With respect I feel like the author is missing a whole bunch here about the point of a website.
It's not just content/info/data, it's a performance (in the creative sense).
Brands spend a lot of time honing their appearance - not just fonts and colours but the whole composition and visual pacing - their entire "say something without saying anything at all" aspect etc. Just walk through any place with physical shops and really look at how the stores have worked on their appearance and how they present themselves to customers. They're not just selling a product, they're selling a lifestyle/feeling/etc/etc. They're not just going to give that creative control away to some LLM.
Another way to think of it is instead of people watching a movie or play when they go to the cinema or theater, they're just given the script to read. Same information but the entire artistry of both the performers and the directors is totally absent, leaving it up to each reader to imagine the delivery of lines or the scene's setting etc.
I think on HN and in tech in general people seem to forget that "the first bite is with the eye", and that is why "normal people" never liked or used RSS. The desire to leave our mark and to create (and view!) visually appealing things seems to be pretty innate in humans - we've been doing it since cave paintings. I struggle to think of a world where we just hand that over to AIs and humans have zero creative control.
To add to this, the OP's vision benefits the user -- reducing a business's value to its actual raw value as a service rather than a brand. For me, it sounds great.
But the business's incentives are in the exact opposite direction. That opposite direction is the whole point of branding. They want their service to have a vibe, a personality, something you irrationally value beyond its raw value as a service.
> something you irrationally value beyond its raw value as a service
Sometimes that feeling is the value. Sure my plants don’t care if they live in a cheap plastic pot off Amazon or a nice pot from the overpriced gardening store selling at a 200% markup, but I care. Sitting in my balcony surrounded by cheap disposable clutter feels different than enjoying the outdoors amidst quality vibes.
Some things are commodities, some are not. The point is only that it's in the interest of commodity businesses to convince you they are not selling commodities. That sleight of hand doesn't prevent genuine quality and artistry from mattering in many cases, including, in your case, pots.
Nice analogy with movies, but essentially it’s a category error. Movies are media, not interfaces. You consume movies, but _use_ websites. A movie is immutable. A website is dynamic. As a matter of fact, even movies follow a very common structure, from narrative, to format specs and credits. Directors and actors fit their performance to these constraints. Movies are arguably way more standard than websites.
This is a very good point, however I'm on the other side of it (or at least across the boundary, perhaps not polar opposite).
Performance may be worth a lot today but I feel it will be less and less. I mean "we" don't like the "performance" of Windows (copilot everywhere, a setup process taking ages with dozens of offers you don't want), we don't like MacOS' performance (weird corners ;), inconsistent icons, icons disappearing behinds notches, no tiling)
I like Hackernews because it's so minimal, I just changed the bar to be gray instead of orange, otherwise it's perfect for my needs. Imagine some performer making this a beautifully crafted site, I'd go for any of the alternatives we see coming by every now and then.
Movies are perhaps different, although for me they are often about the lessons, did they change my view on things? That can often be condensed a lot more (for me that usually means drop a lot of the emotional finery, ie, I like TNG and Voyager more than Discovery because there is less crying and close-ups of crying people's eyes, ok, Discovery also has a lot less moral discussions).
Maybe I'm not normal, but to me my own UIs sound good, more efficient, more (useful-) information dense, so I need to spend less time navigating. It's why I use Nix and Gnome and (to a lesser degree) FireFox. It clicks more for me, but I can think of ways to improve them (yes I will soon try Niri). It's why people like chatting with their agents that are hooked into everything (Home Assistant to email to joblisting sites). Where's your beautiful UI in that workflow? Just give me a good API. Personal assistant/agents may be toys for nerds at the moment, but they're going to be big imho.
One argument against mine is perhaps that I also get used to tools and setups at some point, even though I don't consider them optimal at first, they become optimal. Perhaps because there is a deeper vision behind them.
All in all, perhaps we're both right. But people here seem to be very much on the company side (not surprising), but I don't care about your company, I care about information. That's why I have ad blockers, throw articles and long lists into LLMs and increase the contrast on your "beautiful" gray on gray text.
Buried in your prose is certainly a point shared with your average website visitor: they want the information, they don't want to be wowed with complex animations. But they also don't want no styling. There is a middle ground between looking like lynx and having some flair.
> Maybe I'm not normal
You definitely are not normal, if we define normal as "the vast majority of people". If web developers took your feedback seriously it would be detrimental to the experience of almost everybody. But I think that you knew that.
<devils-advocate-mode>Meh, more and more people will get information relevant to the decisions they have/want to make via their agents, not via your work-of-art-website. Deal with it.</devils-advocate-mode>
I think OP has envisioned a situation half way between what we now have and what will soon be the actual reality and that his idea will just be quickly skipped over.
I don't think website design will go away, but I do expect that people will soon be ordering products and booking holidays through AI chats instead of doing it themselves, which will require the kind of manifests he's talking about, but will skip the UI layer completely.
> I think on HN and in tech in general people seem to forget that "the first bite is with the eye", and that is why "normal people" never liked or used RSS
I think HN reflexively shoots down any idea or prediction with a bias to the incumbent.
Generally, a technological advancement will render some previous ways of working useless or outdated. People value convenience way more than a curated experience but I'm not disagreeing that brand differentiation would still exist.
A company that offers a meaningfully better experience in the long term will outcompete a company that focuses too much on aesthetics.
If they get generative UI right, where the UI provider can also give their own flavour and have some differentiation but also allow enough personalisation to afford the user better experience, it will happen.
Some bets don't work (like RSS) but some bets have worked - like the Amazon e-commerce model. A person in 1985 could have shut Amazon's idea down the same way you have.
Convenience is often curated experience. That is apart what was already there when humans emerged (which is of course still the biggest part), everything human experience was curated by previous humans. But of course even curated crafts get thrown and replace with other different experiences.
You're right, but are also ignoring that branding, appearance, etc., is simply not important to some people. They prefer function over form, which is where I think the author is coming from. They're wrong in thinking that most people share this opinion, and the idea of LLMs creating UIs seems awful to me, but as you can see from the comments here, this is appealing to some. It's niche, but this website is not exactly mainstream.
I partly share this opinion because most branding and UIs, products that are primarily marketed as a "lifestyle", etc., are obnoxious. Yes, appearance is a factor of anything we interact with, but when using technology my primary thought is if it solves a practical problem. Not if it's broadcasting an image, or even if it's enjoyable to use. The latter is important, but often companies prioritize it over functionality, which is backwards to me.
So starting with a mostly functional product, and giving me the choice of how to style it, is appealing to me. This is why I still use RSS, custom style sheets, the CLI and simple GUI wrappers, etc.
There is an audience for this type of product, but it's of the magnitude of a rounding error, so naturally most companies don't, and likely shouldn't, focus on this segment.
I totally agree that there is an often loud minority calling for this sort of thing: "I am an expert. I don't need styling or white space. I want every last square centimeter of space filled with 8pt font. I demand information density!" (aside: these are also the same people who say that JS-based UIs are slow and server-side HTML is faster, despite the fact that backend latency is 99.999% of the problem but that is another discussion...)
And yet, in my lived-experience at an unnamed Big Co when we did lots of UXR work in the on-call, monitoring, and incident management software/tooling, when it came to people being the primary on-caller handling a page for an incident when the company is losing millions for every minute of downtime that the 8pt font information dense UI they said they wanted actually led to increased stress, more mistakes, longer time-to-mitigation etc. Turns out that a carefully and deliberately designed UX and information architecture and - gasp - white space (that was all carefully and minutely tuned to specific CUJs over many rounds of research and prototyping) is really important.
Even if you have all the information available, just throwing stuff at the screen doesn't always help IME. Less is often more.
What the article misses is that generating a good UI is not easy, a good interface conveys so much more semantic information then just it's underlying API, and it does that without the user needing to concisely interpret the information.
And it's not just semantic information, presenting any kind of information in a way which enable the user to seamlessly interpret and use it is not an easy task.
AI, definitely lowered the bar for making some UI, but it doesn't help with the fundamentals challenges of making a UI, at least not more so then it helps with the fundamental challenges of any other job in our industry.
I've been thinking about this a lot about what are the benefits of UI, GUI, TUIs in the wake of AI.
I've come to the thought that we built interfaces honed to the way we visualize and work. We simplified many things down to just clicks. A really good UI captures intent and use very well. It doesn't require constant written language input when one mouse click is enough. Also it appeals to the aesthetic experience too in some ways to enhance the brand.
Design might be subjective but that's a good thing to act as a differentiator for companies.
Too many people assume pretty or nice looking is what design is. It's a very empty statement that disguises how important design works to make our lives easy.
And so many UIs stink, so auto-generating them might, on average, not be worse.
> a good interface conveys so much more semantic information then just it's underlying API
Yes, a good UI contains institutional memory on rendering that particular information in a bigger context than the pure data. I’m not sure how to convey that best.
In most cases, the deeper into the stack you look the more you'll find out why it sucks. You might be taking for granted how hard it is to implement some features.
Just because you can superficially design a better UI doesn't mean it would work as intended. It's amazing really how much can go on behind the scenes just for a seemingly trivial button that looks like shit.
1. Branding. Companies want to control their interfaces for all sorts of reasons. Branding is a big one. Clarity and comms are another.
2. LLMs in the hot path. LLMs are expensive, a hell of a lot more expensive than executing some Javascript locally. Hell, you'd still probably need to do that under this model anyway. We're likely to see LLM usage filter into the right places, use-cases with higher leverage, LLMs to create a UI that is shipped to all users over LLMs creating UI on the fly every time. Costs and time will dictate this just like they have dictated how every other technology is used.
I believe branding is still possible, though more of a capability and platform approach like Stripe or Shopify. Local LLMs will likely make it more feasible in the future as costs decrease.
The branding part might be less relevant to B2B products, but it's critical for B2C. And the clarity/communication aspect is acutely relevant to both. I don't know how you handle that with everyone generating their own interfaces.
Imagine policy compliance too? You need a cookie banner for legal reasons, how do you enforce that everyone's interfaces add the cookie banner? I hate cookie banners, but it's a clear example of where compliance dictates UI, and there are others (sales, insurance, contracts, etc).
As for costs decreasing, sure, and local LLMs improve things... but building the UI once will always win out on costs. Even with local LLMs we'll still cache UI creation, so then why not share that cache? Maybe it takes a bunch of prompting to get the exact accessibility stuff you need in the UI, so now you share prompts for generating the bit you need.... why not just share the actual output, why not just use the one the service provides?
I think there's a version of this focused on customisation that I can see happening, but otherwise all I see is a ton more code, a ton more liability, and products being on the whole worse.
Look I'm sure it'll happen in some places, it's just not going to be a major shift in the way this post describes.
If the table stakes for using my API is a local GPU to build the interface, my competitors who used their GPUs once to create the interface for their customers will win. If the API getting started guide involves going to a user-contributed list of prompts to put together a set of things I want in the interface, the competitor who doesn't have that step and provides a default interface will win.
Default interfaces being provided is not going to change, and the universal truth of defaults is that most people stick with them.
Power users modifying their interfaces has always been a thing and is easier with LLMs, but it's going to remain niche, as in, something that power users/hobbyists do, or companies might provide their own internal UI to some external API, but again that already happens extensively.
Amazon is the wrong analogy I think, because delivery is in some ways cheaper than every individual going to the store themselves, storing in warehouses is cheaper than storing in stores. In fact in some ways the Amazon analogy better fits the other way around. Not a perfect fit.
I appreciate this idea. I don’t think it fits our current mental model of the web (or mobile), which makes it thought provoking. If you squint, it’s like the optimistic Web 2.0 era of open APIs expecting a bunch of various UIs and mashups to spring up. The business model could be challenging with the client-centric focus though, unless the adaptive browser slips ads in, which is an unpleasant thought.
Big brands spend millions establishing a particular look, style, format. They don't want you to treat their sites as merely a set of APIs to scrape and customize based on your own style preferences. They want you to have a branded experience.
You’re absolutely right. But consider big brands make for a minor percentage of sites on the web. Also recall that all those big brands have standard profiles on social media and they share the very same layout as your local dog shelter. They have no problem with that.
It doesn't really matter what they want. Chat interfaces are doing this from the opposite direction, pulling the data down and explaining it to you, it's not a big leap for LLMs to turn their markdown responses into a slightly richer experience you can browse natively.
When the canned sentence structures of LLMs are frequent and unaltered throughout an article like this, I always wonder whether the thinking also has been done mainly by the machine.
Focusing on SaaS rather than B2C.
A clear advantage of getting users build their own UI is that processes emerge as a consequence.
Specific roles within a team don’t use every feature in an UI, and often compose a series of actions in a workflow.
Letting them build the UI to aggregate and automate leads to being able to extract business knowledge in the UI as well as the reasoning the user has with AI about what to build.
Put that in a SaaS for an office and the outcome is the true representation of work being done in that office, plus clear signals about edge cases (aka “the user is not using his custom built flow, why?)
Stability etc can be handled post-hoc: once a customized ui proves some benefits (via user adoption, or whatever you think efficiently measures productivity gains), it can be formalized by a human coder, who gets the full picture and has all the domain knowledge baked in, as long as you don’t capture UIs only but also the reasoning that built it.
Back to article: smart to think this in terms of browser, since that crosses the boundary between SaaS
It could work if it makes production and distribution of content easier and cheaper. All social media sites without exception have standard layout and usability. There, brands encode their aesthetics through media, and brands are much more alive in these channels than on their own websites, which often lags behind their own platform profiles. Company websites are expensive to build, maintain and update. Even for a design company, say Pentagram, it’s much better to follow their work on the standard architecture of Instagram than on their own handcrafted and “beautiful” website. The relevance of corporate websites as a means to retrieve essential information is decaying. Economic factors ultimately drive decisions. If something like this existed in a solid form, it would be hard to justify spending thousands of dollars on a website. As a matter of personal opinion, UI should never be a place to express creativity. Media is a much better substrate to express personality than through user interface affordances. Nowadays all my corporate clients develop websites on the expectation that they will grant them legitimacy, and they don’t actually expect anyone to actually use or read them. As a user, I actually do prefer when a supplier has an Instagram page because their sites, if they even have one, are 100% going to be awful to read and navigate, not to mention they’ll almost certainly be outdated. The greatest barrier to something like this is simply tradition. The general idea is perfectly defensible and logical. We should be reminded that standard websites are never going away, so this is not to be a replacement, but could open the doors for small businesses and non-profits to spread rich structured information in a cheap and sovereign manner. The argument that businesses are averse to being scrapped is only true for elitist corporations. Most businesses stand to gain tremendously from having their data highly accessible from anywhere. And it’s damn easy to convince them of the benefits. Even more so considering that, if they want, they also could have their handcrafted website, which by the way would be simply a thematic structure over the very same API. You could argue that this is inevitable long term. But regarding the OPs prescribed timeline of couple years, I think it’s just naive. For this to become mainstream it would take at least a decade, if not more. Just writing the specs and tools for this would take years, easily.
The author emphasizes accessibility and coherence as a benefit but another interesting one is composability which does not emerge naturally in the world of UI. Create a UI for a pair of websites like a command line for grep and wc. LLMs already provide that but under the natural language interaction primitive. UI could allow for branded experiences, ad delivery and whatnot in ways that natural language doesn't.
I do like this idea, and agree on the timelines of the world grappling with what to change and what to keep with these new capabilities.
Having a traditional web page with styles and assets AND the spec allow LLM's to be a bit more guided by the original site's design and intent. More of a remix or like Arc's boosts/skills feature.
There's also the reality the a lot of the things you'd want to be promptable (sorting, functionality, enrichment) couldn't be done on just the front end. You need some mix of UI and API logic to be promptable...
I love this. It sparks so many ideas about how I could take control of the data in the world around me. And of course, open source apps built in this way, through shared repositories of ui prompts.
Cool idea and line of thought, obviously rough and early but it gets you thinking. “Software as clay” is obviously where the industry is heading, and as you say we’re approaching this from multiple angles… applying it directly in the browser is certainly an intriguing idea.
Why’d you make the prototype a separate browser instead of implementing with a chrome extension? Something like greasemonkey but with an LLM generating the scripts on the fly..
I like it. I feel like this is a possible evolution of the browser.
Going further, AI internet browser could be an entirely new app to break from the legacy.
I feel this with coding agents, so often where it fetches web data and interprets it, html in that loop is only occasionally additive. Feels quite futuristic
A webui halicunated just for you? I think this must be by far the scariest and from my point of view worst idea you could ever come up with, no offense. Why would you ever want your web-experience to be non deterministic. A different ui every time you load? Even longer loading spinners then with mbs of react dependencies today? This is wrong on more levels then I could count
What is the browser built the company, the country, the international exchanges and the biosphere for you? I noticed there's a lot of redundancy in these things.
So for the sake of argument let’s establish this: a vibe coded browser presents a vibe coded UI based upon vibe coded backend. Surely this will work great!
I'm struggling to understand what's being described here.
If it's personalized clients, that's what we already had for most web services before the iPhone and app-ification of everything. It failed because making things compatible is a hard problem and a highly political/bureaucratic tarpit.
> most SaaS products still ship hand-crafted React apps, each building its own UI, its own accessibility layer, its own theme system, its own responsive breakpoints
Contrary to popular belief on HN, building these React apps are not "bullshit jobs" in the broader corporate world, nor going to be replaced by AI. They're the backbone of all ecommerce today and the ground floor for business operations because they keep us out of the tarpit. The implementation details are irrelevant here anyway. The actual problem was always how a business retains full control of its brand and UX.
"and honestly, ..."
"heres the thing ..."
[em dash every paragraph]
Smells like default Claude voice. I like the ideas, but if someone can't be bothered to proof read their own article, then I don't know why we should trust that any of it was human generated.
I estimate far above 90% of frontends do the same thing you could do with .jsp or .jsf 20 years ago and yet here we are still not having perfectly reusable frontend primitives and everyone doing custom development. We were closer to that with bootstrap than now with tailwind.
I am convinced neither client side nor backend side AI solutions will solve this.
Fully on topic: It would be naive to believe that serious web offerings would allow you to do this. Reality is moving in a different direction: Try applying custom css and js to reddit, for example: The website is a nightmarish matryoshka of shadow dom components and that‘s only the beginning of the flashification and silverlightification of the web.
For anyone interested in this, log on to claude.ai and ask it to teach you something using "your generative UI elements" and watch it work. I do think this is the future in some ways although this specific feature in claude will eat up your tokens so beware.
With respect I feel like the author is missing a whole bunch here about the point of a website.
It's not just content/info/data, it's a performance (in the creative sense).
Brands spend a lot of time honing their appearance - not just fonts and colours but the whole composition and visual pacing - their entire "say something without saying anything at all" aspect etc. Just walk through any place with physical shops and really look at how the stores have worked on their appearance and how they present themselves to customers. They're not just selling a product, they're selling a lifestyle/feeling/etc/etc. They're not just going to give that creative control away to some LLM.
Another way to think of it is instead of people watching a movie or play when they go to the cinema or theater, they're just given the script to read. Same information but the entire artistry of both the performers and the directors is totally absent, leaving it up to each reader to imagine the delivery of lines or the scene's setting etc.
I think on HN and in tech in general people seem to forget that "the first bite is with the eye", and that is why "normal people" never liked or used RSS. The desire to leave our mark and to create (and view!) visually appealing things seems to be pretty innate in humans - we've been doing it since cave paintings. I struggle to think of a world where we just hand that over to AIs and humans have zero creative control.
To add to this, the OP's vision benefits the user -- reducing a business's value to its actual raw value as a service rather than a brand. For me, it sounds great.
But the business's incentives are in the exact opposite direction. That opposite direction is the whole point of branding. They want their service to have a vibe, a personality, something you irrationally value beyond its raw value as a service.
> something you irrationally value beyond its raw value as a service
Sometimes that feeling is the value. Sure my plants don’t care if they live in a cheap plastic pot off Amazon or a nice pot from the overpriced gardening store selling at a 200% markup, but I care. Sitting in my balcony surrounded by cheap disposable clutter feels different than enjoying the outdoors amidst quality vibes.
That's not irrational at all!
Some things are commodities, some are not. The point is only that it's in the interest of commodity businesses to convince you they are not selling commodities. That sleight of hand doesn't prevent genuine quality and artistry from mattering in many cases, including, in your case, pots.
Nice analogy with movies, but essentially it’s a category error. Movies are media, not interfaces. You consume movies, but _use_ websites. A movie is immutable. A website is dynamic. As a matter of fact, even movies follow a very common structure, from narrative, to format specs and credits. Directors and actors fit their performance to these constraints. Movies are arguably way more standard than websites.
This is a very good point, however I'm on the other side of it (or at least across the boundary, perhaps not polar opposite).
Performance may be worth a lot today but I feel it will be less and less. I mean "we" don't like the "performance" of Windows (copilot everywhere, a setup process taking ages with dozens of offers you don't want), we don't like MacOS' performance (weird corners ;), inconsistent icons, icons disappearing behinds notches, no tiling)
I like Hackernews because it's so minimal, I just changed the bar to be gray instead of orange, otherwise it's perfect for my needs. Imagine some performer making this a beautifully crafted site, I'd go for any of the alternatives we see coming by every now and then.
Movies are perhaps different, although for me they are often about the lessons, did they change my view on things? That can often be condensed a lot more (for me that usually means drop a lot of the emotional finery, ie, I like TNG and Voyager more than Discovery because there is less crying and close-ups of crying people's eyes, ok, Discovery also has a lot less moral discussions).
Maybe I'm not normal, but to me my own UIs sound good, more efficient, more (useful-) information dense, so I need to spend less time navigating. It's why I use Nix and Gnome and (to a lesser degree) FireFox. It clicks more for me, but I can think of ways to improve them (yes I will soon try Niri). It's why people like chatting with their agents that are hooked into everything (Home Assistant to email to joblisting sites). Where's your beautiful UI in that workflow? Just give me a good API. Personal assistant/agents may be toys for nerds at the moment, but they're going to be big imho.
One argument against mine is perhaps that I also get used to tools and setups at some point, even though I don't consider them optimal at first, they become optimal. Perhaps because there is a deeper vision behind them.
All in all, perhaps we're both right. But people here seem to be very much on the company side (not surprising), but I don't care about your company, I care about information. That's why I have ad blockers, throw articles and long lists into LLMs and increase the contrast on your "beautiful" gray on gray text.
Buried in your prose is certainly a point shared with your average website visitor: they want the information, they don't want to be wowed with complex animations. But they also don't want no styling. There is a middle ground between looking like lynx and having some flair.
> Maybe I'm not normal
You definitely are not normal, if we define normal as "the vast majority of people". If web developers took your feedback seriously it would be detrimental to the experience of almost everybody. But I think that you knew that.
<devils-advocate-mode>Meh, more and more people will get information relevant to the decisions they have/want to make via their agents, not via your work-of-art-website. Deal with it.</devils-advocate-mode>
I think OP has envisioned a situation half way between what we now have and what will soon be the actual reality and that his idea will just be quickly skipped over.
I don't think website design will go away, but I do expect that people will soon be ordering products and booking holidays through AI chats instead of doing it themselves, which will require the kind of manifests he's talking about, but will skip the UI layer completely.
> I think on HN and in tech in general people seem to forget that "the first bite is with the eye", and that is why "normal people" never liked or used RSS
I think HN reflexively shoots down any idea or prediction with a bias to the incumbent.
Generally, a technological advancement will render some previous ways of working useless or outdated. People value convenience way more than a curated experience but I'm not disagreeing that brand differentiation would still exist.
A company that offers a meaningfully better experience in the long term will outcompete a company that focuses too much on aesthetics.
If they get generative UI right, where the UI provider can also give their own flavour and have some differentiation but also allow enough personalisation to afford the user better experience, it will happen.
Some bets don't work (like RSS) but some bets have worked - like the Amazon e-commerce model. A person in 1985 could have shut Amazon's idea down the same way you have.
Convenience is often curated experience. That is apart what was already there when humans emerged (which is of course still the biggest part), everything human experience was curated by previous humans. But of course even curated crafts get thrown and replace with other different experiences.
You're right, but are also ignoring that branding, appearance, etc., is simply not important to some people. They prefer function over form, which is where I think the author is coming from. They're wrong in thinking that most people share this opinion, and the idea of LLMs creating UIs seems awful to me, but as you can see from the comments here, this is appealing to some. It's niche, but this website is not exactly mainstream.
I partly share this opinion because most branding and UIs, products that are primarily marketed as a "lifestyle", etc., are obnoxious. Yes, appearance is a factor of anything we interact with, but when using technology my primary thought is if it solves a practical problem. Not if it's broadcasting an image, or even if it's enjoyable to use. The latter is important, but often companies prioritize it over functionality, which is backwards to me.
So starting with a mostly functional product, and giving me the choice of how to style it, is appealing to me. This is why I still use RSS, custom style sheets, the CLI and simple GUI wrappers, etc.
There is an audience for this type of product, but it's of the magnitude of a rounding error, so naturally most companies don't, and likely shouldn't, focus on this segment.
I totally agree that there is an often loud minority calling for this sort of thing: "I am an expert. I don't need styling or white space. I want every last square centimeter of space filled with 8pt font. I demand information density!" (aside: these are also the same people who say that JS-based UIs are slow and server-side HTML is faster, despite the fact that backend latency is 99.999% of the problem but that is another discussion...)
And yet, in my lived-experience at an unnamed Big Co when we did lots of UXR work in the on-call, monitoring, and incident management software/tooling, when it came to people being the primary on-caller handling a page for an incident when the company is losing millions for every minute of downtime that the 8pt font information dense UI they said they wanted actually led to increased stress, more mistakes, longer time-to-mitigation etc. Turns out that a carefully and deliberately designed UX and information architecture and - gasp - white space (that was all carefully and minutely tuned to specific CUJs over many rounds of research and prototyping) is really important.
Even if you have all the information available, just throwing stuff at the screen doesn't always help IME. Less is often more.
[dead]
What the article misses is that generating a good UI is not easy, a good interface conveys so much more semantic information then just it's underlying API, and it does that without the user needing to concisely interpret the information.
And it's not just semantic information, presenting any kind of information in a way which enable the user to seamlessly interpret and use it is not an easy task.
AI, definitely lowered the bar for making some UI, but it doesn't help with the fundamentals challenges of making a UI, at least not more so then it helps with the fundamental challenges of any other job in our industry.
I've been thinking about this a lot about what are the benefits of UI, GUI, TUIs in the wake of AI.
I've come to the thought that we built interfaces honed to the way we visualize and work. We simplified many things down to just clicks. A really good UI captures intent and use very well. It doesn't require constant written language input when one mouse click is enough. Also it appeals to the aesthetic experience too in some ways to enhance the brand.
Design might be subjective but that's a good thing to act as a differentiator for companies.
Too many people assume pretty or nice looking is what design is. It's a very empty statement that disguises how important design works to make our lives easy.
> generating a good UI is not easy
And so many UIs stink, so auto-generating them might, on average, not be worse.
> a good interface conveys so much more semantic information then just it's underlying API
Yes, a good UI contains institutional memory on rendering that particular information in a bigger context than the pure data. I’m not sure how to convey that best.
In most cases, the deeper into the stack you look the more you'll find out why it sucks. You might be taking for granted how hard it is to implement some features.
Just because you can superficially design a better UI doesn't mean it would work as intended. It's amazing really how much can go on behind the scenes just for a seemingly trivial button that looks like shit.
I doubt this will happen for a few reasons:
1. Branding. Companies want to control their interfaces for all sorts of reasons. Branding is a big one. Clarity and comms are another.
2. LLMs in the hot path. LLMs are expensive, a hell of a lot more expensive than executing some Javascript locally. Hell, you'd still probably need to do that under this model anyway. We're likely to see LLM usage filter into the right places, use-cases with higher leverage, LLMs to create a UI that is shipped to all users over LLMs creating UI on the fly every time. Costs and time will dictate this just like they have dictated how every other technology is used.
I believe branding is still possible, though more of a capability and platform approach like Stripe or Shopify. Local LLMs will likely make it more feasible in the future as costs decrease.
The branding part might be less relevant to B2B products, but it's critical for B2C. And the clarity/communication aspect is acutely relevant to both. I don't know how you handle that with everyone generating their own interfaces.
Imagine policy compliance too? You need a cookie banner for legal reasons, how do you enforce that everyone's interfaces add the cookie banner? I hate cookie banners, but it's a clear example of where compliance dictates UI, and there are others (sales, insurance, contracts, etc).
As for costs decreasing, sure, and local LLMs improve things... but building the UI once will always win out on costs. Even with local LLMs we'll still cache UI creation, so then why not share that cache? Maybe it takes a bunch of prompting to get the exact accessibility stuff you need in the UI, so now you share prompts for generating the bit you need.... why not just share the actual output, why not just use the one the service provides?
I think there's a version of this focused on customisation that I can see happening, but otherwise all I see is a ton more code, a ton more liability, and products being on the whole worse.
> LLMs in the hot path. LLMs are expensive, a hell of a lot more expensive than executing some Javascript locally
1. The costs are not that high
2. They are going down constantly
3. The ease of using LLM's and the personalisation will result in more interactions which means companies don't mind subsidising
This is like saying Amazon e-commerce can't work because logistics are expensive and its hard to pay for a delivery person to come to your doorstep.
Look I'm sure it'll happen in some places, it's just not going to be a major shift in the way this post describes.
If the table stakes for using my API is a local GPU to build the interface, my competitors who used their GPUs once to create the interface for their customers will win. If the API getting started guide involves going to a user-contributed list of prompts to put together a set of things I want in the interface, the competitor who doesn't have that step and provides a default interface will win.
Default interfaces being provided is not going to change, and the universal truth of defaults is that most people stick with them.
Power users modifying their interfaces has always been a thing and is easier with LLMs, but it's going to remain niche, as in, something that power users/hobbyists do, or companies might provide their own internal UI to some external API, but again that already happens extensively.
Amazon is the wrong analogy I think, because delivery is in some ways cheaper than every individual going to the store themselves, storing in warehouses is cheaper than storing in stores. In fact in some ways the Amazon analogy better fits the other way around. Not a perfect fit.
I appreciate this idea. I don’t think it fits our current mental model of the web (or mobile), which makes it thought provoking. If you squint, it’s like the optimistic Web 2.0 era of open APIs expecting a bunch of various UIs and mashups to spring up. The business model could be challenging with the client-centric focus though, unless the adaptive browser slips ads in, which is an unpleasant thought.
Big brands spend millions establishing a particular look, style, format. They don't want you to treat their sites as merely a set of APIs to scrape and customize based on your own style preferences. They want you to have a branded experience.
You’re absolutely right. But consider big brands make for a minor percentage of sites on the web. Also recall that all those big brands have standard profiles on social media and they share the very same layout as your local dog shelter. They have no problem with that.
Too bad, I control the client and not them.
That may not be for long. How far are we from requiring age check and ID for every curl request?
Have a proof of concept?
Why not both? Look at stripe or shopify for example
So?
It doesn't really matter what they want. Chat interfaces are doing this from the opposite direction, pulling the data down and explaining it to you, it's not a big leap for LLMs to turn their markdown responses into a slightly richer experience you can browse natively.
> So?
The point of the OP is that the companies would willing cooperate and replace their websites with LLM consumable APIs.
It's a different question whether this will happen despite their objections, as a kind of logical conclusion of the greasemonkey plugin.
When the canned sentence structures of LLMs are frequent and unaltered throughout an article like this, I always wonder whether the thinking also has been done mainly by the machine.
maybe one day our llm browser ui will remove those too ;)
Focusing on SaaS rather than B2C. A clear advantage of getting users build their own UI is that processes emerge as a consequence. Specific roles within a team don’t use every feature in an UI, and often compose a series of actions in a workflow. Letting them build the UI to aggregate and automate leads to being able to extract business knowledge in the UI as well as the reasoning the user has with AI about what to build.
Put that in a SaaS for an office and the outcome is the true representation of work being done in that office, plus clear signals about edge cases (aka “the user is not using his custom built flow, why?)
In a sense, related to https://danieldelaney.net/normal/
Stability etc can be handled post-hoc: once a customized ui proves some benefits (via user adoption, or whatever you think efficiently measures productivity gains), it can be formalized by a human coder, who gets the full picture and has all the domain knowledge baked in, as long as you don’t capture UIs only but also the reasoning that built it.
Back to article: smart to think this in terms of browser, since that crosses the boundary between SaaS
It could work if it makes production and distribution of content easier and cheaper. All social media sites without exception have standard layout and usability. There, brands encode their aesthetics through media, and brands are much more alive in these channels than on their own websites, which often lags behind their own platform profiles. Company websites are expensive to build, maintain and update. Even for a design company, say Pentagram, it’s much better to follow their work on the standard architecture of Instagram than on their own handcrafted and “beautiful” website. The relevance of corporate websites as a means to retrieve essential information is decaying. Economic factors ultimately drive decisions. If something like this existed in a solid form, it would be hard to justify spending thousands of dollars on a website. As a matter of personal opinion, UI should never be a place to express creativity. Media is a much better substrate to express personality than through user interface affordances. Nowadays all my corporate clients develop websites on the expectation that they will grant them legitimacy, and they don’t actually expect anyone to actually use or read them. As a user, I actually do prefer when a supplier has an Instagram page because their sites, if they even have one, are 100% going to be awful to read and navigate, not to mention they’ll almost certainly be outdated. The greatest barrier to something like this is simply tradition. The general idea is perfectly defensible and logical. We should be reminded that standard websites are never going away, so this is not to be a replacement, but could open the doors for small businesses and non-profits to spread rich structured information in a cheap and sovereign manner. The argument that businesses are averse to being scrapped is only true for elitist corporations. Most businesses stand to gain tremendously from having their data highly accessible from anywhere. And it’s damn easy to convince them of the benefits. Even more so considering that, if they want, they also could have their handcrafted website, which by the way would be simply a thematic structure over the very same API. You could argue that this is inevitable long term. But regarding the OPs prescribed timeline of couple years, I think it’s just naive. For this to become mainstream it would take at least a decade, if not more. Just writing the specs and tools for this would take years, easily.
The author emphasizes accessibility and coherence as a benefit but another interesting one is composability which does not emerge naturally in the world of UI. Create a UI for a pair of websites like a command line for grep and wc. LLMs already provide that but under the natural language interaction primitive. UI could allow for branded experiences, ad delivery and whatnot in ways that natural language doesn't.
I do like this idea, and agree on the timelines of the world grappling with what to change and what to keep with these new capabilities.
Having a traditional web page with styles and assets AND the spec allow LLM's to be a bit more guided by the original site's design and intent. More of a remix or like Arc's boosts/skills feature.
There's also the reality the a lot of the things you'd want to be promptable (sorting, functionality, enrichment) couldn't be done on just the front end. You need some mix of UI and API logic to be promptable...
I love this. It sparks so many ideas about how I could take control of the data in the world around me. And of course, open source apps built in this way, through shared repositories of ui prompts.
Cool idea and line of thought, obviously rough and early but it gets you thinking. “Software as clay” is obviously where the industry is heading, and as you say we’re approaching this from multiple angles… applying it directly in the browser is certainly an intriguing idea.
Why’d you make the prototype a separate browser instead of implementing with a chrome extension? Something like greasemonkey but with an LLM generating the scripts on the fly..
I like it. I feel like this is a possible evolution of the browser.
Going further, AI internet browser could be an entirely new app to break from the legacy.
I feel this with coding agents, so often where it fetches web data and interprets it, html in that loop is only occasionally additive. Feels quite futuristic
A webui halicunated just for you? I think this must be by far the scariest and from my point of view worst idea you could ever come up with, no offense. Why would you ever want your web-experience to be non deterministic. A different ui every time you load? Even longer loading spinners then with mbs of react dependencies today? This is wrong on more levels then I could count
What is the browser built the company, the country, the international exchanges and the biosphere for you? I noticed there's a lot of redundancy in these things.
So for the sake of argument let’s establish this: a vibe coded browser presents a vibe coded UI based upon vibe coded backend. Surely this will work great!
lol imagine the support burden (docs and help) once everyone generates their own ui and shit breaks. hard enough already.
I'm struggling to understand what's being described here.
If it's personalized clients, that's what we already had for most web services before the iPhone and app-ification of everything. It failed because making things compatible is a hard problem and a highly political/bureaucratic tarpit.
> most SaaS products still ship hand-crafted React apps, each building its own UI, its own accessibility layer, its own theme system, its own responsive breakpoints
Contrary to popular belief on HN, building these React apps are not "bullshit jobs" in the broader corporate world, nor going to be replaced by AI. They're the backbone of all ecommerce today and the ground floor for business operations because they keep us out of the tarpit. The implementation details are irrelevant here anyway. The actual problem was always how a business retains full control of its brand and UX.
Self-describing API endpoints... is the server side for this basically just HATEOAS?
Or SOAP with WSDL.
"and honestly, ..." "heres the thing ..." [em dash every paragraph]
Smells like default Claude voice. I like the ideas, but if someone can't be bothered to proof read their own article, then I don't know why we should trust that any of it was human generated.
Reinventing what REST should’ve been.
I estimate far above 90% of frontends do the same thing you could do with .jsp or .jsf 20 years ago and yet here we are still not having perfectly reusable frontend primitives and everyone doing custom development. We were closer to that with bootstrap than now with tailwind.
I am convinced neither client side nor backend side AI solutions will solve this.
Fully on topic: It would be naive to believe that serious web offerings would allow you to do this. Reality is moving in a different direction: Try applying custom css and js to reddit, for example: The website is a nightmarish matryoshka of shadow dom components and that‘s only the beginning of the flashification and silverlightification of the web.
For anyone interested in this, log on to claude.ai and ask it to teach you something using "your generative UI elements" and watch it work. I do think this is the future in some ways although this specific feature in claude will eat up your tokens so beware.
So...XForms[0] but involving AI.
[0] https://www.w3.org/TR/xforms/
Why limit to a browser? Why not the whole system. Check out this horrible thing I'm building. https://abject.world
This is incredible.
This is actually really innovative! I like it a lot
This is ridiculous and I’m into it.
[dead]