"Great design prompts require design vocabulary. Most people don't have it."
Vocabulary is just the surface. Beneath it is an understanding of how to achieve your goals with design. How to make things that are easy to use, accessible, that create a certain impression.
Does this website (presumably made with the help of these AI tools) show this kind of understanding of design? Not really. It's chaotic, the text is often hard to read and there is a ton of fluff, both in terms of visuals and copy.
There is a "Frequently Asked Questions" section and a "Popular" $100 tier in the "Support the Project" section, even though this project seems to be brand new. Why lie to the reader?
Yes, but with LLMs, sometimes simply mentioning the right words is enough to prime the model in the direction you want to take it. If you start a prompt talking about leading and type pairings, it will take greater care with typography. You don't need to be an expert typographer to take advantage of this phenomenon.
How will an LLM "take greater care with typography" if it can't see the page it is creating? How will it "improve" leading if you need a human to see that there's too much distance between lines or too little?
Because humans have already annotated diagrams and examples of what ‘too much’ and ‘too little’ look like, and these have been incorporated into the model. It tries to reproduce the content that is associated with humans indicating that they are taking greater care, and that content has the ‘not too much / not too little’ judgement already baked into it.
Or, said another way, vocabulary provides us with the words (semantics). You also need a grammar (syntax), which itself needs to be ordered toward an end (pragmatics).
This isn't really what I said, and is considerably less clear. What I meant is that you can't boil design down to certain stylistic flourishes and words denoting them (e.g. "vertical rhythm", to take an example from the linked page). Whatever you're doing, it involves understanding how the viewer will react to what you will show them and why.
I had to go back and check, with "modern invisible scrollbars", and those useless theme settings at the bottom I assumed the page was just some css demo that ended there and left.
That landing page example is devastatingly bad. You start with a page that has usage numbers, uptime, support 24/7 and a customer rating above the fold. You end up with a page that lacks all of these advantage and instead looks bland and has horrible typography and even less text contrast.
In line with that, the Dashboard looks more organized in the "after" picture, but that's because it lost most of its useful information.
Concept seems fun, and I'm expecting we'll see a bunch of those in the next few weeks/months. UX of that specific page seems broken, however, as the container for the explanation of each "function" doesn't scroll along with the rest of the content (stays stuck at the top) and makes it impossible to see.
I can confirm the broken UI. The demo container disappears as you scroll down, leaving a blank space that takes up most of the screen. I want to make a snarky joke about this but I'm just tired at this point.
The "Before" examples look infinitely better than the "After" examples. Tells you all you need to know. Wouldn't be at all surprised if this whole thing was a ridiculous joke or a satirical commentary on pretentious design.
Define “better”. Some people actually prefer to wake up and make toast and eggs and coffee in their own kitchen, instead of just buying affordable, professionally barista-assembled grab-and-go from the tone-balanced local Starbucks on the way to work. It’s a deviant preference, really.
agreed, the information density on the "after" example is much worse for most dashboard use cases. Way too much space, not enough info. But I guess I'm not exactly surprised based on the style of the page being both zoomed in and spaced out
It's interesting to see people creating and 'selling' agent skills. This one asks for donations, but I was expecting to see a stripe link and 'download for 4 dollars, yours forever' (personally I think that would convert better...)
I wonder if there will be full-blown skill marketplaces soon. Would that be a way for some experts to recoup some (presumably very small portion) of the income they might lose due to generative AI market effects?
Come one, there are things to say about this project but the Download section is pretty clear. It installs commands for your LLMs and AI-based IDE.
It states that clearly in the section.
If you’re not familiar with what a /command is in the context of LLM, this may just not be for you and that's fine, but the purpose is clearly stated.
To get something usable out of an LLM (aka vibecooding, vibe engineering et al), it works best if you're an expert yourself -> a.k.a you need to know the "lingo".
So there's the possibility of skipping the intermediate work in between by exposing yourself to just the input and the output of the process for certain domains, this is for frontend I think.
The Form UX one is hilarious. It took a streamlined form used to convert and added enormous marketing copy that's more attention grabbing than the form itself. If you look closely they ran the `/simplify` command, haha.
This is the most egregious one in my eyes, too. I've run A/B tests on a few signup forms and without fail it validates the standard practice: the lowest drop-off rate comes from removing every possible obstacle and distraction. I'd bet a few dollars (which is as much as I'll ever bet) that design update would perform worse. The tool is almost intriguing as a _reductio_ of certain design practices.
The "after" designs all replace the rather generic "SV startup with a tailwind UI" with this serif font, parchment color look. It looks very similar to Anthropic's branding. I guess it looks marginally more distinctive? Though it seems to replace one knock-off visual identity for another. But the claim is that the tool here is implementing best practices through a sophisticated "design vocabulary", and in that sense the examples strike me as manifest failures. I find the general legibility of the "before" designs to be much better.
I'm glad it's not just me. One would hope that `BEFORE` and `AFTER` would imply `WORSE` and `BETTER`, but from their examples they somehow they managed to shoehorn `MEH` in there.
I think the difficulty for AI to learn this, in general, is the missing out of the day-to-day experience living as a human, because that is what shapes our viewing habits. And those are what a good graphic design interacts with.
They set themselves up for a fall when they named themselves "Impeccable Style"
The mix of sans and serif fonts on their website is a mess. There's too much negative space, and it's inconsistent. Too many font sizes, and some that are so tiny they're illegible.
In the landing page before/after example, I think the "before" design looks more appealing.
Vocabulary is just the surface. Beneath it is an understanding of how to achieve your goals with design. How to make things that are easy to use, accessible, that create a certain impression.
Does this website (presumably made with the help of these AI tools) show this kind of understanding of design? Not really. It's chaotic, the text is often hard to read and there is a ton of fluff, both in terms of visuals and copy.
There is a "Frequently Asked Questions" section and a "Popular" $100 tier in the "Support the Project" section, even though this project seems to be brand new. Why lie to the reader?
Roundabout what I would expect as a result from the prompt "make a website that demonstrates how LLMs can better designs"
Yes, but with LLMs, sometimes simply mentioning the right words is enough to prime the model in the direction you want to take it. If you start a prompt talking about leading and type pairings, it will take greater care with typography. You don't need to be an expert typographer to take advantage of this phenomenon.
That landing page example is devastatingly bad. You start with a page that has usage numbers, uptime, support 24/7 and a customer rating above the fold. You end up with a page that lacks all of these advantage and instead looks bland and has horrible typography and even less text contrast.
In line with that, the Dashboard looks more organized in the "after" picture, but that's because it lost most of its useful information.
It's interesting to see people creating and 'selling' agent skills. This one asks for donations, but I was expecting to see a stripe link and 'download for 4 dollars, yours forever' (personally I think that would convert better...)
I wonder if there will be full-blown skill marketplaces soon. Would that be a way for some experts to recoup some (presumably very small portion) of the income they might lose due to generative AI market effects?
If you’re not familiar with what a /command is in the context of LLM, this may just not be for you and that's fine, but the purpose is clearly stated.
So there's the possibility of skipping the intermediate work in between by exposing yourself to just the input and the output of the process for certain domains, this is for frontend I think.
Renaissance Geek (noun)
A person who moves fluidly between art, technology, narrative, and systems — guided by curiosity instead of specialization.
With AI as their amplifier, this breadth makes them dangerous enough to build the future rather than be shaped by it.
Of user interface style: low contrast and hence poor readability, with excessive white space.
The dashboard might even be funnier, though.
And this is what the creator chose to demo.
The "after" designs all replace the rather generic "SV startup with a tailwind UI" with this serif font, parchment color look. It looks very similar to Anthropic's branding. I guess it looks marginally more distinctive? Though it seems to replace one knock-off visual identity for another. But the claim is that the tool here is implementing best practices through a sophisticated "design vocabulary", and in that sense the examples strike me as manifest failures. I find the general legibility of the "before" designs to be much better.
And if they need to explain it... ;-)
Tufte it isn't.
I think the difficulty for AI to learn this, in general, is the missing out of the day-to-day experience living as a human, because that is what shapes our viewing habits. And those are what a good graphic design interacts with.
The mix of sans and serif fonts on their website is a mess. There's too much negative space, and it's inconsistent. Too many font sizes, and some that are so tiny they're illegible.
In the landing page before/after example, I think the "before" design looks more appealing.