Common Mistakes With Vibe Coded Websites

It's easier than ever to design with AI tools, but that doesn't mean you should always accept all changes. So today,

I'll be joined by Raphael Shod to look at the good, the bad, and the ugly of AI assisted design and sort through how to use these superpowers to stand out.

Welcome to another design review.

All right. Today I am thrilled to be joined by Raphael Shad, one of the newest visiting partners here at YC. And

Raphael was the co-founder of Kron, a modern calendar tool, which you sold to Notion a while back and are also one of the top designers that I know and have worked with. So, uh, Rafael, it's an

worked with. So, uh, Rafael, it's an honor to have you here and it's an honor to work together now at YC. Yeah, super

fun to be now back in uh in in YC uh and helping founders, next generation of founders build iconic companies.

>> Well, um you're the perfect person to have on here because uh the topic that we're going to be talking about today is AI design trends. On the negative end, there's the AI design slop and on the

positive end, there's a lot of AI uh design trends that that now are easier to build for anybody even if they're not a designer. So, what are some of the

a designer. So, what are some of the common things that you've seen recently?

a lot of the AI top AI design trends that are standing out that that you'd want to call out for the audience before we jump into some sites.

>> Yeah, sure. Um, this all kind of started when I had like a late night thought and tweeted that I see a lot of dumb hover effects on landing pages of startups these days presumably white coded. And

so I was kind of like curious to peel the the the layer back there. It's like

how did these like what I thought were dumb effects um how did they make it into LLMs and why are they everywhere?

Now, a couple other trends that we then identified was kind of like purple gradients and all of a sudden all startup websites um had purple gradients everywhere or these sections that can like fade as you go in as you scroll and

they fade in and fade out.

>> Yeah. And it's not so much that those inherently are bad. It's not purple gradients are bad. Nobody should ever use them. It's that now they're just so

use them. It's that now they're just so ubiquitous everywhere that they kind of uh lose all meaning and specialness and originality.

>> Totally. And one of the key things was when there was a good website kind of establishing a trend, it took a while um in the old world uh for others to kind of like copy these trends. But now with

LLMs, if there's a good website with a purple gradient, it makes it into the LLM because the LLM gets trained on like the good examples that get linked to a lot. Um and then all of a sudden, like

lot. Um and then all of a sudden, like the next week, all the startup websites look the same.

>> Yes. And so this is both good and bad if this power is used in the right ways.

So, we've got a bunch of YC companies that have submitted their websites that were all built with AI coding tools. So,

this should give us good grounds to figure out uh ways that they have been used really, really well and things that we would do differently if we were designing it ourselves. So, let's uh let's dive into some of those.

>> Let's have a look.

>> All right. Okay. First up, we've got new.ai.

new.ai.

Solve your game testing with AI agents.

Save countless hours of manual game QA by using AI agents performing end-to-end tests at scale.

>> Purple gradients.

>> The first thing that jumps out. It's a

very purple site. Um, which, you know, we were just talking about that's one of the most common things that you see. So,

it's interesting like prelooked at this and gone like, "Wow, this looks really nice." And now I look at it and

really nice." And now I look at it and I'm like, "They used AI to design this."

Maybe they didn't, but like that's what it screams to me. Okay. So, one thing that's interesting is we as you start scrolling here, we've got this line that's following us up and down. Um, and

it's interesting because I actually find this this line to be a little distracting. Um, I'm paying more

distracting. Um, I'm paying more attention to the line than I am the the things that I think that it wants me to pay attention to.

>> It's it's almost like something that it's kind of like probably hard to implement for a human. And you would have never thought to actually code this up um before AI. Um, and just because we

now can, just because LM are kind of good at these type of like SVG, you know, buildups or transforms, doesn't mean that it's actually a good design and helps you convert potential, you know, visitors into into customers,

basically.

>> Yeah. Yeah. I don't know anybody that's like, we need a line that goes through here. Let's start coding that from

here. Let's start coding that from scratch. Like, you would only implement

scratch. Like, you would only implement something like this because it was so easy to do.

>> Um, but I don't think it actually adds value. If anything, I think it distracts

value. If anything, I think it distracts from the rest of the page. Um,

I I can't recall anything from these sections based on this. Um, okay. So,

now we're down to everything in one place. You I find this hard to read.

place. You I find this hard to read.

It's uh it's it's very light. Um, maybe

I'm just old. Um,

>> yep. Contrast could be improved there a little bit. So, these cards are

little bit. So, these cards are interesting, right? Where

interesting, right? Where >> this is cool.

>> Um, this is actually pretty good. Uh

this is uh probably something that would have been maybe too expensive to do or only like the very best or like website designers would have kind of got gotten to the lengths of of doing something like that these hover animations for the

cards. But AI is so simple to do and

cards. But AI is so simple to do and this is actually like a pretty tasteful um way of using leveraging so like the powers of LLM uh to to build more compelling websites.

>> Yeah. What's cool about it is it feels like it helps establish the brand and it's unique and it's creative and it's fun. Um, and it it's actually trying to,

fun. Um, and it it's actually trying to, you know, multiplayer like two controllers with lightning connecting them. Um, and it actually helps

them. Um, and it actually helps reinforce the points that I think that they're trying to make here. Um, and

this is a great example of, you know, where the line that was following you down the page didn't add much value.

This is this would have been really hard to do. And now, um, you kind of get it

to do. And now, um, you kind of get it for free.

>> Um, you don't have to go from scratch building a lot of this stuff.

>> One of the things I want to try out is, um, up here. Um, so we here have the main navigation. One of the things that

main navigation. One of the things that jumps out here is that as you hover, you generally kind of want to make things pop, but here it actually kind of disappears. So instead of inviting to

disappears. So instead of inviting to click to talk to them, um, it kind of like goes into the background. Same

here. it kind of fades out as you activate one of these menu points. Um,

this is something that I think a designer would almost never choose to do. And now uh and the default of

do. And now uh and the default of browsers also doesn't work that way. And

now lens just somehow decided that that's a good uh hover kind of like effect. Um, but it makes no sense. Um,

effect. Um, but it makes no sense. Um,

think about it that way. Like the

browser already has like a built-in hover effect for free even without CSS.

It turns the cursor into a hand, right?

So that already indicates like I can click here. So just leaving it white in

click here. So just leaving it white in this case or just kind of like the the bright purple and then having the hand already tells me like I can click on this thing. So if you want to add a CSS

this thing. So if you want to add a CSS effect then you know make it maybe like a little pop. Maybe make it like one shade lighter um or more active or add like a little you know a little uh um

glow around it but don't make it go away.

>> Yeah. Do you have any rules of thumb when you know you're designing something and you're trying to come up with hover effects? Like what are your goals when

effects? Like what are your goals when you were trying to come up with the right hover?

>> Yeah, make it sort of like clickable. Um

Steve Jobs famously said you know he wanted to make things lickable with the Aqua interface back in the day. I think

you know make things you know clickable kind of like inviting to to click stick to sort of like pretty standard again kind of I think the hand indicates that something is clickable and so I wouldn't go much beyond that. Another thing that

I see hover being used um is a lot to kind of like reveal additional information. So, not just kind of like a

information. So, not just kind of like a hover effect to make the UI more feel alive. Like hover is great to kind of

alive. Like hover is great to kind of make UIs feel more alive, right? Um, but

I see it increasingly used to kind of like uh disclose additional critical information or functionality. And I

don't think that's the best use for hover. This whole trend uh that maybe

hover. This whole trend uh that maybe like kind of like Mac OS kind of started to over the years get more and more into kind of like very clean interfaces, get the functionality and the tools and the buttons all out of the way so there's

all this space for content. I don't want to have my computer to be just like all content. I actually want it to be, you

content. I actually want it to be, you know, a bicycle for the mind, the tool.

Uh, and so, um, revealing additional functionality only on hover and you have to go hunt for what the tool can actually do, I think is kind of like an anti-attern.

>> Yeah. Not to mention that, you know, there isn't a concept of hover on mobile as well. So,

as well. So, >> and long press, which is kind of the equivalent of of hover, um, uh, never really catch caught on on mobile >> probably because it's so hard to discover. Yeah. You you don't want to

discover. Yeah. You you don't want to sit there long pressing all the time to try to figure out if there is any additional >> think of a desktop desktop interface a tool where you have to hover around everywhere to discover where this stuff

actually lives. You know the disclosure

actually lives. You know the disclosure arrows or like hiding something or getting to a dot dot dot kind of menu to get to extra functionality. If that's

all hidden behind hover then, you know, you're just not as efficiently communicating the functionality of your software.

>> Yeah. Overall, I think this is um a very nicely designed site. I think it's easier than ever using AI design tools and LMS to get a professional looking site. There's almost no excuse for not

site. There's almost no excuse for not having it now. Yeah. And especially as we're reviewing YC applications, if you look at a demo and that demo uh does not have a base level of quality design, it

seems like the person just didn't even try >> because it's so easy to do it now. So,

Noo-Noo, um really cool site and uh some some really interesting uh applications of using AI here. So, well done. All

right, next up we got Rosebud AI. Create

games with AI. And and once again, like you start to see the trend here. We've

got the purple gradient again. We've got

a very similar um kind of pinkish purple accent color up here. Um and that's why like in isolation with one of these, like it looks cool, it looks modern, everything, but you know, one after the

other, they all start to kind of look the same, and it takes away a lot of the core brand and originality and things like that that I think most founders want for their site.

>> So, let's dig in here. This is the first time we're looking at these live.

>> Yeah.

>> Um, so rose by AI, create games of AI.

And there's a prompt to to to prompt to game kind of interaction, I would assume.

>> Um, >> that's cool. There's something you can instantly play.

>> Yeah, >> let's try it.

>> Whoa. It's like a 3D game in the browser. That's cool.

browser. That's cool.

>> There's person walking around.

>> Can I use the arrows?

>> There we go. Oh, wd. We have a old school game here.

>> You can jump.

>> All right, let's uh escape out of this.

Um, but cool. Definitely kind of like engaging. Um, I wonder whether this is

engaging. Um, I wonder whether this is actually their like product code running on the website here or this is like bitecoded >> sandbox environment. Um,

>> yeah, I would assume that this is the type of thing you can create with their product. But um that's an interesting

product. But um that's an interesting point is that they don't really tell us, >> you know, it doesn't say like play a game made with our product right now. Um

it's interesting. Is this following us?

This top bar.

>> Yeah. These are just anchors jumping vertically around.

>> Yeah. So what's interesting is I didn't realize this at the time, but when we went to that one we were just playing that was under modify games. And so what I wonder is if this is a game that has

already been created that we can modify in real time and that's what they're trying to show off rather than the creation part.

>> Yeah. So maybe kind of like a thing to take away here is that using this non kind of like non-standard navigation um is tripping us off a little bit um on how to use this website. Another thing

that kind of jumped out here at the top is like I don't think anyone would think of combining a red logo with purple as an accent color necessarily. Um it's not

necessarily complimentary. Uh and uh and

necessarily complimentary. Uh and uh and I guess whenever I see the use of emojis, even though they're not the system emojis, I feel like it's a little lazy. And so I feel like LLMs kind of

lazy. And so I feel like LLMs kind of take the easy path because they don't have any IP really themselves. They're

just like big models kind of cobbling things together. They use like these

things together. They use like these standard icons everywhere and it's just immediately kind of like a tell.

>> Yeah, I think we could just use more context up here too. And and you know the the name of the company is is here.

It looks like it's almost part of the headline rather than I would expect it to be up here by the logo.

>> Yeah. And then the value prop kind of being more clear. Um you know uh uh create games with AI something something. So it's kind of like what's

something. So it's kind of like what's important for the H1 is typically kind of you know what is it? Who is it for?

and to what end? Why should that person that it's for care? Um, and if those three things plus a call to action um are not kind of above the fault, then it's a harder time to convert uh convert

visitors.

>> Yeah. Um okay, we've got some I guess examples of games. Yep. So games created with Rosebud AI. Um okay, this is interest. So there's another effect

interest. So there's another effect here. You can see the light following my

here. You can see the light following my cursor um around the edges. And yeah,

this is the kind of thing where, you know, you're like, "Hey, it costs nothing to do this." Then maybe you're like, "Yeah, sure. Let's throw it in.

It's fun and it's playful." Um, but if you had to spend a week coding this and, you know, getting it to work, right?

You'd be like, "That's not worth it."

Okay. So, we've got a lot of a lot of game examples. And it looks like maybe

game examples. And it looks like maybe this is just the rest of the site here.

Okay. And then we hit a footer. There's

a >> like remixing.

>> Oh, yeah. Okay.

>> Some some of the some of the um projects you can remix and then it kind of puts up the conversion wall. Okay. Um, that

would go beyond like landing page.

>> Very cool. Awesome. Well done, Rosebud.

Okay. So, next up we've got Get Crux.

And one of the first things that I noticed here was kind of the fade in. It

was this automatic fade in. Um. Oh, and

another thing that I noticed is my scroll is hijacked here.

>> Yeah, it's maybe a little hard to um to see on screen. Uh, but there's certainly some scroll jacking logic going on. Um,

and there's also like a state where you see both the detached and the fixed uh header menu all like both at the same time like right right around here. Yeah.

Um, so typically kind of like they should just smoothly I like the sticky header. Um, but it's just like smoothly

header. Um, but it's just like smoothly transition one to another and here you see sort of like that you know they're clearly two distinct menus that just happen to kind of be the same.

>> I'm sorry I'm so distracted by the button that's chasing me around the screen here. Um,

screen here. Um, yeah. There's I almost feel like it's

yeah. There's I almost feel like it's hard to get the button to click on it because it's constantly moving.

>> Does it make you want to talk to the founders more or less?

>> It makes me distracted. I would say I'm not paying attention to what they do.

I'm just uh I'm like, why is this button following me around? Um, it definitely got my attention, I would say. Um, one

of the other things that I noticed first too is is almost these like meteors that are, you know, shooting down from the angles. cuz you can kind of see him over

angles. cuz you can kind of see him over there.

>> Um, that's another kind of thing where like it doesn't feel like it actually adds a lot of value to the product. And

again, if you had to build that from scratch, like it would never be worth your time. Um, but you it's so easy to

your time. Um, but you it's so easy to do it that people are like, "Yeah, let's throw that in. It looks cool." Um, but I actually find it a little distracting.

Um, I don't think it adds value to the the product here.

>> Then the hero screenshot here is um I guess a poster image for a video. Um

uh and it goes into what looks like a completely different video. Yeah, that

was confusing.

>> And the poster image is just very blurry here. Um I don't know whether this is

here. Um I don't know whether this is like an artifact of like uh using using AI or not. Um uh but just as a general design principle, I think all all your assets should be, you know, high-res,

high quality. Um and this just looks

high quality. Um and this just looks like something very blur blurry.

>> Yeah. And again, back on the button, too, because I still see even when we're down here, it's like still chasing us around. You know, this is another

around. You know, this is another example of the type of thing that you would only do because it's easy. But

just because something is easy doesn't mean it's worth doing. There's a lot of things that are now at your fingertips with AI that, you know, anything imaginable is possible. Just because

something is possible doesn't mean you should say yes to it. Here we're kind of like hitting a little bit on on just a lack of visual uh consistency. Like this

looks like an entirely different visual language than anything we've sort of seen seen above. Um,

>> where do you think that comes from?

>> Yeah, I wonder. Um, maybe different sections of the website being made by like different parts of the tool. Um,

maybe it's also just kind of like a function of like this here looking very different um than like the website content. Yeah. Sometimes just like a

content. Yeah. Sometimes just like a clear image with a little bit of text, maybe a headline and then some subtext um goes a long way.

>> Yeah.

And I gotta say, we're like halfway down the page right now. And um if you asked me what they did, I would not be able to tell you. Let's go back up to the top

tell you. Let's go back up to the top and re remind ourselves. AI creative

strategist to make ads that work. Ads

that. Okay. So, it's something to make ads. It seems like

ads. It seems like >> it's interesting. They they do kind of have like a visual um like style to it with a lot of these uh it's almost like a a vertical motion blur or something on

a color gradient that we see in some of these images. um like these right here

these images. um like these right here and we saw the same thing in these here and these here um and these here. Um but

it's it's constrained to some of these elements and it feels like you know it's maybe a little overused here because it's hard to read these and this feels like more like a rainbow rather than calling the attention to it. But up here

does not have that style at all, which was maybe, you know, the point that you were making earlier. And I could see something where the background here is more reflective of this type of style.

And then that might carry it through all the way through the rest of the page.

>> Yeah, let me quickly reload the page and try something. So, we see all these

try something. So, we see all these sections kind of like fading in. Um, and

here again, it's just like fading in.

And one of the things that I noticed, I just don't see the advantage of having these things not just right there as I scroll because scrolling already like is the motion. Why does it have to like

the motion. Why does it have to like also then can like fade in? Um, and one of the things that you get is these weird um, for example here, right? We

say we're here to answer all your questions and then there's one question and we we we break here earlier and I was like, wow, that's a lame FAQ if there's only like one Q and one A,

right? Um, but then I was like, "Aha, it

right? Um, but then I was like, "Aha, it just hasn't come in yet." Um, and so not a fan of that pattern.

>> Yeah. What bothers me with the fadeins a lot of time, and I'm not sure if this does it or not, is when they're on a a timer, sometimes I'm scrolling really quickly and I'll get all the way past

the section as it's starting to fade in up at the top and I miss the content altogether. So, I'm curious to try that

altogether. So, I'm curious to try that here. Um, and see how they did this. So

here. Um, and see how they did this. So

like if I'm just kind of zooming down um so these seem to fade in um at least when it's at the bottom of the page um >> but it still feels like going through

molasses because it's all scroll checked like hijacking the sort of like actual native browser scrolling um to do some fancy thing with JavaScript to actually have the hooks to do all these animations.

>> Yeah.

>> All right.

>> Super interesting. Very creative. Uh so

well done get crux. Okay. Okay, next we got Sphinx, your last compliance hire.

All right, we've got some animation thing going on here. Um, I think one of the trends that we've seen so far is just like way more animation these days than there were, you know, three years

ago probably because these AI design tools make it so easy to do this.

>> Yeah, let's quickly talk about hierarchy here. Um, so we start with um a logo

here. Um, so we start with um a logo type, kind of a stylized version of the name of the company, Sphinx, right?

That's cool. Top left, pretty standard.

But then we have the H1 is also good and the subtext. But also we have like this

the subtext. But also we have like this kind of like weird in between like fourth style, right? Um not something that I typically ever saw designers or

engineers kind of come up with their own landing pages. is only something kind of

landing pages. is only something kind of like these many many styles mixed is something that I increasingly see kind of LLM do because like somehow the LLM thought it's clever to have like an

extra label here in an extra style meets thinks but we already have the company name we already have you know three different styles basically if you count count this here it's like four styles

this just adds a fifth um complicating the hierarchy not really adding it adds a bunch of vertical space like we could just like move the H1 a little bit up and then you know that value didn't trust it um logos would already be like

a little higher. Um so this is something that I increasingly see kind of like LLM's dream up that just isn't really tight information hierarchy and graphic design.

>> Yeah, I I think I going through these I I keep coming back to the concept of like it's easy to do anything. You

really have to be um opinionated as the human that's in the loop in designing these around what you think is the right thing. Not just kind of saying yes,

thing. Not just kind of saying yes, accept all changes.

>> Yeah. You're now the editor suggestion.

Yes. Yes. And and another thing that I'm noticing happening here is this section on the right, it just keeps swapping around. Um if we watch this animation

around. Um if we watch this animation here for a second, this button over here, the big star kind of button. Uh

yeah, like moves around and and now it's a circle button. I think that's a button. Yeah, I guess that's clickable.

button. Yeah, I guess that's clickable.

Um and then it's going to come back over to the left side and it's going to be a big round rectangle kind of button. And

so we have a lot of different button styles. I'm not sure why things are

styles. I'm not sure why things are moving like this. Um, and it's very confusing.

>> It's actually not a button. Like, yes,

it shows the hand, but that's a head fake.

>> Yeah, I guess if you click it, it doesn't really Well, I guess it's kind of like a next button, >> but then also on auto advance.

>> Yeah, this is very confusing to me. I'm

not sure what it's showing. And And I can't tell what's happening in this animation either. It seems to be some

animation either. It seems to be some kind of workflow.

>> Um, >> actually, you can just click anywhere and it it it advances. So, it's not even like this button or the whole thing is a click target.

>> Yeah.

>> This is again one of those things where it's hard for me to imagine a human just being like, "Let's have three different style buttons that every time you click it, they all reorient and shift and you know, it's it feels like the type of

thing that uh an AI design tool would have suggested and you would have said, "Sure, that seems cool."

>> Yeah, it feels like the visual manifestation of LM hallucinations.

Um, let's go a look at the menu here.

Um, all right.

Pretty standard stuff. I never like when like the menus can like go go away like in between. Yeah.

in between. Yeah.

>> Like there's the gap and then sometimes goes away.

>> What's happening on the right side of a couple of those? There were two of those and there's there's kind of a grid on the right. Yeah. With an icon.

the right. Yeah. With an icon.

>> Is that interactive or anything or is that >> Nope. It's just something the LM thought

>> Nope. It's just something the LM thought was cute.

>> How about if you hover over on Yeah.

Okay. So, it's just giving a different icon.

>> Icon. Now, the icons are actually kind of cute. Um, this seems somewhat

of cute. Um, this seems somewhat original and on brand here. Um, I would just never thought to to do this little little uh like if if you want to make a

point that on board is like four circles in the square arranged in that way. Like

if that was important, which it doesn't seem like it, like you probably just put the icon here kind of back to the like why reveal stuff as you hover that is essential, but it's probably non-essential. So, I can probably just

non-essential. So, I can probably just like leave it out.

What happens when we go further down?

Valued and trusted by. Okay, we got a bunch of lines going everywhere.

>> All right, so we've got that line following you down the page pattern again.

>> And now we're scroll jacked. We're

locked into this uh position here of the website in order to build up this animation.

>> Mhm.

>> Um and I wonder what it wants to tell me. like why is it important to capture me here on this point to like build out this animation versus it's just like showing it already

built in the in the built out state >> and I find the animation is getting all of my attention rather than what it says all the way over here on the left side so I'm not even noticing any of this and

this is not on the right side it's not giving me enough visual information to communicate something valuable about what they do or how the product works so I just kind of miss everything over here

on the side the animation is too distracting Yeah.

>> Outcomes.

>> It's also with these with these landing pages that kind of like build up and then keep you in a in a section for a little while. It's really hard to know

little while. It's really hard to know when you're like how far into the book you you have read. Yeah.

>> Right. Like if you have a physical book, you're kind of like, oh, halfway through. here. It's kind of like I don't

through. here. It's kind of like I don't know, maybe it feels like we're halfway through, but maybe, you know, it's gonna keep me for another 15 seconds on some some like really rich buildout um of a

of a landing page. So, the scroll indicator um uh is is a really important tool, I think, kind of like to get a a grasp on sort of like the you know, the

landing page. And here it's it's um it's

landing page. And here it's it's um it's completely failing me kind of like where am I in the website?

>> All right. Overall, like visually, I I think it looks nice. Um, it looks modern. There's a theme that they're,

modern. There's a theme that they're, you know, you've got the orange accent color. You've got the black as kind of

color. You've got the black as kind of your primary brand color. Um, you've got the line that's kind of following you down. And, and my hunch is that that is,

down. And, and my hunch is that that is, uh, sort of reproducing what the product looks like because we keep seeing all these workflows with orange lines in the product itself. And I think that's what

product itself. And I think that's what they're trying to communicate here. Um,

and so, you know, that makes sense to me. their headline up at the top. I

me. their headline up at the top. I

think uh does a great job of explaining what it is and who it's for. Um so I think they really nailed that. Awesome.

Sphinx, great work. Next up, we got build zero. Let's take a look here. Uh

build zero. Let's take a look here. Uh

build custom internal apps with AI. Um

so right off the bat, we got some purple gradients.

>> Yeah, there there we go again. Um and uh let's see here. So the menu nicely morphs. Oo, and then my pet peeve um

morphs. Oo, and then my pet peeve um dumb hover effects. Um just doesn't add anything. this vertical movement is just

anything. this vertical movement is just distracting and not cool. Uh it pushes also there's another thing where um because of the effect it kind of like

>> horizontally shifts the whole menu which kind of just seems like a bug. Um and

then here's another like effect um kind of like the arrow has to move somehow vertically. Why? Um

vertically. Why? Um

>> the arrow at least should go to the right also. It feels like when you're

right also. It feels like when you're hovering >> it goes like backwards. Um this hover effect here I don't mind. You know, it's like it's a black and white button and then it turns into the primary brand color. In this case, kind of purple.

color. In this case, kind of purple.

Like that's not a bad hover effect here.

This here pretty bad.

>> Um then what else have we got? Uh we got a little like interactive element here.

Can click through it. Ooh.

>> Okay. There seems to be some like bug with the selection here. Yeah.

>> Um if this was handcoded, like no way someone would have not noticed it. I

almost wonder like if you just oneshot landing pages um do you actually go and really like use it with the same attention detail as you feel if you had

like built it um from scratch and then because you don't um you just ship uh things that are that you wouldn't have um because you don't even notice the the little bugs that the LLM made.

>> Um go back to the engineers tab here.

Now, if this is their product, one of the other things that stands out to me is this kind of dashboard. Um, that's

got, you know, it's got like the red, green, blue, purple kind of callouts up here. That's one of the hallmark classic

here. That's one of the hallmark classic things that a lot of, uh, AI design tools will will actually create. Um,

>> yeah, every fake dashboard looks basically like like something like that.

>> They've got the icons that are a darker version of the lighter background color.

Um it's it's usually like the Google colors, you know, it's like red, yellow, green, blue prize primary.

>> Yep. Yep. So that that that we tend to see a lot. Um ideas to working apps in minutes transform.

>> Yeah. And these like kind of like bento bento boxes um is is not a bad pattern.

Um but it's also very non-original. Um

and so this LMS often times kind of come up with these just like icon at the top.

Uh and then a bit of text and you know u in this case kind of like a 3x two um that that's yeah quite common.

>> Yeah.

>> Uh we've got some quotes. We got the purple gradient.

We've got some frequently asked questions and then more gradients.

>> Yeah. you know, we're I don't know, a few into reviewing these now and and I think we just keep seeing a lot of these common patterns. Um, which I think just

common patterns. Um, which I think just highlights how ubiquitous they've become and how um it it almost like takes away from the brand and originality that I think most of these companies want to

create.

>> Um, when it it has elements that every other site and every other product has too.

>> Yeah. Look, I think the really cool thing about AI tools is that it kind of frees you from having to kind of like, you know, fiddle with like the technical details

so it can really work on the hard hard kind of questions of your offering, of your product, of your company such as like what are we making for whom, why is this valuable to them, right? And that's

the cool aspect. But if you then can like just use it um to create uh what we sometimes call like AI AI slop, I think this idea of having sort of like you know all this power that you should use

responsively um gets a little bit wasted. And so that's the fine line to

wasted. And so that's the fine line to walk I feel for for builders in 2026.

>> Yeah, there's there's nothing inherently bad about this. You know, it's just that we've been through a few and we've seen this before. You know, that that's kind

this before. You know, that that's kind of what it feels like is I've seen this before. And for any anybody else in the

before. And for any anybody else in the audience, like you lose a little bit of credibility, I think, you know, when your customers are are viewing your product and they're saying like, "H, this looks like a bunch of other things

I've seen. They probably just use AI to

I've seen. They probably just use AI to build this and design it." If that's the case, like how good is the actual product, you know? Um, or are they vibe coding that too?

>> Use AI to build this by any means. Um

but just kind of like you know really then evaluate the output and uh use AI to make it even better and more original and spend the time that you saved oneshotting it or whatever um to then

really think about your messaging and um and you know designing a wonderful product that you can then showcase etc. >> Yep. That's great advice. Awesome. Build

>> Yep. That's great advice. Awesome. Build

zero. Thank you.

>> Thank you.

>> All right. Next up we got Zarna AI associates for private capital markets.

Oh we're scrolljacked again.

Oh yeah, >> it's 10x everything. All the things.

>> Unlocked irr this this site. Yeah, it feels like I'm in molasses. It just feels so clunky to

in molasses. It just feels so clunky to use. Not to mention there just isn't a

use. Not to mention there just isn't a lot of like information. Um I feel like I have like even this is deal volume 10x like uh

>> this all everything 10x just feels made up. Um and it feels like there's just so

up. Um and it feels like there's just so much space. It feels like I have to

much space. It feels like I have to scroll through this molasses forever to get to anything that helps me understand what this is.

>> Yeah. One of the things I personally like, if we go to the top, um, right now the, uh, sort of background here uses 100% of the vertical space.

>> Um, which kind of like almost like vis visually kind of locks me in and then there's like nothing more.

>> Um, and I hate when when people can like put like scroll down to see more because like of course we're on a website. Of

course I know how to scroll down. So,

one of the things that I personally kind of like um to interrupt the visual um a little bit is kind of like to start a landing page like maybe like like this.

You know, now I have kind of like something here at the top. Um but I know there's like more good stuff down here.

The good stuff better be good because close better deals faster doesn't, you know, tell me much, but maybe there's like, you know, like 10 pixels of like an awesome hero image kind like peeking up or something like that. That would be

cool. And then I immediately without any

cool. And then I immediately without any indicators like know and want to can like scroll down and discover it, right?

>> Yeah. Well, we've stopped at the perfect point here because you cannot see anything in the top navigation bar.

>> Yep. It's uh you better know what's in the background and if there's a dynamic background such as a video playing which might have all sorts of bright colors.

Um all of a sudden your bright menu isn't uh readable anymore.

>> Okay, so this is interesting.

>> So here it picked it up. So it's smart, but it's not smart when there's video behind it.

>> Ah, you're right. That is the the nuance, huh?

Okay, these I can't click on. I would

have expected that I could click on any of these.

>> Maybe the chevron.

>> Oh, I don't know what just happened.

Okay, now I can click on those. Yeah,

some of it is clickable and some of it is not, it appears. And also, it's like jumping me up and down on the screen.

Um, it seems to just be moving on its own timer and not letting me do anything, which is confusing. Um, and

again, this is one of those things where it feels like you wouldn't intentionally design it to work this way. Instead, you

just it created a component and you went and it does sometimes work.

>> Only on those nested ones. If you try the other ones, it doesn't do anything.

>> Yeah. And this better label here is um super super hidden. There's sometimes a chevron, sometimes not. Um, yeah, this

feels maybe like a little bit sloppy.

>> Yeah, one agentic layer generate an investment memo.

Um, yeah, even the little bits of motion here is the kind of thing that like AI will just give you that for free and you're like, "Yeah, okay. Let's go with that." But you would never create

that." But you would never create yourself. Um, yeah. I can't even tell

yourself. Um, yeah. I can't even tell what I'm supposed to do with any of this. It hovers and expands a little

this. It hovers and expands a little bit, but like nothing is clickable.

We got some fadeins.

This just feels like it's lacking a lot of like useful information presented in the right ways. Um

and at least not purple though. Got a

lot of blue.

>> Yeah. Yeah. Yeah. Fit and finish as well as kind of like meaty content.

>> Yeah.

>> Um like clear messaging is maybe lacking a little bit. Um the visual style if I didn't have to scroll through it um is actually kind of fresh. Yeah. Yeah. Like

it it reminds me of uh lovable or or things like that that kind of have that like grainy zoomed in kind of picture in the background.

>> Not not sure what these boxes um here represent uh in the background.

>> Yeah. Some weird artifact.

>> All right, Zarna. Thank you. Okay. Well,

thank you everybody for submitting your sites and uh Rafael, we we've seen a bunch of them now. What are some of your takeaways here?

>> Yeah, I think definitely lots of purple gradients. Um, and in general kind of

gradients. Um, and in general kind of like a lot of kind of like the same the same patterns. You know, we see

same patterns. You know, we see animations kind of like for animation sake just because it's uh it's possible now. I would have killed to have these

now. I would have killed to have these kind of like AI superpowers when I built my first website. So I think it's wonderful that we have them, but you are still kind of like responsible to not

outsource your thinking to LLMs um and actually just use them as tools to get your brilliant ideas, designs um novel maybe kind of like interactions

um out on the web um to dazzle your customers and uh and I think that's what a landing page should do in the end.

It's kind of like a customer acquisition channel, right? This is not your

channel, right? This is not your product. This is not a web site as in

product. This is not a web site as in kind of like a cool thing on the web.

there's space for that as well, but these are like startup landing pages and they're basically customer acquisition channels. So, and that's how I would

channels. So, and that's how I would really um use these tools.

>> I totally agree with all of those.

Couple others I would add. You know, one I think it's really important that uh you QA everything. It's very easy to create something, but we went through and we found, you know, some quirks or bugs or things like that here or there

or things that were confusing.

ultimately you know you the founder you the human need to be the one that goes through and make sure that um you know the interactions and and everything else um work right. The other one is um you

know it it's important that your site and your brand they come through as the thing that you want to represent your company and that originality and something that feels different and it's

very easy using the same tools that everybody else is using to end up in the same place. you have to be intentional

same place. you have to be intentional about ending up in a different place.

And so, um, rather than start from what does the AI spit out and then let's tweak that, I would start from, you know, what color palette do we want to use and what is our brand and and how

can we feed that into the system and use that to make sure we end up at an end goal that represents us rather than just going with, you know, whatever uh the AI spits out there. We've seen a lot of

people do that and I think you know like you said one of the most powerful tools ever to create incredible designs and do things that would have been impossible or taken forever that you can do just

like this. And so um to use that for

like this. And so um to use that for ways that that actually help you move the lever uh upward and make the sites more engaging and convert better is one of the most powerful things that you can do.

>> Yeah.

>> Well, thank you so much for joining uh Raphael. It's been uh amazing to get

Raphael. It's been uh amazing to get your perspective on all of these and uh thank you to everybody that submitted websites and we will see you on the next design review.