How AI is Changing Design Workflows

Welcome to Dive Club. My name is Rid and

this is where designers never stop

learning. This week's episode is a

special one because it's a live

recording of a panel that I hosted with

Maven all about how AI is changing

design workflows. We get to hear from

three of my favorite guests in the

entire industry. The first is Henry

Modiset who's the VP of design at

Perplexity. The second is Prenati Perry

who's leading the design of Vzero at

Verscell. And lastly, Nick Patterson,

who's done the branding for Lovable,

Craft, Maven, and many of today's top

startups. So, without further ado, let's

dive in. I want to start off with you,

Nick. You know, the reality is, you

know, I spent too much time on Twitter

and many of today's top startups are

turning to you for branding. So, I I

just want to know like what how the

sauce is made and some of the things

that you're doing behind the scenes that

are allowing you to execute at this

level. So I asked people to kind of

prepare a little bit of screen shares

just to really dive into workflows and

what's happening. So I'll let you take

it over and we can start there.

>> So first of all, great to be here with

you all. Um so yeah, I thought that what

I would do is just do a quick

walkthrough of a project that we just

completed that we haven't actually

released yet. Um, we've been doing this

new thing probably for the last like 2

weeks, uh, where when we're starting to

build a graphic system, uh, we're

actually building the tools for the

graphic system so that we can scale them

pretty quickly. And this is one that I'm

super excited about. So, I thought that

I could walk you through it. And when I

say building these tools, I mean we're

building them in Lovable. or using G GPT

to support that so that we're talking to

lovable in like a super

clear effective uh way that it can

understand really well. Um so I thought

I'd just walk you through this briefly

and and show you how we're um applying

it and also giving these tools to

clients when we're done. Um so real

quick, Flow Glad is a payments processor

tool. Uh at its base it's very precise

and modular and mathematical. um the

founders are of Middle Eastern descent

and so what we unearthed in our first

call probably in the first like 20

minutes this doesn't always happen is

that they like the idea of Islamic

geometry is a possible graphic system uh

it's symmetrical it has lines that flow

and expand like the name flow glad it's

also could be infinite and so there's a

ton of possibilities for patterning uh

so we started to dig into a lot of

inspiration and found uh these patterns

early Um, a second thing that they liked

were these subway lines as well. So,

we're trying to think about how to

correlate both of those things together.

Um, and you know, if we went away and

had to actually draw all of this, it

would take an extremely long amount of

time. And that's not that we couldn't do

it. Um, but it's not super scalable. So,

um, we actually had the idea that we

could build first this pattern architect

tool, which I'll show you in a minute.

first um based on some of that geometry.

Um we used this core pattern here. So um

we built this in about an hour um and it

has like lots of uses. Um you can kind

of zoom in here and just use like um

single line patterns and adjust the uh

radius and the star density, zoom in and

out, work on spacing. Uh but you can

also change it over to where is this

thing a primary shape here and um we can

work on line thickness and gaps etc.

Very very fast way for us to work on uh

some of this patterning. Um so very

early on this is you know we do these

sprints in two weeks which is 10 days.

So we have tight timelines in order to

build and develop these things. So I

thought I'd just show you kind of day by

day. So this is the second day where

we're working on um some of this. We're

already built building some of these

patterns which we're starting to like

and then we're starting to integrate

some typography typography and think

about um some of that subway path line

stuff as well. And so we're able to

create those patterns and integrate it

really quickly. Um but then we also

wondered if maybe we like um sort of

dithered this and digitized it. Uh, so

we were able to build a second tool even

faster that was this half te pattern

tool. Um, which started to create some

really interesting results. And I think

like the main thing here is also being

able to just experiment and have fun and

play. Um, so we were all able to work on

this. We have four people on our team

and everybody's able to like start

creating these patterns. So only within

day two we're doing this. So, you know,

I think I think the cool thing is that

we're showing up for our mood board

presentation now. And it, you know,

moods for us used to be uh a lot of

found things, but we're starting to find

that we can produce things faster than

saying, "Hey, what if we did it like

these other designers who did it?" So,

our mood boards, I just wanted to show

you looked actually quite polished. Um,

you know, we're coming in and we're

starting to show them some of these

patterns early for our second direction

as well as our first direction. you

know, we're integrating this work into

those workflows and we're already

starting to get the vision. So, like it

becomes easier for us to believe in the

vision vision to get them on board,

especially when um they've already

expressed to us that this is direction

they want to go and it's just up to us

to execute something impressive and

interesting. Um so, we're actually like

getting pretty deep in the weeds and

woods. We just start executing uh

visuals that are integrated into web and

like how these patterns can work. Um, so

yeah, and I just wanted to like show you

show you briefly how this all unfolded.

>> Real quick message and then we can jump

back into it. By now you know that

Jitter has been my go-to tool for

animation for years now, but they are

still shipping like crazy. I mean, just

this summer they've released comments,

pen tool, morphing, text gradients,

Google fonts, and a bunch more. So if

you haven't yet, I promise you will be

shocked at just how easy it is to bring

your designs to life in Jitter. So, go

ahead and give it a try today. Just head

to dive.club/jitter.

I saw a scrolltoppping tweet the other

day. The creators of Tailwind are

working directly on paper to train the

output to be perfect. They even invested

in the company. So, just think about the

possibilities for a second. In the

future, you could design something in

paper and then just rightclick and copy

the perfect Tailwind as if the creators

themselves wrote it by hand. Or maybe

you take an existing code component and

import it into paper to make edits

directly on the canvas. I mean, this is

going to totally change how we design

and deliver UIs for the web. And it's

just another reason why I'm betting big

on paper as the next great design tool.

You can try it out today. Just head to

dive.club/paper.

Okay, now on to the episode. They love

this stuff. They had some feedback, but

they wanted to go with the second

option. Um, so we worked on some initial

patterns. We started to develop vectors

with the first tool that I showed you

and then we started to dither them with

the second tool that we showed you. So

we were able to like build build the

initial versions and dither them off.

But then we're starting to have like

clear conversations with them like they

felt like this sort of patterning was a

little bit like a ro shark test, right?

So like, okay, the tools giving us that.

How do we pull off of that? Becomes a

lot faster for us to be able to create

um patterns with the first tool that

don't have that. Uh so by day five, we

were honing in on these patterns that

felt much more clearly based on the

Islamic tile. Um and then we're honing

in on core patterning. Um and then that

allows us to back off of a lot of the

patterning and start to think about

applications in a more thoughtful way.

So getting into website design and app

design and also integrating these subway

patterns. So you know we could start

thinking through these additional layers

on top of it and then uh eventually get

to those executions and outputs thinking

through how these things can all

integrate uh into the website into the

app into data etc. Um, and I think that

the most exciting thing is that I just

have been, you know, getting on calls

with clients and on day three I'm like,

we built a tool for you. If you like

this, the tool is yours. So,

>> it's so cool.

>> This tool is theirs now. And so, I think

that there it's not just about how fast

we work. It's also about um, you know,

handoff, which is tough for companies

who are moving fast. We build a brand in

two weeks and, you know, they have to

implement it. And so I see this as a

real way that we can uh hand this stuff

off to clients and they can begin to

implement and scale instantly and also

play in a way that has parameters. Um

and even just giving clients the ability

to articulate what they want, you know,

having gone through this process

multiple times, the amount of times that

I get two options and then I'm like, I

think it's somewhere in between

somewhere, you know, but now I actually

have a tool to dial that in and express

my intent as the client. Super super

valuable. I haven't seen anything like

this in the brand process before. It's

really cool.

>> Totally. Yeah. Big agree.

>> And like thinking about the moves,

right? Like you know, how can I

extrapolate this brand that I've now

been given and having tool as a part of

handoff? I mean, you said that this is

like what a couple weeks old. Like is

this something that you're still pretty

green in? Like where do you see this

going?

>> Yeah, I you know, I'm not quite sure

yet. This is probably the second tool we

built, but uh we've done a couple more

since then. And even some directions

that clients haven't chosen, we have

tools for now. So, we sort of have this

graveyard of tools. So, we're we're

getting better at um building them and

expanding off of them. I mean, I think

that right now some of this feels like

it's in its infancy in terms of just

using it for patterning, but I've

started to wonder how we could expand on

this. If we're doing identities that

are, you know, dimensional and 3D, not

just flat 2D, how can we uh push that

and create really dynamic systems that

we can pass on to clients? So, I think I

think this is just the beginning for us.

It it feels very young in terms of what

we're uh tinkering with

>> and a lot of clarifying questions in the

Q&A right now. This is all lovable like

what we're seeing right now. Like you've

just built zero to one in these tools.

>> Yes, correct.

>> Well, let's just keep the pace going

then. I want to use this as an

opportunity to toss it to Pron who's

building one of these tools at V0ero.

And maybe you know as you're kind of

bringing up your own workflow stuff,

let's just start like high level too,

right? So we've have the example from

Nick about this idea of using these like

custom internal tools as a part of the

brand handoff process. You're obviously

approaching it from many different

angles, including more of like the

product design angle. So when you're

doing research to understand how teams

are using these AI prototyping tools,

what are some of the key trends or

signals that you're noticing that are

influencing the way that you even think

about product strategy for Vzero?

Yeah, I think um initially the tool I

think kind of started a lot as like a

designer tool where we would see a lot

of people create like really interesting

graphics, really interesting landing

pages, um shaders which I think VZ is

very good at as well. Um but recently

we've been in kind of investing a lot

more effort into things like

integrations because a lot of people I'd

say of like the sales or PM archetype as

well have been actually using it to

create not personal software but I would

say personal software to help them like

work better in their job. So a lot of

PMs will like kind of create interactive

mockups to help supplement their PRD.

Um, and it's just kind of like the

bridge tool that helps you kind of, it's

not quite Figma and it's not quite dev,

but it's like in between that just gives

a better idea of like what you actually

want to build. So, it's very cool to be

able to hook those things up to like

live data. We also have really great AI

integrations, too. So, like a lot of

people are doing image generation or

like building their own chat bots and

stuff like that, too. So, it's I think a

step above just a static prototype if

that makes sense for a lot of the

builder type people. you said an

interesting phrase which is like it's

not quite Figma, it's not quite dev,

which I think is spot on. It's like this

middle ground. And I think that anytime

you're dealing with that ambiguity and

kind of this gray area, it's a little

bit unclear for someone who maybe

considers themselves a professional UX

designer, when the heck do I reach for

this type of tool? So, you have like the

double perspective of like the the

industrywide perspective and also, you

know, you are a designer working on one

of these tools. When do you reach for

Vzero in your own practice? Like are

there slam dunk use cases and maybe are

there other times you're like, "Okay,

you know, it's probably not the right

tool for the job there."

>> Yeah, I actually had a really good

example. Just

>> let's do it. Go ahead. Yeah, go ahead

and screen share.

>> Personally, actually recently, I was

kind of doing some exploration for my

portfolio as every designer is doing

every month. Um, but I came up with

these kind of nifty little cartridge um,

like components that were SVGs that I

created or like vectors I created in

Figma that I exported to SVGs and then I

would generate images to like place on

top of them.

>> So, and I was like it would be really

cool to turn this into a tool where I

can just like dynamically generate a

bunch of these cartridges with like

slightly different tints and stuff like

that.

>> So, I actually made a tool for this. Um,

and then I made it so the text is

dynamic. So, I can like put in red. Um,

I can change the colors, which was

literally just a single prompt with

Vzero to be like, "Hey, can you tint all

of the gradients in this SVG to be like

correspond to the color that I want to

create them?" And even cooler was that I

can made like a transparent effect where

there's like a little CPU image like

behind the cartridge and then lowered

the opacity of the SVG. Um, but it

becomes like a fun tool that you can

actually just use to like generate

dynamic different things. And like I

said with things like integrations, I

could actually, right now this is just

static image upload, but I could

actually do things like connect it to

foul and like have image generation and

all of a sudden this just becomes like

an app that was something that I use for

my portfolio, but also becomes something

like that people can play around with.

Um, similarly, I've also been really

into using uh creating sprites. Um, so

something I've found that obviously

you're creating each of these like um

frames in Figma or AE Sprite or whatever

tool you have, but it's really fun to

create like these kind of like micro

tools again to like actually play

around, see what they look like in

production. And something I want to do

in the future is create my own RPG game

like on the web. So this is something

that I think works really well for that.

Um, and both of those things I like

obviously started Figma and I do all of

my ideation in Figma, asset generation,

all that stuff. and then I pull it over

to V 0ero to actually take it to the

next level. I think one application I've

been enjoying recently is I actually

have been kind of every week as a means

of debugging VZ as well or like catching

dimes is recreating like interactions

that I'll see in other products just to

understand how they work. So this one I

created recently is like the DIA new tab

prompt uh new tab animation um

completely made in VZO assets in Figma.

Um but it's super fun just to like

analyze the motion. Now you have this

asset that you can kind of like learn

from in the future. Um,

>> can I pause you and get really specific

for a second because I think people see

this and they're like out dope. This is

really, really cool.

>> How did you start? Did you start with

like this all in like comprehensive

prompt? Did you kind of chip away at it?

Like, do you feel comfortable scrolling

up in this chat? Cuz I think that's

always interesting and helps it feel a

little bit more like an on-ramp for

people.

>> I think go all the way to the first

chat. Um, sorry, this is going to be

like really embarrassing to see like me

like yelling at the LLM.

We scroll up and it's just all caps.

>> It's It's a pretty It's a pretty long

chat, but you can see things like I'm

like the borders are much better, but

I'm still seeing the gradient inside the

input, if that makes sense. And I'll

like give it screenshots of things that

it's just doing incorrectly. Um, and I

actually try to like kind of test my

test like the limits of Vzero. So, I

won't edit the code at all. Like I'll do

everything in chat just to see how close

I can get to it just from spoken word if

that makes sense. Um, something that I

also think that obviously people are

really encouraging like just completely

non-technical people like being able to

like wield these tools to do whatever

they want, but I actually find that

understanding at least a little bit of

like what is going on and like reviewing

your own code and PRs and stuff gives

you like a bit more like language and

like propensity to like guide the LM to

what it might be doing wrong, especially

when it comes to debugging. So, you're

not just building on a house of cards um

and like bandage over bandage when

you're debugging. actually can give it

constructive criticism. Um, but yeah,

this is like I'm I'm giving it like hex

colors that it should change it to. Can

we line up um animation timings and all

of that stuff? So, yeah, I get pretty

detailed, I would say, but um it's it's

definitely a little bit hard, I would

say, to manage context and like also

this all costs money. So, um prompt like

and like tweaking your prompt to be able

to get the most out of each each message

is something that I think people are

learning too.

>> Cool. I didn't want to interrupt you too

hard. Are there other examples that you

want to walk through?

>> So the the last two I kind of wanted to

show are Verscell specific. So one of

our brand designers at Verscell, Dan, um

uses Vzero a lot for like asset

generation for our conferences and

things like that. Um and very similar to

what Nicholas actually showed. Um we

just did Next.js conference I think last

week. Um and there were these assets

that we would use for like ads and email

headers and things like that. It was

like this dithered cube effect. And

rather than creating a bunch of

different iterations of this in Figma,

he created a tool in B 0 to like let you

adjust the density of these. There's

actually design mode where you can like

draw um you can actually draw these

cubes on this. Um and then you can copy

it to clipboard, paste it into Figma,

and then we have a bunch of these

different like headers that we use for

all of our conferences and stuff like

that. So that's an example of again

something really similar to what

Nicholas does of how our brand designers

use VZero just in their day-to-day

lives. Um, and again another example is

we have this signature big CTA that we

use in like a lot of our conference

sites. Um, we'll just iterate on these

in Bzero and hand them directly off to

design engineers. Um, especially because

our brand designers are so aware of how

they want the timing and the hover

effects and all of that stuff. They get

it all right in VZO and then just hand

it.

>> I love the example because it's totally

something that I would like to be able

to bring to the table as a designer, but

I otherwise have no business doing this.

you know, like I wouldn't make this

happen in something like Figma, but if

all of a sudden I can use natural

language and play a little bit freehand,

it makes a lot of sense. So, I I kind of

want to tap into Henry's perspective on

this because I think there's a lot of

talk about how AI is making us faster,

but that's actually less interesting to

me. Like what I'd like to hear from you

maybe is what are the ways that AI has

increased the scope of deliverables that

designers are bringing to the table at

Perplexity.

>> Yeah. I mean what I love about um both

hope what you all just showed is like

it's it speaks to um the fact that like

being a designer actually means like you

you you actually have a potentially a

lot of different roles. Um, sometimes I

think design is a practice of pure

problem solving and other times it's a

practice of communication of like

vision. Sometimes it's your vision,

sometimes it's PM, sometimes it's the

founders, whatever. But like designers

are like probably the best in an orc at

like figuring out how to like put it

into reality and and piece it together

like in you know some coherency. And so

sometimes you use a design tool to

literally mock up what the app will be

or whatever you're making. And sometimes

you use it to like explore and

communicate and tell a story and

visualize like you know a path and and

it's not about making it perfect not

about thinking through every state and I

think you know given that like a lot of

us are making software and it is

interactive it really helps to have to

be able to have a tool that generates

interactive things but you but that only

meant to be used as like a communication

tool like you can you know make

something in v 0ero

just as a way to communicate like

something that you're thinking or just a

way to like roll out something you're

thinking you can do that so much faster

um and it's like less dangerous and

expensive than like you know asking

engineers to work on it right because a

lot of times I think that there's a you

know we don't know what we don't always

know what's right uh I mean I we as

designers obviously and and we shouldn't

think that we do and so it really helps

to explore um and if you can explore

something interactive and cheaper and

faster that's like so valuable. So I

think like just the spectrum of like you

know every company works differently

what designers do in companies is so

different all across the board.

Hopefully not hopefully your job isn't

to just make presentations of potential

ideas. Um and actually getting to solve

problems too but it it's it's great to

have like tools that let you kind of

work in all the ways you need to

>> speaking and working all the ways you

need to. I want to kind of tap into your

perspective as more of the design

leader. You know, we've kind of had the

ISB IC perspective here. So, what are

some of the workflow shifts that you've

noticed at Perplexity recently?

>> Yeah. Uh I'll maybe separate brand and

product. Um the brand design team, we

have this sort of principle of of uh

both velocity and volume. Um we want to

be able to make a lot of stuff and put

it everywhere. It needs to be that like

Perplexity is like a brand that is just

totally ubiquitous. We're a small

company. Like we need to be loud and the

only way to do that is to like basically

have tools that let us do that very

fast. And so we've definitely like

really embraced every tool, explored

every combination of tools. We'll we'll

blend things together, merge things

together, whether whether we're making

something static or visual or or a

website. I think like the brand design

team in particular is just like fearless

with tools and trying things and

stitching them together mostly as a way

to just like push the boundaries of like

how a brand might be represented and

also how much you can make in a week. Um

there's like a freedom built into the

brand values of like we don't really

have like a pretty like hardline

guideline uh brand system that like you

know prevents exploration. Um so there's

like that side of it. On the product

side, uh I have we have designers that

are in our codebase writing production

code every day and so they're using

cursor and cloud code and whatever to

like accelerate like a like a proper

engineering workflow. Um there are

designers using tools to like prototype

ideas or visualize them and you know

making like a little um slice of

something as a deliverable. Um, so you

know, a lot of our like crazier like

animated stuff, sometimes it's just nice

to go into a sandbox and just like

tinker with it until it feels cool and

then use that to hand to an engineer.

Um, sometimes designers will just build

it themselves effectively using the same

method. It just depends on the designer.

But I I think we're able to talk we're

able to do, you know, I think like the

the theme of all of this is not being

boxed in by just a single tool that

makes static images, right? We'll we'll

use anything that lets us communicate

like the the actual complexity of

software you know the all the states and

all the animations and you know because

all the stuff at this point like the

amount of things you can render in a

browser is astounding especially when

you get into like 3JS and beginning to

do 3D stuff like the only reason we're

not doing it more is probably because

it's like the tools don't really work

that way. Um, the rendering abilities

are there. So,

>> um, we're trying to like, you know,

remember to push the limits on that. Um,

as long as we're not making things

awfully confusing. Hey, really quickly,

let me tell you about the allnew Dive

Talent Network. I've hand assembled over

a hundred of the most talented designers

and builders that I know so I can

recommend them to my favorite companies.

So, if you're listening to this and

you're open to new opportunities, the

talent network is anonymous and super

low pressure. It's just an easy way to

see what's out there without having to

post on social media. So, if you're

interested in joining or maybe you're

looking for your next hire, head to

dive.club/talent.

For me personally, like you all are kind

of leading the charge in terms of just

inspiring me to get out of my comfort

zone of existing tools that I've become

so familiar with over the last 10, 15

years. And I look at people like, you

know, Gunner bringing shaders to the

table or ESHA bringing music to the

table and all of a sudden it's clear,

wow, that's part of the user experience.

Like that that is part of the UX that

someone is touching and feeling and

hearing and it was entirely

conceptualized by a designer and

executed by a designer and it has

nothing to do with rectangles or even

CSS like that. That's the future.

>> Absolutely. Yeah. we've been uh

essentially have as we've worked on um

games uh which we've we've launched one

um it built in the comet but that's like

like I have a there's a game designer on

the team who does like the game design

you know the mechanics of the game and

then all the other designers get in

there and add like you know then when

they're thinking about the UI and the

and they're generating the music and

like are composing it and uh it you just

add like you get like so many different

brains from different worlds combined

And then then the start things start to

get special.

>> I want to talk to on tooling for a

little bit and I even like the language

that you use. We talked about kind of

stitching together or chaining together

tools because that's almost the new art

of it is figuring out what do I reach

for and how can I get things to work

together and it's a little bit like the

wild wild west. So, I'm sure at least

some percentage of people who have tuned

in are curious what are the tools that

are becoming more interesting or that

you have your eye on or that people are

exploring. And I'm going to open this up

to everybody pron and and Nick too in

case that you have answers, but you

know, what are you seeing or tinkering

with that you think is worth paying

attention to as a designer, especially

when sometimes it kind of feels like

we're drinking from a fire hydrant of

new tools right now? I would say like

the most sort of fully adopted and

integrated is cursor um on the product

design side. Um and some people have

just a variety of setups related to that

like they may be using cloud code in

cursor whatever but um just as like a

pure accelerant of like exploration. Um

otherwise people do use V zero or

lovable or just like anything that lets

them kind of throw an idea together um

to see it. Um especially if they're want

to see like especially like a very small

slice like you know an animation of a

moment um and they don't have to worry

about like the rest of the codebase.

It's very useful uh for that stuff.

Otherwise on brand I mean I think any

tool that's ever been made they've

probably used it. Um, I mean all these

things have to come together in the end

um in like a in a traditional tool like

even when we're making videos like it

all ends up in like frame.io know um in

the end but like so many other things

happened at the beginning you know maybe

something started with midjourney then

it went into Craya or went into became

animated um and then you know then maybe

someone sorry Sora and then someone used

to make some music there's like

everything is kind of being experimented

with and then but then it all needs to

get stitched together in like a a proper

tool so where where it all goes is

interesting because you know there's

some like you said there's an art to

like knowing what works well for what

and that's just like late nights and

exploration. Eventually, I think this

stuff will get streamlined.

>> I mean, I feel like we're kind of like

uh lightweight on tools right now. I

mean, we use MidJourney. We were big

Visual Electric users. Uh

>> RIP electric uh

>> Henry Stolen.

>> Love, they're busy.

>> They're busy. Yeah.

>> Um but yeah, we use both of those and

then a lot of lovable. I mean, mainly

just because we we've been in there for

a while now. I would say that that's

pretty much the extent of what we're

doing now, though. But I'm very curious

about Flora. Um, I'm going to their

event on Thursday. I think they're doing

some really interesting stuff and I'd

like to learn more about it. So,

>> digging into that a bit too.

>> Pr I have a question for you as kind of

the representative IC product designer,

right? So, how do you think about when

to reach for different types of tools

based off of what you're working on and

where you're at in the design process?

When are you in more traditional Figma

land? And I'm talking about like for

like your professional deliverables at

Verscell. When do you reach for Figma?

When do you reach for Vzero? How often,

if ever, do you jump into more

production codebased tools?

Yeah, I think um my stack I'd say I'd

like split it up into like pre during

and like postdevelopment. It's like pre

I'm like in V 0 and Figma like

ideulating about like what we even want

a feature to look like between Figma and

Vzero. Um I like to say that we rarely I

think use V like Figma as a source of

truth for a lot of our UI these days.

Um, we have like a very basic obviously

component library that we use to like

quickly create components and stuff like

that in Figma, but for the most part

code is like the source of truth because

everyone's kind of collaborating. Um,

I'll use Figma or sorry m0ero for things

like billing flows and stuff that like

you're not doing all of like the

spaghetti prototyping in Figma. It's

just much easier to like spoken word be

like okay open this modal if you press

this button and then it's like very

clear to like an engineer. They can play

around like see all the edge cases and

implement it themselves. um during I'm

like a huge flaw code user. Um I think

we implement I think the myself and

Pablo who just joined the VZO design

team we implement most of our designs I

think I would say at this point which is

pretty awesome to be able to like see it

from like end to end. Um, and I think to

Henry's point about moving fast, I think

the one thing for us that has really

improved there is because implementation

is now like table stakes and and is so

fast, it gives us a lot more time to

think about like the creative side of

things of like, okay, now that this is

like implemented, what's like the next

step, which would usually happen like

two or 3 weeks out of like shipping a

feature, but instead it's pushed up to

one week because implementation is so

fast. Um, and then post, like I was

saying before, um, is mostly like

versatile agent, which we have, which

leaves comments on PRs and like gives

you an idea of what you're doing

correct, what you're doing wrong, and

all that stuff. But all that to say, I

think like from a product design

standpoint, I am using AI a lot

throughout the ideation process, I think

I tend to lean on it more once I have a

better idea of what I want a feature to

look like and I'm like in the edge case

stage. Um because it's just much easier,

like I said, to just speak the edge

cases than it is to like mock each one

up, especially when you already know

what you want it to look like and and

then it gets boring of like this but fun

leads to this, this button does that,

like all that stuff. So that's where

it's kind of helped my life as a product

designer. I would say

>> I have a very pinpointed question that

I've been thinking about recently, which

is in the world of using

AI prototyping tools, vzero, lovable,

that kind of a thing. There's a very

wide spectrum of the size of a

prototype. You could make something

that's a pinpointed interaction, right?

It's like, I just want to make something

that I'm considering part of the handoff

process and it's how this thing feels

and moves and then there's like a I'm

going to make this entire flow, right?

And I think you sometimes run into

issues the bigger you get. How do you

think about where you want to fit into

the spectrum and how far down this, you

know, I'm prototyping whole flows do you

go do you go with these AI tools?

>> Yeah, I mean I think we're kind of I

think on on our team and I think a lot

of other teams too, we're past the point

of like really prototyping entire flows.

It's like once you're getting to the

point where you're like building the

whole product in like a prototyping tool

and you might as well just implement it.

>> Jump in the cloud.

>> Exactly. like halfway through once I

like get a general idea, I'll usually

just create a PR. Um, but I do think we

have a lot of customers and we have a

lot of people that use Vzero that are

building like entire products like

within VZ and like huge apps, hundreds

of versions and like

>> that's so cool.

>> They're saving a lot of time. Like

there's a lot of like I'd say like

products that are very very UX focused.

So think like um healthcare, banks, like

stuff like that that UX is like mission

critical for them. It's like so easy to

use Vzero because you're like speaking

the UX rather than like having to create

it all by hand. If that makes sense.

>> Are there teams doing that like

together? Like a full team only in VZ?

>> Um I I think so. I think like forking

and remixing chats is like a really big

like thing for users of Ezero where it's

like someone will create something and

there's like millions of different

versions that like people will go off

and do on their own where they like want

to work on an isolated page of a website

someone's creating or something like

that. So we see people collaborate like

that. Um but I think real-time

collaboration is still something that

we're trying to figure out.

>> Yeah, you have to like reinvent version

control and commenting and yeah

>> slippery slope. I'm just going to

underline what you said, Pride

point at say, well, AI prototyping tools

are a waste of time is like they get

stuck in every little piece of surface

area that you have to account for and

fine-tune in order to make the entire

flow feel real. And so I was kind of

interested in your perspective almost as

someone like maybe you dog food the heck

out of VZ and do everything in VZ. you

know, I wouldn't have been surprised if

that's your answer, but for you to even

say, "Hey, this is really good at

focused prototypes, but I'm quick to

jump into Claude and owning more and

more of the front end," I think is

really noteworthy. And I remember

talking to you, Henry, gosh, probably

was over a year ago now, which is

insane, when you were dabbling a little

bit more in IC land and and now I know

you refer to yourself as a professional

email sender. So from your vantage point

with the rest of the team, how often are

designers owning the front-end code and

actually implementing their designs at

Perplexity?

>> It's totally uh on a spectrum. There

there are some designers who have like

years of writing production code and

they kind of view the this moment as

like a major accelerator like the same

way an engineer might that was my

personal experience. I was used to it.

Um, and used to being like, you know,

essentially like part of the engineering

team and used to dealing with all the

stuff that sucks about it like dev

environment and going through PRs and

whatever. Uh, and all of this like was

an accelerant. Uh, there are designers

that are um more that have never done it

before and we just have like such a a

culture of like learning and teaching

and embracing new tools. And so there

are designers that they're like, "Okay,

well, can I learn? Can I learn how to

write a little bit of code? And and and

that that's mostly coming through and

like fit and finish like that last mile

of getting the animation right or adding

like some extra weird juicy detail. Um

that is like you know in my experience

like it's hard to unless you're building

like a full prototype of the app like

it's kind of hard to even feel out some

of those things um unless you're doing

it in code. So I think that's just added

like an extra level of um I would say

like brand in the product. Um so

designers on my team like Gunnar have

who have never coded before are now like

writing Smith UI CL you know with the

help of cursor and I think it's like

pretty cool and then there are designers

who are just using it as like a sandbox

and they'll just like you know visualize

an animation or prototype an idea and

it's very much like more of a

communication tool. Um, so we we really

have like the full spectrum, but uh on

our team like anyone that wants to learn

and is uh embraced, which is cool.

>> I'm going to highlight this question

from Diana, and I think she hits on

something that I see a lot, which is

almost this pressure to

use AI as a designer. Top down, the boss

is like, "We need to use AI, we need to

use AI." And there's this built-in

skepticism because it's, you know, by

nature almost feels just about cost

cutting and efficiency for the sake of

margins. for someone in that position

where they're not working for themselves

like Nick or in more innovative

companies like Verscell and Perplexity.

What are some practical ways that a

designer can get momentum with some of

these tools and just start to

incorporate them into their practice

when they don't have the concrete

guidance of a culture that is making a

use of them and figuring out where the

ceiling exists? I my perspective is like

you know the more you can communicate

the more empowered you are. I think like

being relegated to making rectangles is

actually like there's not much power to

that. And you know, like if if you've

got like a vision for how the product

should work or how the brand could feel,

being able to like make that to bring

that to fruition um with like something

very robust like a prototype or like you

know being able to quickly put together

like a whole different brand direction

like you're actually whatever is in your

head is more likely to then be

communicated with with these tools. And

so I purely see it as that. I've always

felt like the benefit of designers

coding is that like it's kind of hard

for me to explain what I think how I

think the software should work and it's

easier for me to do it myself and I

think with all these tools you're just

able to make something way more um you

know it have way more depth and way more

interactive and uh and you can do it

with you know in a couple days like I

think that's that's the reason why we

you all should do it is so that you can

just like show more and show what's in

your head easier. I think the answer is

gravitating towards code. The more and

more that I have these conversations,

like the reality is AI doesn't

necessarily make you faster at doing the

old thing that you were doing in Figma.

It opens the door to new types of

deliverables and what you can bring to

the table. So, Pan, maybe I can double

click on your experience as a way to

follow up with that because I mean,

gosh, your first full-time design role

was what, like 2021, right? like you you

went from there to playing at like the

highest stakes table and I know you had

a little bit of like a engineering

background maybe. So I I want to tap

into your perspective as somebody who

can kind of see both sides of the table

but is still you know you're not like

looking for your exit ramp for your

career you know like you're thinking

about where is this going to go for the

next 5 10 plus years. How do you think

about where the new technical threshold

is for designers to be able to make the

most out of these tools, adopt these

workflows, and really set themselves up

for success in this new era of design?

>> Yeah. Well, I mean, I think like 80% of

design is just like having the idea and

like the the creativity to like know

what you want to like make and then the

other like 20% is going to be like an

everchanging like suite of tools like

that everyone's using. Like it's I who

knows if like any of the AI tools that

we're using now will even be around like

two or three years from now. So I think

like back when I was doing a lot of I

was actually just talking to someone

about this the other day of like coding

without any sort of AI assistance like

feels so insane to me now even though

just two years ago like we were doing

that. Um so it's like I think things

will just become the new norm of like

now I like implementation is table

stakes. So it gives you a lot more time

to think about how it should feel, how

it should look, how it integrates into

like your your daily lifestyles like the

tools that you're building. So I think

like I said just having the idea is like

the hard part and like learning the

tools I think is the easiest part

especially when like most of our

interaction with AI now is just spoken

word like if you just have a list of

running ideas and you just like chat

about it that can now turn into a

prototype. Um so yeah I think like it's

much easier today to create like a

prototype or create a proof of concept

than it was I think even just two or

three years ago which is makes me really

hopeful. I think about the future of

like younger designers and stuff too

>> puts more of the emphasis on

communication at that point.

>> Yeah.

>> I'm gonna toss it to Nanny who had an

interesting question here kind of

touching on the fact that this

democratization of building makes it so

that the lines are a little bit blurry

between design and front end but

definitely design and maybe a PM. when

both people can spin up a prototype, how

does collaboration change and what

should designers be thinking about when

it comes to just existing at a high

level inside of an org? Maybe Henry, if

you want to kick us off there.

>> Yeah, it's tough. I mean, in in some

cases that's like a total disaster. In

some cases, it's amazing and you end up

making better software. My my

perspective has always been like it

doesn't make any sense that we use a

tool that makes images as a way to plan

out how software should work, right?

First principles like like I started

when I started my career it was like you

know we were using HTML and just writing

Dreamweaver or using Dreamweaver to

write HTML and then we all use Photoshop

and it felt crazy to me. Now we have a

Figma which is like the first tool that

kind of became really popular for

product design but it never made any

sense. So if you really think like how

do we make software it does make sense

that people will be building interactive

prototypes as like the the way to design

the software. Um if anyone can do it

that can be great. It can be like

confusing. I think like it really just

is going to depend on the org. And I I

think like the the roles should blend

more and it should become more about

like you know a culture of being

decisive and having clear ownership.

That's certainly what we've what we've

had to do ultimately like the people who

have been designers have like a very

specific skill of like comfort with

ambiguity and being able to like pick a

direction in in a in a moment where

there's a million choices like the it's

easier to make stuff. So then we have to

be more decisive because there's there's

more choices. So I I kind of feel like

the people who have like ended up being

designers will probably still be

designers or whatever the you know like

the role of design and the type of

person in there will be the same as

always. It's just like the level of

decision they're making a little bit

different. I'm going to end on one last

question from uh Marique here and again

you know a lot of talk about designers

being in the codebase owning more and

more of the implementation and my my

TLDDR of the question is basically how

do you not mess it up you know like what

do you what have you learned about

getting into the codebase making changes

working with these tools that is

impacting the way that you everything

from prompt to strategize where to

insert yourself. Uh I I'm sure there's

people listening who maybe even are a

little bit excited about getting into

the codebase, but they want to make sure

that that first experience is one that

makes it so that their engineering

counterparts want to continue to have

them participating. So if you have any

advice for those people who maybe will

hang up here and take their first step

the next day. I I feel really lucky to

work at a company that just lets me like

do whatever I want the codebase um in

the name of like experimentation. Um so

I've like you just have to make mistakes

to like I think learn from them. And

like the biggest thing I've been kind of

pitching like recently is that black

coding is like awesome and I think it's

been getting a lot of people past like

the initial hump of like learning how to

like make your ideas real. But I think

like the next level is like actually

what comes after it of like are you

learning from like what the LLM is

actually writing? Are you just like kind

of blindly shipping code? Um cuz the

more specific you are about like what

could be going wrong or like what

exactly you want to happen from a

technical standpoint, the better the

LLMs are. So I think it's like right now

we're at the stage where like most

models are like interns and you kind of

have to like guide them in the right

direction and the only way that you can

do that is by like still kind of keeping

up with what actually happens

technically um for what you're creating.

So I think there is like a suite of

tools that is now emerging in like the

like review stage of like PRs and stuff

like that. Um but I think the learning

aspect of all of this will never go

away. It'll just be like a different

type of like learning of

>> you're not learning how to write the

code. You're learning what the code is

doing. Um, so yeah,

>> that's something that I've been doing

even in like clawed memory recently is

like putting it at the top level and

saying I want to learn like I am here

like break it down for me. Don't just do

the thing but look for any opportunity

to teach me what's going on. And I was

recording an episode right before this

and making the joke that at the same

line item or I'm just like oh also for

context I'm a designer and my goals are

two-part. One to make the thing and two

to not make my engineers angry. So make

my footprint as minimal as possible and

I've had a lot of success with that as a

practical tip as well. There are we

started a few minutes late so I know

we're late. Everyone is sticking around

for the most part. Henry, can I ask one

more question before we ride off into

the sunset? And I want to tap into your

perspective as someone who is hiring and

thinking a lot about the evolving talent

landscape and how to pair people

effectively. So given, you know, I think

somebody in the chat used the word

uncertainty. Given the times that we're

in and everything kind of being up for

grabs a little bit, what are some of the

main signals that you are looking for

when reviewing design candidates?

There's uh there's certain values I look

for that are like we most of the design

team is has some background in either

being a founding designer or like a solo

freelancer. So there's like a someone

who's kind of used to just being in a in

an environment where they have to make a

lot of decisions by themselves. So there

there's that type of thing. In terms of

skills, you know, the thing that won't

ever go away is product intuition. the

90% of product design work is picking

the right thing to build in the first

place and saying no to like most ideas.

And it's especially true on consumer.

And so we really like we really

emphasize that it's really like product

intuition. Why would someone use this?

How does it going to work? What makes it

intuitive? How does it fit in the

market? Uh what makes it compelling? Um

all of these things that are really like

they're they're certainly not visual

design. They're not interaction design.

It's really just like what is the thing

and why would someone want to put it on

their home screen? Um that's the thing

we test for the most. Um the other thing

that won't go away is just like the the

sort of basic craft of like I mean I

would almost just call it graphic design

like color and spacing. Like we we we

need that. We need people to know what

what good and bad is and what what

interesting is and what you know what it

means for something to have character

and that that that's stuff that is like

that's a an AI is going to give you

choices and a designer needs to make a

choice and it's just sort of like you

know there's a duality to that like

product and uh you know the kind of art

direction like someone needs to make

those calls and typically people who

have chosen these designers already are

pretty good at making those calls and so

I we really look for those things aside

from that someone who just wants to keep

learning because the tools are changing

like really quickly.

>> Yeah, I think that's a perfect line to

end it on even just reflecting on the

nature of the demos that we shared like

there's a freshness to them, right? Like

you are very clearly experimenting and

tinkering and trying new things and the

last thing that you can do is sit still

when everything's changing so quickly.

So, I appreciate that and and we'll kind

of use that as a way to wrap up. Thank

you so much everyone for joining. This

has been a ton of fun. Appreciate all of

the questions and getting to hang out

with many hundreds of people like this

is is pretty cool. So, don't take it for

granted. Appreciate y'all. Before I let

you go, I want to take just one minute

to run you through my favorite products

because I'm constantly asked what's in

my stack. Framer is how I build

websites. Genway is how I do research.

Granola is how I take notes during crit.

Jitter is how I animate my designs.

Lovable is how I build my ideas in code.

Mobin is how I find design inspiration.

Paper is how I design like a creative.

And Raycast is my shortcut every step of

the way. Now, I've hand selected these

companies so that I can do these

episodes full-time. So, by far the

number one way to support the show is to

check them out. You can find the full

list at dive.club/partners.