>A few hours used to get a programmer a script; now it gets a polished app, and it gets a non-programmer something instead of nothing.
I think "polished" is doing some very heavy lifting in that sentence.
Don't get me wrong: I literally spend every weekday generating web UIs with Claude: I'm not anti-AI at all.
But to claim anything Claude produces as "polished" means limiting yourself to a very narrow definition of that term! If it didn't, people like me wouldn't get paid six figure salaries still (we'd be replaced by the AI that produces "polished" code).
Claude and Codex produce far more polished work than any of my side projects. Are they competitive with entire teams of professional developers working on standard (multi-week) project development schedules? I don't know. But it's weird to suggest that "unpolished" is their big problem.
I'm sure it helps to have some design sense, or something. Whatever it is, I don't have it.
Each time I've tried to get some Web UI going with the AIs, they all look terrible. They may work, but, they're terrible. I know they're terrible. I can't really express why they're terrible, just that I don't like them. So, I cannot really explain to the AI why they're bad so they can fix it.
Its just a blindness of mine that I've had forever.
I keep trying this for very simple android apps and every pass between the Claude design tool and Claude code results in an uglier and less performant app. Woe is me, I have to actually learn this kotlin design / templating system.
Hard agree. Modern frontier models will give you a surprisingly solid first pass at a concept if prompted with a clear initial description and a handful of well-guided feature refinement rounds in an afternoon. That said, if you have a high bar for things like concise, well-architected codebases and strong, intuitive UX (the two main components of “polish” imo), you’ll find it takes a long time to get from Pareto-decent to truly polished.
The lack of polish is usually in the actual usability of the feature set (see yesterday's "polished photoshop clone" where the keyboard shortcuts just don't work)
I think you can use these tools to get something very polished. I have found that a lot of people get "tricked" into thinking that their result is good, but the people making these things and the people with expertise in the domain are often not that close to each other.
It is valuable for a thing to get used to figure out if it's actually any good.
Though the obvious glib thing is a lot of people build these tools and then don't use it.
(No not you, person who decides to reply to this saying 'but _I_ use it')
We're entering the age of personal software. Everyone who is motivated will have a catalog of unreleased custom software to solve their own problems. No need for support, no need for community management. It's often said that LLMs can get you to an 80% prototype very quickly but polishing to a public release is really hard and still takes a ton of time. Well, 80% is good enough for my own use, which means solving problems quickly, for myself.
Indeed! After years of wanting to finally I was able to create my own budgeting app with my own needs with auto ingestion of transactions, an Uscis case tracker, gh stars organizer, you name it. The list of ideas I accrued over the years is just now 1 prompt away from being a testable iterable app that I can self host and use behind tailscale. Good enough for my wife even. There is something more addicting about building for others though, I did open source my apps in case it helps anyone, but I presume its why chefs become chefs, put something out and immediately see its effect on others. I guess now marketing will be even more important. Even for free and genuinely morally good ideas that don't have an agenda behind them and I am a bit scared that most people who care about marketing that much are not trying to spread free public utility around...
I posted some free little games/utilities i made just below in this thread and can attest that in general you do have to have thick skin and brace yourself for derision, abuse and downvotes on say Reddit or Linkedin if you dare to acknowledge you used AI in any way to create any form of content right now.
Which is silly because it doesn't help anything to deride creators that use the same tools the big corporations use.
Absolutely! And who wants to deal with warts you couldn't even imagine but will likely encounter if project was github hosted, gained traction, and gets the typical commentary/requests/abuse from randoms.
I've been considering opening a 2nd GitHub profile just for AI stuff. I'll link to it from my regular profile, I'm not hiding it! If someone's checking me out for a job or whatever, it might be nice to have differentiate between ideas of mine, and things I've spent time making.
I’ve taken the viewpoint that a domain is basically a cup of coffee per year in cost and static web hosting is literally free so have been putting a ton of stuff up. None of these have ads or make me any money, just listing here as an example of the ease of sharing;
tfmbot.com - an implementation of my favourite board game complete with ai that plays against you
thedailycheat.com - a save game editor for a game Newstower that was fun but a little grindy ao i asked ai to figure out the save game format and make this online save editor
grandcheaten.com - a guide and save game editor for jagged alliance 3
Basically all personal projects but i find sharing them out rewarding. Occasionally i’ll get abuse or large amounts of downvotes for daring to share ai generated content but, shrug, it’s useful to me and i’ve also had lots of messages of appreciation.
In the past, the effort it took to make a polished tool was much higher and hardly worth it for one self alone. We've pushed that bar up significantly.
I believe this also: any domain name lookup service is actually a domain squatting service. When you look up a domain they buy it, then sell it to you for a markup.
If you look up the domain using their system and it doesn't exist, the person running the system could come in and buy it out from under you, and then offer to sell it back to you at a higher price. tbc I'm not accusing anyone of doing this, but that's the failure mode people are afraid of.
LLMs causing a flood of apps is an interesting problem to watch the indiehackers around me try to navigate. I've watched them try to sneak around automods on Reddit or post just the right way to not piss people off, do Instagram ads, enter game jams... One guy actually had a lot of luck on Facebook groups, actually got some paid members for his Chinese learning app, which surprised me because basically every white guy that moves to Taiwan creates a Chinese learning app. Couldn't believe one actually got revenue.
Idk, maybe this is why compared to my peers I don't have much money, or the co-op I founded doesn't generate as much revenue as it probably could, but something just feels icky about it. I don't want to try to stick my head above the water of ten thousand of the same people doing the same thing. The effort to payoff ratio seems broken. I just want to build things, some of which are useful or fun. Maybe I'm too much of a dirty anarchist but it seems a lot more fun to me to make something and give it away for free that someone else is trying to charge for, than to try to monetize everything I do.
But I do wonder what's going to happen to indie game devs and hackers. I suspect that the burn-fast quick ramp up and validation strategy will begin fail, and what will make software worth paying for in the future is software that grew slowly and proved out its usefulness, maintenance, and development cadence over a long period of time. Before, you could validate an idea relatively quickly, maybe now you can just pick a couple ideas, stick to them, and cross your fingers.
Self proclaimed Indiehackers for the most part don't produce anything of value anyhow and have been in the flood the market game for a decade. It's very much a case of X made $1000 doing Y so now I'm doing Y but for Z and when it doesn't get instant traction I am going to stop updating it/stop paying for the hosting in 2 months.
It's how we ended up with more niche job boards with scraped postings than there are altcoins and seven million sites that are nomadlist but in a different niche with bad data.
It looks like you are downvoted and flagged here. Maybe that should be your signal.
>I’m surprised you care more about someone pointing out slop than actual slop.
Your comments are slop and I do care about it. I care that when I open the comments on HN now I have to scroll past a bunch of people loudly proclaiming they were unable to read the article or how it sounds like AI, etc, etc.
You are copying and pasting the same thing all over the place, that is slop/spam/what have you but it's not contributing.
The analysis of existing tools and why they are annoying is spot on, no one is paying for it, so existing solutions try to recoupe 4.57$ monthly revenue from ads.
But that means that the product you are launching is trying to undercut 4.57$/mo for 0$/mo, which is a losing proposition. Any proposition that tries to undercut a market that has already been driven to bottom prices will die of hunger.
Also, in terms of execution, there's a design choice "recent lookups" that is quite unconventional, and will kill the trust of your product, and your personal reputation.
By leaking your user's input and data, you have outed yourself as someone that cannot be trusted with user or commercial data.
You have died twice, first from hunger and then from reputational death. I am in a better position for having launched nothing than you are in for launching this.
Did you read the article? The entire point of building this was to solve a problem for myself. If anyone else wants to use it, that's also fine with me, but the goal of this was not to monetize this nor turn it into a serious project or business.
>A few hours used to get a programmer a script; now it gets a polished app, and it gets a non-programmer something instead of nothing.
I think "polished" is doing some very heavy lifting in that sentence.
Don't get me wrong: I literally spend every weekday generating web UIs with Claude: I'm not anti-AI at all.
But to claim anything Claude produces as "polished" means limiting yourself to a very narrow definition of that term! If it didn't, people like me wouldn't get paid six figure salaries still (we'd be replaced by the AI that produces "polished" code).
Claude and Codex produce far more polished work than any of my side projects. Are they competitive with entire teams of professional developers working on standard (multi-week) project development schedules? I don't know. But it's weird to suggest that "unpolished" is their big problem.
If anything: too polished!
(Like the words in this blog post, by the way).
Polished turds maybe. Making something actually good (stable, intuitive, not lying to you in some way) still requires a lot of attention and effort.
Sure, of course it does. It's just different effort now.
I'm sure it helps to have some design sense, or something. Whatever it is, I don't have it.
Each time I've tried to get some Web UI going with the AIs, they all look terrible. They may work, but, they're terrible. I know they're terrible. I can't really express why they're terrible, just that I don't like them. So, I cannot really explain to the AI why they're bad so they can fix it.
Its just a blindness of mine that I've had forever.
Give screenshots to Claude and ask for design review. Surprisingly effective. At this point opus 5.5 can do a lot with that exact type of feedback.
I keep trying this for very simple android apps and every pass between the Claude design tool and Claude code results in an uglier and less performant app. Woe is me, I have to actually learn this kotlin design / templating system.
I wouldn't read too much into it, as that sentence is also clearly AI juice.
Hard agree. Modern frontier models will give you a surprisingly solid first pass at a concept if prompted with a clear initial description and a handful of well-guided feature refinement rounds in an afternoon. That said, if you have a high bar for things like concise, well-architected codebases and strong, intuitive UX (the two main components of “polish” imo), you’ll find it takes a long time to get from Pareto-decent to truly polished.
The lack of polish is usually in the actual usability of the feature set (see yesterday's "polished photoshop clone" where the keyboard shortcuts just don't work)
I think you can use these tools to get something very polished. I have found that a lot of people get "tricked" into thinking that their result is good, but the people making these things and the people with expertise in the domain are often not that close to each other.
It is valuable for a thing to get used to figure out if it's actually any good.
Though the obvious glib thing is a lot of people build these tools and then don't use it.
(No not you, person who decides to reply to this saying 'but _I_ use it')
I have taken to describing Claude as adding a veneer, not polish, which is exactly what makes it so insidious in the workplace when misapplied.
We're entering the age of personal software. Everyone who is motivated will have a catalog of unreleased custom software to solve their own problems. No need for support, no need for community management. It's often said that LLMs can get you to an 80% prototype very quickly but polishing to a public release is really hard and still takes a ton of time. Well, 80% is good enough for my own use, which means solving problems quickly, for myself.
Indeed! After years of wanting to finally I was able to create my own budgeting app with my own needs with auto ingestion of transactions, an Uscis case tracker, gh stars organizer, you name it. The list of ideas I accrued over the years is just now 1 prompt away from being a testable iterable app that I can self host and use behind tailscale. Good enough for my wife even. There is something more addicting about building for others though, I did open source my apps in case it helps anyone, but I presume its why chefs become chefs, put something out and immediately see its effect on others. I guess now marketing will be even more important. Even for free and genuinely morally good ideas that don't have an agenda behind them and I am a bit scared that most people who care about marketing that much are not trying to spread free public utility around...
I posted some free little games/utilities i made just below in this thread and can attest that in general you do have to have thick skin and brace yourself for derision, abuse and downvotes on say Reddit or Linkedin if you dare to acknowledge you used AI in any way to create any form of content right now.
Which is silly because it doesn't help anything to deride creators that use the same tools the big corporations use.
Absolutely! And who wants to deal with warts you couldn't even imagine but will likely encounter if project was github hosted, gained traction, and gets the typical commentary/requests/abuse from randoms.
It is partially the result from chilling effects of hosting AI generated code on your public portfolio. Not worth the ire.
I've been considering opening a 2nd GitHub profile just for AI stuff. I'll link to it from my regular profile, I'm not hiding it! If someone's checking me out for a job or whatever, it might be nice to have differentiate between ideas of mine, and things I've spent time making.
I’ve taken the viewpoint that a domain is basically a cup of coffee per year in cost and static web hosting is literally free so have been putting a ton of stuff up. None of these have ads or make me any money, just listing here as an example of the ease of sharing;
tfmbot.com - an implementation of my favourite board game complete with ai that plays against you
thedailycheat.com - a save game editor for a game Newstower that was fun but a little grindy ao i asked ai to figure out the save game format and make this online save editor
grandcheaten.com - a guide and save game editor for jagged alliance 3
Basically all personal projects but i find sharing them out rewarding. Occasionally i’ll get abuse or large amounts of downvotes for daring to share ai generated content but, shrug, it’s useful to me and i’ve also had lots of messages of appreciation.
I'm going through the same process.
I think the author nailed it with this sentence:
> The audience can be one person.
In the past, the effort it took to make a polished tool was much higher and hardly worth it for one self alone. We've pushed that bar up significantly.
This was why I built https://checkna.me
Every time I see a domain name input checkbox I can't not think about domain squatting.
Don't worry, they won't squat the domain, but they will leak your domain in the "recent lookups" section so that others may squat it.
What do you mean?
I believe this also: any domain name lookup service is actually a domain squatting service. When you look up a domain they buy it, then sell it to you for a markup.
If you look up the domain using their system and it doesn't exist, the person running the system could come in and buy it out from under you, and then offer to sell it back to you at a higher price. tbc I'm not accusing anyone of doing this, but that's the failure mode people are afraid of.
LLMs causing a flood of apps is an interesting problem to watch the indiehackers around me try to navigate. I've watched them try to sneak around automods on Reddit or post just the right way to not piss people off, do Instagram ads, enter game jams... One guy actually had a lot of luck on Facebook groups, actually got some paid members for his Chinese learning app, which surprised me because basically every white guy that moves to Taiwan creates a Chinese learning app. Couldn't believe one actually got revenue.
Idk, maybe this is why compared to my peers I don't have much money, or the co-op I founded doesn't generate as much revenue as it probably could, but something just feels icky about it. I don't want to try to stick my head above the water of ten thousand of the same people doing the same thing. The effort to payoff ratio seems broken. I just want to build things, some of which are useful or fun. Maybe I'm too much of a dirty anarchist but it seems a lot more fun to me to make something and give it away for free that someone else is trying to charge for, than to try to monetize everything I do.
But I do wonder what's going to happen to indie game devs and hackers. I suspect that the burn-fast quick ramp up and validation strategy will begin fail, and what will make software worth paying for in the future is software that grew slowly and proved out its usefulness, maintenance, and development cadence over a long period of time. Before, you could validate an idea relatively quickly, maybe now you can just pick a couple ideas, stick to them, and cross your fingers.
Self proclaimed Indiehackers for the most part don't produce anything of value anyhow and have been in the flood the market game for a decade. It's very much a case of X made $1000 doing Y so now I'm doing Y but for Z and when it doesn't get instant traction I am going to stop updating it/stop paying for the hosting in 2 months.
It's how we ended up with more niche job boards with scraped postings than there are altcoins and seven million sites that are nomadlist but in a different niche with bad data.
But I do wonder what's going to happen to indie game devs and hackers.
We will find out who's good at game design, and who was just OK at game design and really good at coding.
We'll also hopefully find a few thousand people who are really good at game design who were previously gatekept by being really bad at coding.
hell yes we will
You won't find them, because they will be buried under a giant mountain of slop generated by those who aren't good at either.
It's not that hard to crowdsource curating of slop.
Isn't it? Do you have an example of it being done?
You're currently on a website that crowdsources the curation of slop :)
Whether you think it's doing a good job or not, I suppose that's your call
[flagged]
[flagged]
Thank you for your service.
Can you stop spamming articles with this. You care, that is great don't read them but you aren't helping anyone with this offtopic spam.
https://news.ycombinator.com/newsguidelines.html
>Don't post generated text or AI-edited text. HN is for conversation between humans.
The issue is AI generated text, the issue is not that someone points out AI generated text.
That is for comments. Click on the comment history this user is spamming the exact message so generated text breaking the rule you posted.
I don't care if you don't want to read ai written or edited posts. You don't need to announce it in every thread just close the article and move on.
Maybe you should try just ignoring the comment.
> you aren’t helping anyone
Are you sure about that? The upvotes say otherwise.
If I start getting negative feedback I will stop commenting. I’m surprised you care more about someone pointing out slop than actual slop.
It looks like you are downvoted and flagged here. Maybe that should be your signal.
>I’m surprised you care more about someone pointing out slop than actual slop.
Your comments are slop and I do care about it. I care that when I open the comments on HN now I have to scroll past a bunch of people loudly proclaiming they were unable to read the article or how it sounds like AI, etc, etc.
You are copying and pasting the same thing all over the place, that is slop/spam/what have you but it's not contributing.
lol
Do you work at Pangram? Most of your comments are links to Pangram and an encouragement not to waste time.
For curiosity's sake—not criticism—how accurate is the Pangram score? If this is AI assisted at all then what was the authoring process like?
No I do not. I wish I did. They have such an amazing product that’s not well utilized in this slop era.
The analysis of existing tools and why they are annoying is spot on, no one is paying for it, so existing solutions try to recoupe 4.57$ monthly revenue from ads.
But that means that the product you are launching is trying to undercut 4.57$/mo for 0$/mo, which is a losing proposition. Any proposition that tries to undercut a market that has already been driven to bottom prices will die of hunger.
Also, in terms of execution, there's a design choice "recent lookups" that is quite unconventional, and will kill the trust of your product, and your personal reputation.
By leaking your user's input and data, you have outed yourself as someone that cannot be trusted with user or commercial data.
You have died twice, first from hunger and then from reputational death. I am in a better position for having launched nothing than you are in for launching this.
> But that means that the product you are launching is trying to undercut 4.57$/mo for 0$/mo, which is a losing proposition.
What if I'm doing this on purpose on my own dime, to kill for profit corporations out of spite?
Aka the Manfred Macx doctrine of startups.
Did you read the article? The entire point of building this was to solve a problem for myself. If anyone else wants to use it, that's also fine with me, but the goal of this was not to monetize this nor turn it into a serious project or business.