning of the end of early stage venture capital in b2b saas. Growth capital will still be there, but increasingly there will be no reason to raise. It will empower individuals with actual skill sets, rather than those with fancy schools on their resume
I think bar for b2c, prosumer, SMB, yes.. folks want to see fast revenue growth (vs eyeballs & ideas)... but enterprise... not as much, and that's half the b2b market
Reverse there afaict, enterprise + defense tech are booming. AI means get to do a redo + extension of the code automation era. It's fairly obvious to buyers + investors this time around so don't even need to educate. Likewise, in gov/defense tech, palantir broke the dam, and most of our users there have an instinctive allergic reaction to palantir+xai, so pretty friendly.
" Do what you do best, and let AI do the rest".
Exactly the approach I'm taking with Tunnelmole, which as of right now is still a one person company with no investors.
I focused on coding, which I'm good at. I'm also reasonably good at content writing, I have some articles on Hackernoon before the age of AI.
So far AI has helped with
- Marketing ideas and strategies
- General advice on setting up a company
- Tax stuff, i.e what are my options for paying myself
- The logo. I used stable diffusion and an anime art model from CivitAI, had multiple candidates made, chose one, then did some minor touch ups in Gimp
I'm increasingly using it for more and more coding tasks as it gets better. I'll generally use it for anything repeatable and big refactors.
One of the biggest things coding wise working alone is Code Review. I don't have human colleagues at Tunnelmole who can review code for me. So I've gotten into the routine of having AI review all my changes. More than once, bugs have been prevented from being deployed to prod using this method.
> "Ushering in a new era."
It's ushering in a new era of valley bullshit. If only journalists tried to falsify their premise before blindly publishing it.
> Jack Clark whether AI’s coding ability meant “the age of the nerds” was over.
When was the "age of the nerds" exactly? What does that even mean? My interpretation is that it means "is the age of having to pay skilled programmers for quality work over?" Which explains Bloomberg's interest.
> “I think it’s actually going to be the era of the manager nerds now,” Clark replied. “I think being able to manage fleets of AI agents and orchestrate them is going to make people incredibly powerful.”
And they're all going to be people on a subscription model and locked into one particular LLM. It's not going to make anyone powerful other than the owner class. This is the worst type of lie. They don't believe any of this. They just really really hate having to pay your salary increases every year.
> AI is sometimes described as providing the capability of “infinite interns.”
More like infinite autistic toddlers. Sure. It can somehow play a perfect copy of Chopin after hearing it once. Is that really where business value comes from? Quickly ripping other people off so you can profit first?
The Bloomberg class I'm sure is so thrilled they don't even have the sense to question any of this self serving propaganda.
[flagged]
New York money is well aware of chump dust, as are the Bloomberg minions.. They know about it and frown on it heavily for real reasons.. 'Angelenos however, probably wise to do a piss test on your sales guys
source: Miami Vice re-runs
Like Johnny Depp in that movie..
How come you put two dots instead of one or three?
I thought I was the only one. :P
You’re gonna have to be more specific
The subhead makes a specific misstatement:
> Startups used to brag about valuations and venture capital. Now AI is making revenue per employee the new holy grail.
The corrected form is:
> Startups used to brag about valuations and venture capital. Now AI is making rate of revenue growth per employee the new holy grail.
Specifically, as with all growth capitalism, it is long-term irrelevant how much revenue each employee generates. The factor that is being measured is how much each employee increases the rate of growth of revenue. If a business is growing revenue at +5% YoY, then a worker that can increase that rate by 20% (to +6% YoY) is worth keeping; a worker that can only increase revenue by 5% contributed +0% YoY after the initial boost and will be replaced by automation, AI, etc. (This is also why tech won’t invest in technical debt: it may lower expenses, but those one-time efficiencies are typically irrelevant when increasing the rate of growth of income results in far more income than the costs of the debt.)
It’s true, especially with the “vibe” movement happening in real-time on X… “you can just do things” — I am building ai app layers b2c/b2b and while I do have an ml technical co-founder, I am largely scaling this with AI from strategy, visuals to coding. For example, with Claude created a framework for my company to scale, then built an AI powered dashboard in cursor around it as the command center. At scale we don’t need a team of more than ~5 to reach 7 fig MRR.
Greg Isenberg has some of the best takes on this on X. He articulates the paradigm shift extremely well.. @gregisenberg — one example: https://x.com/gregisenberg/status/1936083456611561932?s=46)
> 26. lots of first-time founders will build faster than veterans because they are more AI fluent/grew up on vlogging.
Ahh yes, fantastic insights.
time will tell
I guess it will. Maybe the next generation of builders will attribute their success to the formative experience of watching Sargon of Akkad.
I doubt it, but maybe?