btw, someone else having the same idea you have for a saas company has always been the case forever. Individuals taking shortcuts in quality to get to market faster has also been the case forever. There's nothing new about either of those two things.

And yet, Patrick Rothfuss's The Name Of The Wind is the same concept, and sold over a million copes,

Lev Grossman's The Magicians is again the same concept, sold millions, and was adapted into a 5 season TV series for SyFy.

If anything, the success of an idea only leads to a bigger appetite for that idea.

Google was not the first search engine.

It's pretty much the same idea as the above titles but omg it's so well written. Absolute must read!

BTW, The Magicians TV series might be the best thing SyFy ever made. It's got so much heart, it's properly funny, it's creative, it's epic despite a shoestring budget, and the characters stay with you long after you finish.

But yeah up until that point it was great!

The Magicians overdid it though. It's not a musical episode, iirc it's like 2 seasons

Edit: just to be clear, I still watched those 2 entire seasons, because the story is genuinely that good, plus I was invested. But I really wish they hadn't gone musical, it really messed with the mood of the series.

Add to this the 50 bajillion manga/anime's with the exact same trope.

The better comparison is probably with Percy Jackson, which isn't quite the same concept (being an American series, where boarding school fiction isn't quite as well-known a genre) but matches the ages, sense of discovery, and relationships to authority figures far better.

This isn't directly relevant to your point, but I really find it wild that people see two stories that have magic and a school in them and go "look, it's the same thing", especially when the genres and tropes of the two books are so utterly different. For that matter, Harry Potter is also nothing like Earthsea, which is another common reference point. I wonder if Americans just don't have as much experience with boarding school fiction to be able to categorise Harry Potter as a series?

Well, that kind of was exactly my point, although I feel now I didn't make it very clearly.

Someone might shoot down a prospective author who intends to write a book featuring a young protagonist getting a formal education in magic because it's "been done", but the resulting works are very different. It was a counterpoint to the article saying that we should not try to realise our ideas because someone somewhere has had "the same idea". They probably have had an idea which could be described in a similar fashion, but it doesn't really mean its the "same idea"

I probably stumbled a bit describing those books as the same concept when I should have put "same idea" in quotes

A wise friend of mine once said, in regard to "ideas are nothing; execution is everything": you can tell a thousand artists to paint a portrait of an Italian woman with a countryside landscape behind her but, good or bad, none of them are the mona lisa.

If every coked up SDR can build a tech stack, then every junior SWE can get superhuman SEO.

If every product has superhuman seo and engineering, and there are 10 or 100x more products, then probably everyone uses the exact right one for their needs, and quality for your specific usecase is higher. (More competition means more quality, more of every differentiator, including lower prices. )

In a world of zero marginal cost of production (turning ideas into reality with a prompt), maybe it’s hard for anyone to eke out profit margin; I can’t see what anyone’s edge would be in this world. The end state here is much more disruptive than “dang, a coked up sdr out-competed me on my SaaS ide-“.

When you write operational critical code, it matters. No one can blame “the AI made me do it” when things go down and hundreds of thousands of people are without service.

When your code can hurt people, it matters. You can’t burn someone’s eye with a laser then point to some AI agent when lawsuits start flying in.

When millions of dollars in production data is lost or corrupted, who is responsible? Not AI. Quality matters.

I keep hearing this one phrase about code quality again and again. Sure, no one cares about the dumb little linter failing your builds, but when code quality comes to responsibility, it goes hand in hand. It’s either that or your all working on hobby projects.

That's why AI is hurting us so much right now.

We were always trying to have quality in our project, whether it was for readability or for code evolution.

No, Steve, you don't name your 42 variables with only two letter and no you don't use Norse mythology for naming servers in your infrastructure. Yes Odin is the most powerful so it's the production server but Tyr for the source control and print server isn't really obvious.

Well now AI is Steve.

It will create nice little 300 lines functions with a block repeating 6 times. You know that you will have 6 fix to make instead of one if this block was in a simple function.

It's not instinct a this point, is pure knowledge screaming "it's wrong".

And you now realise that the hidden strength from your craft wasn't about coding the best binary tree search algorithm, it was about knowing the underlying soft unknowns that really made it software.

We have a strong feeling that we're watching dozens of kids running with scissors and we don't know whether it's really scissors, we're just getting too old for this shit, or if we should just stop "progress" because we don't like it.

We're the horse breeders when everyone discovered cars.

But that cost is not trivial. In some topics (but not limited) like medial devices, the legal liability would just bankrupt the company. Not so cheap compared to hiring a few humans. I’m picking an obvious cases here, but there are many others.

> Norse mythology for naming servers in your infrastructure

Ouch, yeah seen this a few times, outside of Scandinavia.

Doesn't sound like a hobby side project. Sounds like a business. And then yeah, you get all that comes with it.

Your customers are paying you $1000 a month to handle a process that is worth $10k to them and might be worth $50k to their customers. If you lose that data you haven’t lost $1000 of data. Even if you only made $100 off of the transaction.

for personal side projects of the "handy SAAS thing that gives all your JSON a lustrous sheen" variety, he's probably more on target that quality seems to have gone by the wayside.

https://x.com/karpathy/status/2018043254986703167

I have plenty of other ideas for what to build on top of it: offering an SDK and APIs so you can vibe-code the UI you want, a built-in podcast listener, using news from aggregated feeds to build a personalized AI feed. But the first step is to reach the Google Reader feature set minus social features.

API access is worth chasing. There was something I wanted to do with Feedly (I've already forgotten what it was) but once I saw their APIs were hidden behind some enterprise level plan, that was the end of that. If we're in a world where everyone has a personal AI agent, giving their agent an API key to their RSS sync account... that might have some interest.

Feedly seems hostile to third-party client access (ie mobile & desktop apps), so being friendlier towards RSS clients could be of interest.

Personalized AI feed is a good idea but you don't have all the personalized year of context that my Claude does. My AI agent is (probably) going to do a better job of choosing the most relevant stuff.

And personally, less interested in podcasts in my RSS app. That's something for Pocket Casts / AntennaPod. I like my audio separate from my RSS. But that's me.

Yes, enterprise is certainly where the money is (Feedly's plans start at $1600/month...), but as a solo dev working on a side-project, that's not an accessible market for me anyway. So I try to create a service that's simple and cheap.

> My AI agent is (probably) going to do a better job of choosing the most relevant stuff.

The idea would be basically: the feed reader know the user's interests because of the subscriptions, and knows the last time the user logged in. So it can filter what happened since then; it can also order the posts by relevance, allowing the user to catch up. And in a second step, an agent could even write the posts dynamically, summarizing information gathered from the user's feed, possibly even adjusted to the user's level of knowledge and offering background info where needed.

> And personally, less interested in podcasts in my RSS app. That's something for Pocket Casts / AntennaPod. I like my audio separate from my RSS.

There are some feeds that are more like a mixture of text and podcast. I usually read only the text, but sometimes it catches my interest and I want to listen to one or two posts. That's when I start hating the lack of podcast support in Feedly.

Best of luck though, I think this is a very promising space. (But my bet is you can do all the interesting stuff in vibe-coded thin UI + OSS pipeline.)

Simplicity. I can get you reading your first feed in under a minute. Also, I am not really thinking about monetization right now, but I am building a feed reader I want to use. I wouldn't want to spend $13 a month for it.

> thin UI + OSS pipeline

No, the UI isn't that thin. I am optimizing it to minimize my costs for operating it. Everything I can do inside the client is done inside the client. Interactions with the server are mostly limited to polling every 2 minutes for feed updates, and sending read markers after 3 seconds of inactivity. Feed data is stored on CDN, compressed.

For people who do not want to use self-hosted services (which generally includes me), it offers simplicity. Open the page, choose Google as auth provider, confirm, and you will get a friendly start page. Click on 'follow' on one of the feeds, and you can start reading immediately. The UI is more like Facebook or X, so basically, you just need to scroll. Either in a feed of your choice, or all your feeds. It's designed to work well on small mobile screens, tablets, and desktops, with great keyboard support on the latter. Larger screens use two or three columns.

Margins are worse, but selling is easier. If you've got a thing you can be sure that someone somewhere will give you money for it.

Great point. AI remixes and rips-off existing code-bases in a manner that is impossible to attribute copyright violation making it legal. ie, Perfect cloning. In a world where cloning is legal, the engineering cost of product drops to zero. That is where software production could be headed. What remains is marketing/distribution/sales.

There will remain niches solving "hard problems" which cant be cloned, but those will be rare. Hard problems are where a lot of engineering complexity resides, involving interacting components for which there are no examples in training datasets to copy from. For example, a complex distributed system or hardware with multiple nuanced tradeoffs.

Don't forget liability & compliance :)

There will be people who will pay for "human coded" software if it is better. Quality is always a differentiator that some people will pay for.

Stuff for old men like me

That's what I mean by margins being a significant difference.

You can't hope to succeed by building something cool without distribution already figured out. If you haven't put the work in building a social following, you're pretty much locked into pay to play (which isn't horrible if you target small targeted bloggers/youtubers/etc, but it's not my bag). OpenClaw exploded because Peter has >100k twitter followers and among them are plenty of people who themselves have a ton of followers.

So, if you're building, you also need to focus on building an audience.

The high touch enterprise sales strategy is solid though, and easier to bootstrap. That's why Alex Hormozi and Dan Martell push people getting started that way.

The game has change. The ‘how’ we build it is easy. The ‘what’ we build is and always has been the hardest part of any SaaS or business.

This is what the promptfondlers don't want to admit: the how has been easy for a long time. This last, I dunno, 35 years or so, Visual Basic, Delphi, whatnot, producing code has been very easy. You don't need a fundamentally fascist probabilistic nightmare to do it. The hard problems are indeed is "what" to build and how we maintain it. There's only hype. There's no results. https://mikelovesrobots.substack.com/p/wheres-the-shovelware...

As for fascism, check https://blog.bgcarlisle.com/2025/05/16/a-plausible-scalable-... for example

> By “fascist” in this context, I mean that it is well suited to centralizing authority, eliminating checks on that authority and advancing an anti-science agenda.

Or check Woodrow Hartzog & Jessica Silbey, How AI Destroys Institutions , 77 UC Law Journal (2026). Available at: https://scholarship.law.bu.edu/faculty_scholarship/4179

I’m not so sure about that. It’s very easy to take your own knowledge for granted. Most people can’t do what we do. Most of my customers couldn’t even express what they wanted.

You can test this, if you talk about your idea to three people and one of them says we're already doing that and the other two think you're insane, you'd be safe.

On most days I am resistant to stereotypes about "the welfare state" eroding incentives for entrepreneurism and innovation. But if you're going to wave it in my face like that, I might have to reconsider.

I say the only way to build a successful long term product is by focusing on quality, ESPECIALLY when the competition is shitting out crap.

More so this complexity requires that you have support for your users, and QA of weird functional interactions across systems boundaries that you just can't test for when actually writing the code.

This gets expensive really fast.

Complex software is hard, yo.

Fortunately experienced developers are in the best position to use these tools properly, evaluate what works and what doesn't. We might drown in slop first though.

Let's say you could vibe your own replacement to a $20/month app in 16 hours. Congratulations, you did work valued at an $15/hour less token expenses (over 1 year).

Quality will matter the most in 2026. Specifically because the barrier-to-entry for making software is down there will of course be a lot of poor quality software, which will break, expose customer data, be bloated, etc. Customers will have more options, and this will allow them to be more discerning. Open source, clean code, low dependencies...these are things that can be evaluated by HN crowd types, but it's also something that an LLM can evaluate.

We are entering into an age of software taste. For those of us that have developed taste over the years, we become the taste makers in that we care how things are built, and know what we're looking for. This applies on the supply side, when our taste drives the LLM, and on the consumption side, when we can help the masses evaluate what to use and what not to use.

NB: this is all speculation expressed as fact, in keeping with the OP's style.

Quality isn’t a differentiator if the market is saturated with indistinguishable garbage. Everything is made in sweatshops out of the cheapest plastic available, and I don’t see why software isn’t next in line.

For a long time the stereotypical “young professional” look was tied closely to just a few mainstream retailers (Banana Republic for example), but over the last ~15 years a wider range of smaller or more specialized brands has entered the space: Alex Mill, Spier and Mackay, etc.

But even ignoring that your analogy doesn’t quite fit since price plays a significant role in clothing purchasing decisions: Fast fashion succeeds largely because it is cheap.

If reasonably priced, higher-quality alternatives were accessible people would buy them. It’s partly why certain brands have grown in popularity (Carhartt, for example).

https://en.wikipedia.org/wiki/The_Market_for_Lemons

N of 1, obviously, but this isn’t as outlandish as you wanted to make it seem here.

Lets assume this is true - how on earth are they to determine that your code doesn't have any glaring security holes but the 2h vibe-coded app has more holes than the Swiss is able to put into their cheese[1]?

I really want to know how customers can tell the difference between very pretty crap and your stuff?

-------------

[1] Yeah, I know it doesn't work like that.

What these customers are going to do is do a summary discard of almost all the choices but say 3 to 5 and go from there.

The problem is now how to be consistently on that top list. And that's marketing's problem.

"Dear Claude, please make me a clone of <fancy new saas> but make <these changes specific to my tastes>".

For many things, it's probably not "select the one of 100 that fits my taste", it's probably going to be to just make your own personal version that fits your taste in the first place. And, probably, never share that anywhere.

There was one PM at my ex-job that showed a dashboard for... well... i honestly didnt understand. I think it was some uptime checks. It broke during his presentation.

There's a company I hired at that "built an ERP in 5 days and is shipping the product in June". Same thing happened, it broke when presenting. Basic feature suggestions just returned a "Yes, we can do that!" (they meant they can tell Claude to do it, not that the product could do it).

Maybe at some point non-engineers can prompt build, but for now I'd say we're pretty safe. I think engineers give themselves too little credit. Being able to read code is an amazing tool that can only be sharpened through skill.

Lastly, I think I commented this ~2 years ago as well. If your product is vibe-codable and is replacing customers, it's a shit product. Similarly, if you can outsource your product on fiverr, its's a shit product.

- making a markdown file with all specs, details and plans - asking claude to search online and suggest some approaches

is a better alternative to doing the research yourself.

I like his sense of humour.

Based on Louis Rossmann's rant 3 days ago [1], it seems Gemini has got you covered on that front too.

[1] https://www.youtube.com/watch?v=6uKZ84zwJI0

Build something else!

I mean, it seems at the very least, that open source and in-house production has a natural advantage here? If the marginal cost of software production is now free, then FOSS/in-house just got easier to create and maintain too. Does that make it easier for FOSS/in-house, both available without a subscription to an external third party, undermine "sales critter" SAAS, by the author's own premises?

Isn't that just SAP, er, I mean SAAS as it has been for a decade?

That's funny and all but it can't really be true that quality doesn't matter. It has to matter at some point. Maybe it doesn't matter during the initial sales cycle; I've seen it happen: the CEO sees a slick demo that works, every user / developper rolls their eyes and try to warn them, they don't listen, and the deal is done.

But eventually if the thing that's supposed to be done, isn't, something will have to give. Even if at first they fire all the eye-rollers and replace them with obedient corporate drones, if the think isn't working and it's on the critical path, it will have to be replaced by something that actually does work.

Most people will barely make anything, some will be able to supplement their income, very few will be able to make a living. Even less will become "rich". For every product that blows up, there are thousands that will barely make anything.

But of course, it all depends on what your product does. If you make the millionth TODO / GenAI image editor / food calculator app and hope to make some money, good luck.

Yes, a very hyped mega-corp should be building and replacing all productivity software; why leave room for competitors when a single company can do it all? What can go wrong?

I pay for a SaaS app that tracks my finances, but it's not that great and missing some features I would like. Very soon I expect I'll be able to get a better, local-first replacement tailored to my needs by prompting Claude & Friends.

1) There are a lot of cases where aggregated user data, even if anonymized, allows for insights that you can't get using just your own data.

2) The software is really just a stand in for a process. A way of doing something, like record keeping or tax filing, etc. A lot of times it makes sense to follow an already established process rather than creating your own. You are less likely to encounter unexpected pitfalls that way.

I don't see how you can overcome those just by having an AI that can build simple crud apps at will.

It's a huge trope to think your product didn't work for the market because the marketers beat you. I used to be that kind of developer until I made some products that people actually wanted.

But he's right that the software market is changing. Software will be easier to build and require less people to build it. So more, smaller companies will compete for market share. Margins will be cut and the consumers will get more of what they want for a lower price.

I think this is called a working market. It's what it looks like when capitalism actually works.

This could be the end of enshitification.

The specifics matter here. If you run a CRM for Dentists, can someone replicate your product in a weekend? I'm going to guess that dentists have some esoteric needs related to their CRM, and it's a little more complicated than an outsider might guess.

So what is the threat model? That a dentist is going to get fed up and try to DIY? I think you should encourage that, so they'll see what goes into it. That a 22 year old chooses "CRM for Dentists" as a thing to vibe-code over a weekend? Again, good luck with that.

I really dislike this SaaSocalypse fear mongering, because it's just not based in reality. Show me five examples of established SaaS companies being wiped out by vibe coding.

One guy loses $2400/mo in revenue

200 pool cleaners can now easily track their clients filter change dates without paying $12/mo for a calendar script (something that 20 years ago would have been a one time $3 purchase).