HomeVideos

Paper Is My New Favorite AI Design Tool

Now Playing

Paper Is My New Favorite AI Design Tool

Transcript

721 segments

0:00

Paper is a new design tool, but the

0:02

biggest value I get out of it is that it

0:04

feels like an extension of code. I'm

0:06

able to work as a designer almost like

0:08

an engineer directly inside something

0:10

that still feels like a design tool. If

0:13

you're designer, a head of design, or a

0:15

founder running a product and trying to

0:17

figure out how design fits into this new

0:19

process, this one's for you. Hey, I'm

0:20

Diane. I'm the co-founder of the Design

0:22

Project. We help over 50 B2B SAS teams

0:25

ship products faster with AI. Right now,

0:28

we're heads down building AI managed

0:30

design systems. Our customers have been

0:32

acquired by Nvidia, Slack, and Mr.

0:33

Beast. And if you're interested in

0:35

learning more about how we can help you

0:37

rethink how you build product, there's a

0:39

link in the description below to get in

0:41

touch. Okay, so today I'm going to walk

0:42

you through how I actually use paper in

0:45

my day-to-day workflow. Not just the

0:47

pixel pushing, but how I orchestrate an

0:48

agent to do the pixel pushing alongside

0:51

with me. We go back and forth until we

0:53

land on something and then the part that

0:55

actually matters, I can take it straight

0:57

into code and then I can actually ship a

0:58

PR. The shift is my job isn't pixel

1:01

pushing anymore. And yes, design is way

1:03

more than pixel pushing, but it did take

1:05

up a big chunk of my time. Now it's more

1:08

directing the work, reviewing it, and

1:10

getting it into their product faster

1:11

than a normal design to dev handoff ever

1:13

could. I'll also show you Paper's token

1:15

system. It's brand new. They just

1:17

launched it and it's super cool for

1:19

keeping everything I build aligned to

1:21

our actual repo. I'm also going to test

1:23

the exact same thing in Figma and then

1:25

you can see live side by side exactly

1:27

what paper can do and then what Figma

1:29

can't. Let's get started. Okay, so here

1:31

we are in paper just to show you around

1:33

um really quickly. We're going to be

1:35

working in this folder called carrot.

1:37

We're going to be working on a new

1:39

carrot project and I'm going to kind of

1:40

show you how I start to set it up in

1:41

general. So I'm going to go to my CLI

1:44

whatever I use. I use conductor. Um it's

1:47

one of my favorites. Um I'll link to a

1:49

video below if you want to see a little

1:50

bit more about my favorite tools. But

1:52

what I'm going to do is basically come

1:53

here and say

1:55

I want to create a new project in paper

1:59

in the carrot folder. And the first

2:01

thing I want to do is extract a design

2:04

system from the current carrot product

2:08

and repo.

2:10

So let's hit create. So it's going to

2:12

start this project.

2:15

And let's hope I get it because it's

2:18

carrot, but it's actually cr. So, let's

2:20

see what happens. Um, but what's really

2:22

cool is I'm actually just going and

2:23

telling it how to navigate directly in

2:25

in paper and to create this file. So,

2:29

I'm jumping straight into it. But what

2:31

is super cool is I am telling it to

2:34

extract a design system. Something that

2:36

Paper just announced recently is they

2:38

now have tokens, which is something I've

2:39

been eagerly waiting for. So, um, this

2:42

is kind of showing you how they actually

2:44

are taking, you can actually create the

2:47

tokens directly into paper from the

2:50

codebase, and you'll kind of see it come

2:51

to life in a second, but it's really,

2:53

really cool. Um, and it's also just

2:55

going to spin up a whole design system

2:58

that I can start using to continue to

2:59

build on top of it. So, this is the

3:01

first place that I suggest starting. Um,

3:03

and you'll see why once we get into it.

3:05

What I also want to do is let's compare

3:07

this to Figma, for instance. So let's

3:09

let's kind of ask the same thing. Can

3:11

you use the Figma MCP and can you create

3:14

a new page in Figma and design a design

3:19

system based on the carrot design rules

3:23

that we have in the codebase. So let's

3:26

kind of compare the difference between

3:28

what Figma is going to come up with and

3:29

what what Carrot's going to come up

3:31

with. And we'll kind of come back to

3:32

this um as it goes. So what it's saying

3:34

is it's going into the repo and looking

3:36

at the current tokens so that it can

3:38

pull them out. It found a mismatch.

3:41

Okay, this is good to know. Okay, so J

3:43

mono with J used as a body override

3:46

going to affect the implementation

3:48

system in paper. Perfect. So we'll be

3:51

able to see the difference here. And

3:52

then some of the names are generated

3:54

differently. We'll take a look at that.

3:56

So it's saying dashboard primitives use

3:58

10 compact 10 to 13 operational controls

4:02

status colors and the widget

4:05

design brief for paper.

4:09

So it's defining the color palette we

4:11

have. So it's saying tube black, tube

4:13

soft, phosphorous white. This is how we

4:15

define them directly in the codebase. So

4:17

it's just pulling exactly what is

4:19

defined there. And then it's using the

4:21

types. So implemented to tokens using

4:23

the dice mono and J for longer body copy

4:28

creating a fresh paper file now. Okay,

4:32

so it says it's working on this. Let's

4:34

go back to the dashboard. Oh, here we

4:37

go. So it didn't actually put it in the

4:39

carrot file. Let's see if I can get it

4:40

to move it thereafter. Okay, so it's

4:42

just setting everything up. Let's give

4:43

it a minute. So let's go ahead and look

4:45

at what's happening in Figma.

4:48

So it looks like it is here. Let's open

4:51

Figma. Let's give it a minute. I think

4:53

it might be loading something. So, it

4:56

says it's starting to lay them out.

4:57

Okay, cool. Here we go. You see things

4:59

are happening. Um, this something I

5:01

think is really fun about paper is it

5:03

shows that things are happening that

5:05

that it's working on and you can kind of

5:06

see it come to life, which is always

5:07

super fun. And also, you can see that

5:09

there's an agent in here working as well

5:11

as me, which is really cool. Um, I they

5:14

also have a collaboration feature. Um,

5:16

so someone else can hop in here and you

5:18

can see them. Um, and I thought this was

5:20

fun. You also, if you do slash, you can

5:22

like talk to the person. Hey, YouTube.

5:25

Super cute. Um,

5:29

but it's basically similar to Figma

5:30

where many people can kind of hop in and

5:32

you can work on the same file in

5:33

general. Um, maybe some other things is

5:36

it has some nice features where like

5:38

shift hide. So, if you do command point,

5:42

you can hide the navigation and be able

5:44

to like dive deeper in this. So, we can

5:46

leave it like this for right now, which

5:47

is nice. Okay, so it looks like Figma

5:50

has created this design system board.

5:55

Um, here, let's actually go into Figma.

5:57

So, it's created this and now it says we

6:01

can click send to Figma.

6:04

So, now Okay, so now it's going to open

6:07

directly in Figma. Oh, interesting. Oh,

6:10

it actually went into a page that we

6:13

were already working on. Okay, perfect.

6:15

um for some early stuff we were doing

6:16

and then it created this

6:20

design system.

6:22

So it has colors. Okay, here let's take

6:25

a look here. Um so see it doesn't define

6:29

the tokens. Um so that's something that

6:32

I think is a miss. Okay, so it looks

6:35

like it's finished. Let's take a look.

6:38

So here we are. We have the product

6:40

design system. Um actually let's bring

6:44

back

6:48

um and see here. So it's actually gone

6:49

in and it's created

6:52

my token directly from here which is

6:54

really nice. Um

6:58

it also let's see

7:00

so it has type tokens and scale tokens.

7:05

Um and I can show you kind of what that

7:07

looks like in a minute. Okay. So, here's

7:10

all the color tokens here, which is

7:12

great. Um, and you see here it's defined

7:15

all of the radi. So, I can choose which

7:17

one I want. Um, same with

7:20

all of my spacing here is already

7:22

defined from my codebase. So, it just

7:24

created it directly for me. Um, so

7:26

spacing radius, dashboard status layer,

7:31

right? So, these I actually want to work

7:33

on at one point in the design. Um, okay.

7:36

What's the product actually ships?

7:38

So, it has these

7:41

pills and buttons, which is great. Um,

7:45

so what it doesn't have is it doesn't

7:46

have components, like you can't create a

7:48

component in the same way. You can do it

7:50

with code, um, which is cool. So, I

7:52

could just like keep editing and

7:53

changing these, but it's not exactly

7:55

like Figma quite yet where you can like

7:57

actually build out a component library

7:59

in the same structure and way. Um,

8:03

comment, popover shell. Um, yeah, this

8:05

one is actually one I want to play

8:07

around with in this video. Um, okay. And

8:10

how to build the next paper project from

8:12

this system. Okay. Interesting. It gave

8:13

us like a how-to. So, it's really

8:15

helping us to get to the next stage. So,

8:17

this is really cool. I think the biggest

8:18

takeaway here is like

8:21

that you can get a fully functioning

8:24

token styling so that you can just start

8:26

building directly. And then if we go

8:28

back and we compare to Figma,

8:30

um I think that there's more direct

8:32

rules that are associated here which

8:35

make it nice. Um we could potentially so

8:37

this is foundations

8:39

component primitives and then product

8:41

surface inventory which

8:44

is fun but I don't really need. So let's

8:46

keep going here. I want to continue to

8:50

design and build off of like a pro a

8:53

feature that I'm already working on. So,

8:55

I want to do um the comment popover

8:58

shell. I want to work through these a

8:59

little bit. So, what I'm actually going

9:00

to do is go back into

9:04

my codebase. I want to play around with

9:05

the comment popover shell. Can you

9:08

design all of the different states that

9:10

we have right now so I can adjust and

9:13

play around in paper?

9:16

Okay. So, while it's doing that, also a

9:17

couple things that you can do. So now

9:19

you can take your Figma designs and you

9:22

can actually just copy them and paste

9:24

them directly into

9:27

paper. Um, and everything translates,

9:30

right? So that's great. Um, so you can

9:33

kind of play around like anything that

9:34

you currently have in Figma, you can

9:36

just kind of move over into paper right

9:37

now and continue to work through that,

9:39

which is really cool. Um, okay, here we

9:42

go. Starting with the popover states.

9:44

Perfect. Something that is really cool

9:46

also is they have this plugin. So I can

9:48

come here and basically I can capture a

9:50

whole page. Command enter. It's

9:52

capturing this whole page. This whole I

9:54

can go into my browser, go to any page,

9:57

any product, anything. And it's going to

9:59

be able to capture the selection. It's

10:01

ready to paste a paper. And look.

10:05

Oh, it got a little funky right here.

10:07

Oh, it pulled in the sidebar. Okay,

10:08

there is a sidebar feature. That's

10:10

funny. Um, but it pulls in the whole

10:12

product and it's like very it's it's

10:14

spot-on. Um, when I've looked at this in

10:16

the past, I've done a few different

10:17

ones. Like, it's spot on. And look, I

10:19

can just go in. I really like this. It's

10:21

super cool. I can do a bunch of things

10:22

with this. Um, but I will say if I'm

10:24

actually working on a product, I like to

10:26

actually tell it to build it from my

10:28

codebase because then I'm just building

10:30

those tokens and that styling and

10:31

everything I want. So, this is something

10:32

fun for exploration. Um, but if I am

10:36

designing an actual product or

10:38

rethinking it, I want to just pull from

10:39

directly from the code if I have repo

10:41

access. So, that's kind of like the

10:42

difference there. But this is really

10:43

cool and great for experimentation. Um,

10:46

the other things I can do is they have

10:49

like all these containers and then I

10:51

also can do auto layout. So they call it

10:54

flex here and it's basically the same

10:56

command as in Figma or ship day. And

11:00

then basically I can use my typical the

11:05

ones that I defined with my tokens and I

11:07

can read adjust the layout here just

11:10

like you can in Figma which is cool. Um

11:13

so there's that as well. Let's go back.

11:15

Okay, it's still kind of thinking. Let's

11:17

go down. I think it's here. [snorts]

11:18

There we go. It's pulling all the states

11:20

out which is great. So let's let it

11:22

finish up there. Um but these are just

11:24

some other cool things that that I've

11:25

been doing and playing around with paper

11:26

which are really cool. And if you guys

11:27

want access to more tools, we have a

11:29

community. And if you join the

11:31

community, you get access to all of the

11:34

tools and the tips and the resources

11:36

directly from us. It's a Slack group, so

11:38

we're always on there. Me and my team

11:40

sharing our knowledge, getting feedback,

11:42

and helping anyone that's asking

11:43

questions. We also do monthly workshops.

11:45

We have one coming up soon. So, please

11:47

take a look at the description, and if

11:49

you're interested, we would love to have

11:50

you. Okay, so it looks like it is done.

11:53

So, let's go in. Let me tell you a

11:55

little bit about this as well so you

11:57

guys have some context here. So this is

11:59

a current product that we're building.

12:01

Basically what you can do is it is a way

12:03

to leave feedback and comments and

12:05

prototypes. So anything that you spin

12:08

up, you know how in Figma you have

12:10

comments? Um basically it's this idea of

12:12

being able to add comments but to actual

12:15

working prototypes. So you can go to you

12:17

can download the plugin and then

12:18

basically what you can do is you can

12:19

come in and you can leave feedback.

12:21

Anyone on your team can leave feedback.

12:24

make this yellow. Um, and then it's

12:27

going to go directly to the person like

12:29

the designer and the designer actually

12:31

can see all of the feedback and they can

12:33

actually use their agent to

12:36

fix and update based on the changes. Um,

12:39

so more to come. I'll share more in

12:40

another video, but this is basically

12:42

what we're building. And so if I go back

12:44

to

12:45

here, this is see what I did here. This

12:48

is basically the state that I'm I'm

12:49

rethinking. What I love about paper is I

12:52

have all these states here. Um, so yes,

12:54

I could if there are components, I could

12:56

have them in story book, but I'm a

12:57

designer, right? So I want to like get

12:58

into the dirty details here so I can see

13:00

everything. And it really is like

13:02

perfect. It's like pixel perfect. They

13:04

really nail it. I think like where I I

13:06

want to dive deeper and feedback we've

13:08

been getting from users is on like how

13:10

we explain what it is.

13:13

Make this yellow like what is this? when

13:16

I leave feedback screenshot. Um I the

13:19

screenshot feels confusing to me in this

13:22

visual. So what I want to do is I'm

13:26

gonna copy this. Um and I'm going to go

13:29

into paper. I want to rethink how we

13:32

actually show the screenshot. And do we

13:36

actually need a design like a visual of

13:41

it? I feel like it takes away from the

13:44

actual idea of it being a screenshot.

13:46

Um, it's something that we've been

13:47

getting feedback on and that users

13:49

sometimes struggle with. So, I want to

13:51

rethink this a little bit. Can you first

13:53

take a look at this component and then

13:58

can you go use the mob and MCP and can

14:00

you pull some inspo and put it directly

14:03

into this file for ideas of how to

14:06

display when you are making edits to um

14:10

a specific part of the platform and how

14:11

to actually show what that screenshot

14:15

section should be.

14:17

Okay, so let me hit send and then I'm

14:19

gonna explain a little bit more. So this

14:20

is kind of like how I like to work is I

14:23

think that a big part of being a

14:25

designer today is not just trusting the

14:28

AI to make the decisions. So I still

14:32

want to do my process. Um so I still do

14:34

want to go use I use the mob and MCP

14:37

which is essentially it goes into mobin

14:39

and it's actually based on the context I

14:42

give it and what we're actually

14:44

building. It's going to pull some

14:45

examples of how we could potentially

14:47

tackle this. So, I still love to pull

14:48

info and I still want to figure out like

14:50

how as a product designer, I can rethink

14:53

this in the best way possible. So, it's

14:55

trying to go directly into creating

14:58

those specific variants, but I don't

15:00

want it to do that yet, right? Like we

15:01

just talked about, I first just want to

15:02

pull info. I'm the one that's conducting

15:04

this and I want to make sure that I'm

15:06

making the best decisions. Okay, it's

15:08

starting to build it. Oh, and another

15:09

little fun thing is if you doubleclick

15:12

on

15:13

this, it goes directly to that frame.

15:16

So in any aspect so if I'm here say I'm

15:19

here and I'm like oh I want to go in

15:22

boop and then I'm directly in here which

15:23

is a defend the element. Okay so looks

15:25

like it's finishing up. Let's take a

15:26

look. So examples where the screenshot

15:29

reads as an attached object. Okay so

15:32

this is where it's literally blowing it

15:34

up as like making a comment.

15:38

I don't understand WhatsApp.

15:44

Telegram selected image before you send

15:49

then claude prompt input with an

15:50

attachment.

15:52

Oh, so just these don't feel as

15:55

relevant. Let's look at this examples

15:56

where the screenshot anchors a specific

15:58

task,

15:59

right? So this is basically it's just a

16:02

normal

16:06

comment.

16:12

So, it just attaches it as a or it has

16:14

the PDF and you can make the change

16:16

there. Some of these aren't as relevant.

16:20

Wasn't great. I think this one makes the

16:22

most sense. I

16:25

I don't like the inspo. Um, but going

16:28

back to this, I don't think it needs

16:31

this visual

16:34

just now, right? Um,

16:39

okay. Let's go back and actually look at

16:40

the product again.

16:44

So, it says just now and then it shows

16:46

the location of it and then it says a

16:48

screenshot. Then I can send it, right?

16:51

Okay. So, it does kind of show this

16:53

here, which is a little nice, right? So,

16:56

that makes sense.

16:58

This is just showing that it's text. So,

17:00

let's ask it to come up with a couple

17:01

options. I'm going to show these. I do

17:04

like how it's showing that I'm leaving a

17:06

comment directly on the button icon. Um,

17:11

I think that is visually nice. I think

17:13

some of the other screenshots didn't

17:14

feel this way when I did it. So, I think

17:16

there might not be a common pattern. Can

17:18

you look at that? And then also, I think

17:20

what I don't like is when it's just

17:21

text, it has a CR. So, what if we come

17:24

up with a better pattern where it does

17:27

show more of the element when we take

17:29

the screenshot so we can reference it?

17:31

And then if it is text that we're

17:34

highlighting and making giving feedback

17:37

on, can we make it so that it's a more

17:41

visual fun way of showing that it's text

17:43

instead of just the CR, which I don't

17:44

think is very clear. Can you design

17:46

these options in paper? So, what's great

17:48

about paper is instead of it building it

17:51

out in the product and me um starting

17:54

the server and going back and forth, I

17:56

can just see a few different options

17:58

directly in paper, which is really the

18:01

value. So, you can see I'm not

18:02

necessarily like cushing pixels. I'm

18:03

more like chatting with AI and then the

18:06

AI is building it for me, but I still

18:09

have like those full design

18:10

capabilities. So, let's see what it

18:12

comes up with. Okay, so this is

18:14

currently what I have. Current behavior

18:21

context crop. Replace initials with a

18:24

visible selected text moment. Context

18:28

crop. Capture the surrounding sentence

18:30

and highlight the exact selection.

18:33

But it didn't pull it into any kind of

18:35

design that we have. That's kind of

18:37

strange. Um, but okay. Let's see this

18:40

one. Small screenshot behind selected

18:42

text. Enlarge in front line strip. A

18:45

compact crop for text and dense layouts

18:48

or long pages.

18:51

I do like this markup overlay idea.

18:57

So then it does capture the component.

19:01

Quote hybrid. Oh, I like six quote

19:04

hybrid. Let's So these are kind of I

19:07

think very rough ideas. This is great.

19:09

Like let's do this one. Let's go with

19:12

six.

19:13

I like option six. Can you actually

19:15

design it out how it would look in the

19:18

actual card? Oh, interesting. Okay,

19:22

now I think we can get into some design.

19:25

So, this is how it currently looks. I'm

19:26

going to pull it here. And it's not

19:29

really I don't really like I didn't

19:30

understand this is exactly what it

19:32

meant, but let's do this. I kind of like

19:34

the idea of it.

19:40

here.

19:44

Okay, I'm working in this frame.

19:51

It feels like a very look and feel now

19:53

that I see it in here. I still want to

19:55

pull from the look and feel that the

19:58

screenshot with the X and the compon the

20:01

thing above has, but I like this idea of

20:04

saying highlighted copy. So maybe you

20:07

can um do another version where it feels

20:09

more similar to what we had in the past

20:12

but with this copy highlighted instead.

20:16

No, not really. But here, let's play

20:18

around with it. So I think what it

20:20

should say is

20:22

Okay, I like this a little bit more. Oh,

20:25

I see there's some weird padding here.

20:27

So let's go center. Oh, it's still

20:29

working on it, too. So now I'm making

20:31

changes.

20:33

Let's stop here. Okay. Um, oh, it just

20:37

brought it back. Okay, so this is how it

20:38

currently looks, right? Um, and kind of

20:41

based on this ex idea, I think that this

20:45

background could get rid of this as

20:48

well. And then I think it should just be

20:51

here. Hold on. Let me look back. Okay,

20:53

so I think that it could say

20:58

pull this one

21:01

instead of this. I like this idea of it

21:03

saying like highlighted copy and then

21:07

hold on, let's pull,

21:10

can you recreate this in the same

21:12

artboard so I can play around with this

21:14

one?

21:17

Okay, what I wanted to pull was this

21:19

idea. It is full screen. So maybe this

21:23

idea here.

21:28

So what's really cool? So I can get

21:29

something rough, right? like it doesn't

21:31

have to be perfect but I think this idea

21:34

of highlighted copy so let me do this

21:40

let me copy this

21:42

so this is the concept I am actually

21:45

thinking for if there's highlighted copy

21:48

and you want to make changes to

21:49

highlighted copy

21:52

and I this is how I want it to visually

21:53

show I need you to redesign it so that

21:56

it is matching the constraints of the

21:59

padding

22:00

and the right sizing so that it makes

22:03

sense in engineering as a component and

22:06

then can you also can you redesign it or

22:09

can you take it and design it so it

22:11

follows all of these rules so I don't

22:12

have to push pixels so look this is

22:15

super cool it's adding the fixed avatar

22:17

then fixed action button width so it can

22:20

only it's fixed it's rebuildable okay so

22:23

here's the it's almost finished

22:25

redesigning

22:27

oh but it made these too big look at It

22:30

got carried away. Okay. So, let's fix

22:33

some of this. I think maybe because I

22:35

made this larger. So, let's tell it to

22:38

rethink the size. Wait, you need to

22:40

rethink the sizing of the whole

22:42

component. The send, cancel, the avatar,

22:46

the just now, and the slash landing.

22:49

This should all be the same size that we

22:51

defined already. So, I think everything

22:52

needs to be smaller to adhere to the

22:55

current constraints that the product is.

22:57

So as a designer, I'm not pushing

23:00

pixels, but I am kind of like reviewing

23:02

and orchestrating giving feedback to the

23:04

AI. So that's really like instead of me

23:06

spending all that time, I'm shortening

23:09

the amount of time because I'm just

23:11

coming up with an idea quickly and then

23:13

I'm asking the AI to keep fixing it. So

23:15

you can kind of think of the AI as like

23:17

your

23:20

assistant, your design assistant. And

23:22

that's really cool. And it's I'm able to

23:24

do this in such a better way in paper

23:28

um than I have been able to do in any

23:30

other design program. So I still still

23:31

feel like I am the designer and I'm able

23:33

to design and push the pixels when

23:37

needed and come up with the concepts

23:38

because I am visual. So it's not just

23:40

like chatting with an AI actually to

23:42

visually see it and then tweak it. But

23:44

also I am able to

23:48

design it out a little bit more back and

23:50

forth. So, what's great is like the AI

23:52

isn't perfect as you can see here. Um,

23:54

but it is helping me understand like the

23:57

concept in my head and actually build it

23:59

out directly with AI. Okay, so it's

24:01

getting close, but look, it's cutting

24:02

this off again. So, I'm actually just

24:03

going to go on and try to tweak it

24:04

myself instead of asking the AI. Okay,

24:07

so it's definitely much more aligned.

24:10

The copy is cut off. I don't need to ask

24:12

the AI to fix it. I'm actually going

24:14

First of all, I don't like this

24:15

underlying. I think it's something extra

24:16

that's not needed. So, I'm going to

24:17

delete it. Oh, there we go. Okay,

24:19

perfect. Um, cool. I like this. I think

24:22

it's super clear. Let's just go in.

24:26

Okay. It still feels small to me, but

24:29

and I don't think we've defined this.

24:30

So, I think I need to define body is 15.

24:33

Okay. Okay. I tweaked it slightly. Can

24:36

you look at the tweaked that I made,

24:38

which was removing the line under

24:41

try at all? And my other question is the

24:44

highlight copy. We don't have this text

24:48

size yet. Do you think this is something

24:50

we need to incorporate into the product?

24:52

Okay. I don't know if I agree with this.

24:54

It says basically just to make it a

24:57

micro label for this component only. Um

25:02

I might want to just remove it. But

25:05

let's see. Okay, let's build this out. I

25:08

want to actually see it when I am making

25:14

copy changes in the actual product. So

25:16

build it out now as this new component

25:18

for anything copy related.

25:22

Okay. So now we've agreed upon it. Um I

25:24

feel like it could make sense. I still

25:26

have some questions about the

25:27

highlighted copy, but I'm actually

25:28

telling it to create it in the codebase.

25:33

Okay. So what is really really awesome

25:37

about what it's done here is it's taken

25:39

this component that we redesigned and

25:41

it's looked at the codebase and the

25:44

current component and it's made sure to

25:46

create it in a way that it is an

25:48

existing component pulling from the

25:50

design. So it did all of the front-end

25:52

work, right? So you know all the

25:54

inconsistencies that used to happen when

25:55

you hand off designs to dev well this is

25:58

solving all those inconsistencies. Now,

26:00

it's also,

26:02

let's see, it's still working on some

26:04

things,

26:07

but it did. Okay, it should have um a

26:10

place for us to test, which will be

26:11

coming in a second. Okay, so let's test

26:15

it out. Huh. So, we're going to drop a

26:17

comment

26:19

on here.

26:24

Cool. Look. So, it took my highlight.

26:26

Copy. Highlight. Oh, there's something a

26:28

little funky here. So, we should fix it.

26:32

So, hold on. Let's do cancel.

26:37

Change to

26:40

design.

26:43

I want to see what it looks like. Okay.

26:45

So, I think it all looks good except for

26:48

that it's cut off

26:50

a little bit.

26:53

Oh, it showed up here nicely, though.

26:55

Okay. I I like this. I think it looks

26:58

great. The highlighted copy is a little

27:00

small. I probably would still want to

27:01

update it. Um, but really where you can

27:04

see it went a little rogue

27:11

was here. So, I can actually tell it to

27:14

go in or I can just tell it here.

27:19

When I

27:21

copy the text to make the comment, you

27:23

can see that the white box is a little

27:27

it's going into the right. It's cut off

27:30

on the right. We need to resize it

27:32

appropriately.

27:34

But do you see how cool that is? There

27:37

is paper. I think I walked through quite

27:39

a few things we can do with paper,

27:41

right? um from how we can get started

27:44

with the project, the tokens, um how we

27:46

could actually take a screenshot and put

27:48

it directly in, how we could transfer

27:50

Figma in, how we can do auto layout or

27:53

how they call it flex. Um and then I

27:55

showed you details of how I actually

27:57

work where I am pulling inspo and then

28:00

I'm tweaking the design and I'm asking

28:02

code to fix the design based on best

28:04

practices and following formatting and

28:06

then I'm tweaking again and then I'm

28:08

actually saying, "Okay, now build this

28:09

out in code." and then it actually

28:10

changes the whole component and then I

28:13

can ship it to the engineering and the

28:15

product team and that's the whole

28:17

workflow. So paper is a really really

28:19

cool tool. Um download it, explore it,

28:21

test it out, let me know what you think.

28:22

And if you are trying to figure out how

28:24

to add AI into your product process um

28:27

basically from from start to finish or

28:30

the design phase, I would love to hear a

28:31

little bit more about how I could

28:32

potentially help you. We offer

28:34

workshops, we help build design systems,

28:36

we're doing a lot of things. There is a

28:38

link in

Interactive Summary

The video provides an in-depth walkthrough of using Paper, an AI-powered design tool that bridges the gap between design and development. The creator demonstrates how to orchestrate an AI agent to extract design systems from existing codebases, create tokens, and iterate on UI components. Unlike traditional tools like Figma, Paper allows for direct, code-aligned design iterations where the AI assists in building components that reflect actual product constraints. The workflow moves from conceptualizing designs with AI assistance to finalizing components and pushing them directly into the codebase, effectively streamlining the design-to-development handoff process.

Suggested questions

4 ready-made prompts