Y Combinator Startup Podcast - How To Design In The Agent Era

Episode Date: August 7, 2026

AI isn't just changing the tools designers use. It's changing how they build, ship, and stand out. In this episode of Design Review, Stephen Haney, founder of AI-native design tool Paper, joins YC General Partner Aaron Epstein to demo the agent-first workflow that's making Paper one of the fastest-growing design tools since Figma. Using live redesigns of user-submitted websites as examples, they break down the most common AI design tells, show how to fix them in seconds, and explain why the biggest risk for founders isn't moving too slow, it's shipping something that looks like everyone else.

The agents can speed you up. They can save you a lot of work, and they can do tasks like translation or resizing for you. But we're a big believers in the human element of design and how important that is. And so we're trying to create a tool that lets humans work really fast with agents. It's probably very tempting to use whatever cloud design spits out. I don't think it helps you stand out from the crowd. And so I think design is this great differentiator. And if you look at every great company of the last 10, 20 years, basically all of them have exceptional design.

And so I think if you want to be one of those really great companies, you have to put the carry in on design and make sure that you stand out. Today, I'm excited to welcome Stephen Haney, the founder of Paper, an AI-Native design tool that's been taking the design world by storm. We're already using paper a ton here at YC for all of the design work that we're doing internally, and I'm excited for him to show off how the product works to everybody out there. So, Steven, thank you so much for joining. Yeah, I'm honored to be here. I've watched the show a lot, and so it's really nice to be on.

Thank you for having me. Maybe to start, tell people what is paper? Yeah, well, paper is a design tool. It's an agent-native design tool. And what that means is we use HTML and CSS as the actual rendering engine. And that makes it very easy for agents to use because they understand HTML and CSS really well. And historically, a lot of design tools have custom rendering engines. And those are a lot harder for agents to use.

And so you'll see higher token spend, slower times, lower accuracy. So what we're building with paper is basically it's a group of people, is about 12 of us now. all designers and engineers, we basically started this out of love for design. And now we're just trying to build the best design tool possible for this next era. Yeah, it's crazy to think you're only 12 people because I hear about you all the time from everybody who's using you. And especially from so many founders that are just picking the design tool that they want to build and live in all of the time.

And doing that from the ground floor, being AI native themselves, they're looking for the AI native tool that is going to work best. So maybe talk about what was your inspiration for building paper and why were some of the existing tools that are out there not good enough for you? It really was born from a love of design, as I said. I started as a solo founder, and it was really the side project that just kept going. I had months where I was like, is anyone going to care about paper? And then it was like, oh, yeah, people do care.

You know, this is great. It's really taking off. It was the thing I was going to do anyway. It was very much the thing I was going to do anyway, which I really believe starting a startup, you should do that. You should do the thing you're going to do anyway. You should do the thing where you have founder market fit. And for me, this is where I had the strongest founder market fit,

because I love building tools for creative professionals. It's something I'm going to wake up every day and I'm not going to have a bad day building creative tooling. So that came first, even before the market or competitors or any of that kind of thing. And then looking around, it was like, well, gosh, I mean, we're big fans of Figma.

You almost can't talk about design without saying, you know, Figma these days. They've done a lot for the design industry. They have a huge company now, and they're doing very well with enterprise companies. And we were thinking, well, maybe there's time for a new tool to focus on the designer and just really spend a lot of time.

There's so much change with AI, with agents, can we build something that's agent-native, building on HTML and CSS, and just build this next-gen tool? And so that's kind of what we're going after. And so far, so good. Yeah. And what does that mean to be an AI-native, agent-native design tool? It's really about the fundamental technology. So using HTML and CSS as the rendering engine, we actually did it to help with designer developer handoff. We were like, hey, if the designer can have this constraint-free system where it feels like a figma or a sketch, but by the way, it's HTML. They don't have to know that. That's great. The handoff will be better. It turns out,

if you build a great handoff for humans, you've also built a great handoff for agents because agents understand HTML and CSS, and it's in their training data. And so when you ask an agent to work with a paper file, it's literally just reading the CSS out. And agents know exactly what to do with it. You try to basically extend the agent with a visual interface. It's sometimes that a paper is a design tool, but it's also this visual interface for agents. We see sales, folks, engineers, all kinds of people are using it, because prompting isn't always the best input mechanism for agents. Sometimes you want to drag and draw, and it's just faster to be able to do direct manipulation. Yeah, it's interesting because as it becomes easier and easier to build software,

the figuring out what to build and getting inspiration for how to build it seems to be one of the trickiest pieces right now and one of the biggest bottlenecks. How do you think about using paper to be able to solve that piece of the creative workflow? We see there's more software than ever being created and engineering teams are moving so fast now. And I don't know how you feel about this, but I haven't felt like the software is getting better necessarily. I haven't had a great software experience in the last six months blew me away. And so it's important for us. Like I grew up, I'm like a fan of software. Like I love building great software. The best compliment you can give me is that paper is really great software. And so I want more of that for the world. And so I want to

help designers and people with taste keep up, you know, with all of the software that's being created. And so giving them a tool that can move at the speed of prompting. of the agents, you know, rather than being like just pulled into an IDE or just pulled into prototyping, but being able to still do the traditional design work at a faster pace, I think it's really, really important. I'd love to see a demo. Can you show us how it works? Yeah, let's jump in.

So it's a very familiar interface if you've used, you know, Canvas tools. Sometimes I like to start with the visual things. So because we are using basically the browser to render, we can also render shaders really easily. And shaders are basically webjubljointed. g.l animations that we build. This is called paper shaders. It's a library we put together. A fully open source, feel free to use it for whatever you want to do.

And the shaders are just so fun. And brand designers love these because you can get all kinds of different looks. This is just one shader with different settings. And so you can do all kinds of stuff. I know why I see you guys have been using. We've been using these all the time. And I think, you know, for us and I imagine a lot of people out there, this is probably their first introduction to using paper.

This is something that it feels very modern, cutting edge, that would have been really difficult to do a year ago. And then you guys have made this really simple and native to the product to the point where it's showing up everywhere because it's so easy to do it now, right? That was part of our hope was can we make this more accessible to designers? Because if you're a designer and you're living in a traditional design tool all day, you want to create a shader effect. You want to describe how an AI loading state is working. And that's what I kept hearing was I need to make a design. like loading states for AI and how do I do it? And so we just thought it'd be nice to give people these animated effects that they can customize

and not just use the out-of-the-box ones, but you could get in here and like tweak things and make them make them your own. And you can do all kinds of cool stuff. You can do blending modes, of course, to like create combinations. And you can also, these have been really popular, is some of these more subtle effects. This is a half-tone dots. And you can pull other things into the effect. So this is pulling the, what we call neurone noise.

into the half-tone dots and you can try different things and get different effects. And we spend so much time. We actually studied literal paper to study the like the randomness of, you know, actual physical effects and we try to model that in code. And so I think that's one of the reasons our Shader library stands out is we go to ridiculous degrees to make it to make it as good as we possibly can. And so brand designers really love this. So like, you know, one example is quarter.

I don't know if you guys know the quarter posters. they are really, really cool posters that this team makes that they do, I think it's like earning reports. And they've transitioned a lot of these to be built in paper now because they can get this cool retro, you know, vibe that really fast. So I love this team. The work they're doing is really cool. And I'm like honored that they're using paper to achieve some of these posters.

Yeah, it's really cool. We've used it for we have startup school. And so Ev, who leads design here at YC, has actually created custom animated shaders, with the YC branding, with people's names on them, tickets that are, you know, you can play with and animate when people get accepted and send it to them. And it's a great experience for the person receiving it. It feels very thoughtful and intentional and highly designed.

And it's also created something that's very shareable that people want to share out, which helps to spread the word also, all because of the design and because of the shaders. Really fast way to elevate the visuals and help it be a little more special. So shaders are very popular with the brand designers. We also do, this is kind of like a lesser known thing to paper.

We actually have a really great image generation library where you can use all kinds of models all at once. And so, I don't know, we'll just do a thoughtful man staring at the moon illustration. You can imagine for a brand piece or whatever you're working on. And so this will use four different models at once and you can quickly explore the possibility space of whatever kind of illustration you need to do. That's very cool. Yeah. So I see you have it set to variety pack there. Is that the thing that gives you the four at once? You can pick a certain model if you want to. They all have different, strengths and weaknesses, or you can do the variety pack and get, we kind of rotate which models are in there just as we, new models are always coming out.

Yep. And it just lets you quickly explore and you can provide inputs into it. You can, let's like, you know, I'll copy this shader as an image and then you can feed the image into it and make this more colorful, please. So you can do all kinds of different workflows combining these, you know, things to create really cool graphics, really, really fast. Along the same lines, you can also vectorize. you can do this, extract colors is really, well, let's do this one.

So the agent made a really colorful thing and I can extract the colors and it just pulls out the colors. And so I can explore ideas very rapidly this way and generate new textures, things like this. So this is kind of the brand side of paper. This is the graphics. And what we find is like all designers need to make brand assets. They need to make, you know, social media posts or whatever it is. Which is interesting also because, you know, I think a lot of people use and think of Figma's product design. and it certainly can do a lot of brand design element,

but didn't start as much that way. And it seems like you can do product design and there's all these really cool tools to do a lot of brand design work too. Yes. Graphic design, which traditionally I think has been more difficult with a lot of these tools. Yeah, it's part of our values.

We love both, you know. And so I think the product design is very interesting, but we'll always have the brand design and the graphics too. To be honest, the way we think about the shaders is we spend our, this is our marketing budget. We don't think of this as product development. We think of it, instead of buying ads,

let's make this useful library for the world and spend that effort and that money on building the shader library. We had a lot of attention. It works really well. Similarly, like, we develop a font, paper mono is a new font that we're launching. I'm really proud of.

It's really, really good. And we found there's no mono font that is just kind of an all-rounder that you can use in marketing, that you can use in code. Most of them are built for the terminal. And so we wanted one that you can use as part of your brand.

So we made Paper Mono. It's launching a couple weeks. That's marketing for us. Instead of buying an ad, let's build a font. Better for the world, right? So Eve at YC was in New York City at a dinner the other night, and she ran into someone wearing a Paper Mono shirt, which I thought was so cool. There's only, I don't know, 100 of these in the world. Yeah, how do you get one of this? Yeah, well, we can get you one. We'll see it one over. We want to set up a merch shop, you know, paper merch see. But this is Paper Mono, and we have a great mini site launching soon. We spent a lot of time on the legibility of numbers because that's an area that a lot of

fonts really struggle with. We just think this very, very legible, very usable monof font. So I'm really excited about this, right? It's so fun we get to do these things as, you know, just kind of part of our marketing, basically. You mentioned that the product started to help solve the design to dev handoff. Show us how you're able to do that in the product. Well, yeah, so it started as, hey, this is code. This is literally a React component. Why don't we let you copy it as a React component? And so this shader that we're looking at right now, you can and right click and copy as React. And when you paste it, you get the literal code

that we're looking at. And so this was just a really fast way of like, hey, the designers made this really cool animation. They've dialed it in perfectly. They're the designer, right? And let's let them copy that out and ship it. And so we've seen these take off all over the internet, people shipping these React components.

But it goes further. So anything that you're looking at in paper is code. This frame with this box inside of it is code. And it's basically a website already. And so you can copy it as Tailwind, you can copy it as React and just grab out the code. That's where we started. And that was pretty good.

We already saw some people using it. And then in what was this, last December, 2025, the cloud code explosion happened. And I didn't think designers were going to be in terminals. I never thought that would be a thing we'd see. But we clearly did. And so that was the moment where we were like, gosh, we got to get onto the desktop. At that point, we didn't have a desktop app. So it was like, gosh, we got to get onto the desktop.

That's where things are happening now. Local repos, designers have repos. They're going to GitHub. let's be part of this explosion. The way we see it is kind of the new agent stack. And what I mean by that is you'll see, you know, I have cursor, I have cloud code, you have multiple agents, you have paper as a visual way to communicate with the agents. And then maybe you have a GitHub or, you know, the code review. That's why we see builders using now.

I mean, is that what you see too, like the newest YC batches? Yeah, that's a lot of what we see. And we even see a lot of that getting rolled into single apps or friends at Conductor. you can push code and merge code and do everything that you need directly. Yeah, there we go. Yeah, I have Conductor up too. Directly from the app. And so more and more it's kind of being consolidated in apps rather than having multiple

discrete apps as part of the stack too. That's what we see too. And Conductor and Paper are this like pairing right now. Yes. A lot of people in the community are using both together. I think it's so cool. I love Conductor. It's great product.

But we saw this cloud code happening. Let's get on the desktop. We build an MTP server into our desktop app. And I'll give you a demo of some of the stuff it can do. Let's use, we'll use conductor for this. Why not? I'll just say, hey, conductor in paper, please make a small test frame saying hello to YC. And I'll show you what it's doing. We'll get into more complex cases, but just to show you how the basic technology works,

as the agent uses the MTP of paper, which we'll put on the right half of the screen, we give it tool calls to like, hey, which fonts are available? What's the basic setup of this page? And then it literally runs tools like write HTML. This is literally the, you know, it's in JSON format. But it's literally sending us HTML. And the paper engine is just able to take HTML and knows what to do with it. There's no translation there.

The agent writes HTML, paper can render HTML, and you don't have to like switch back and forth between formats. What that does is it saves you tokens. it makes it run faster, and you just get less weird hallucinations. And this is really the moment we took off, because people are doing design to code, or they're doing code to design. I'll give you another example in cursor. This is our actual paper code base, and I'll just say, hey, can you put at colors.cass

onto the paper canvas as a style guide. So this is our actual CSS file from paper, the code-based paper, not the product. And cursor is going to take that and then render it onto the canvas. And so we see a lot of people doing, you know, auditing of their code base, what exists. You can do component architectures, ask for diagrams. And it really just, it brings the design tool into the rest of the stack. We're so used to the design tool being over somewhere else, right? And having this wall between it.

And this is like the design tool is now part of the entire stack. Yeah. It's kind of the goal here. Yeah. I mean, one of the things that's always been broken with that handoff like you're talking about is designers would design a thing. And then they give it to the developers. and then maybe it wouldn't actually be built and shipped

in the way the designer originally intended when they were working in the canvas. Or even if it was, things would just get out of sync and the code would evolve and it would never get wrapped back into the design. And so it's what is the source of truth? And what's really nice about this is the code is the source of truth and both systems know how to work on that same code base.

Exactly. Yeah, we see so many teams maintain two, two copies of their design system. And it's not a tractable problem to keep them in sync. It's impossible, really. And so you have a lot of maintenance costs with keeping both updated, and you just have accuracy problems. So what we see from the newer companies, the more agentic builders,

is they are treating the code base as a source of truth. And, you know, we don't have components in paper yet. It's this common complaint. Like, people want components really a lot, and we're working on that. But you don't need them because your code base is a source of truth, and you can kind of like summon things in as you need them. And we took that a step farther too. We build a Chrome extension where you can grab your live site and copy that into the paper canvas,

which is a really nice way to start from what your users are seeing and build on top of it and iterate. Maybe that's a good segue. Should we review some user submitted sites? I would love to. I would love to use that feature to be able to pull some of them into paper and make some tweaks on it and just see what we can live edit to try to improve some of the sites. I would love to. It's always my favorite section of the show. Okay. Awesome. Let's do it. So the first one we got here is Legion Health,

psychiatry as low as $0 with insurance, online medication management for ADHD, depression, anxiety, bipolar disorder, and more. I don't know. When I first look at this, I see the little pill up there at the top and this kind of, it makes me think this was just like a vibe-coded thing because it matches so much of the style that we see that the agents are just spitting out. There's definitely like a fit and finish thing here. I think little things, like how these sections kind of like leave a little gap. The pill is very generic. I think overall, though, I can understand the value problem very clearly, which I think is great.

You know, that's a big part of design too, is like, can you comprehend what's going on? It shows the benefit over on the right of like being. Being very happy in a field or something, I guess. Yeah, you're having a great life. So I think like those parts are working. I think it's more about just tightening up here again, you know, tightening up some of the design would really help the brand feel more serious, more elevated, a little more trustworthy, I think.

This feels, you know, I don't know if I trust this brand right now. And this is an area where you have to have trust, right? Yeah, so what is the thing when you look at this that makes you not trust it? Some of the design elements is like the alignment of things, the contrast choices, they feel maybe vibe coded or like somebody didn't spend a lot of attention to this. And then your mind goes to what else don't they spend attention. The one other thing is like, this must be a YC company, but I don't see that here.

And I would say that would be a great thing to call out as a little bit of social proof or evidence, but like, this is nice. I mean, there's a lot to like here. I would just, I would work on the fit and finish a little bit. I'll show you paper snapshot. Paper snapshot is a way to grab content from your live site. This replaces taking screenshots of things. So we'll just grab this section because I think this is a cool one to work on.

And we'll paste it into paper. And these are all live sections, right? So this is not a screenshot. This is actual stuff that you can come in here and edit and you can like alt drag and get another copy of it and all these things. So let's ask the agent to work on this a little bit. And one of the things I like to do is actually just ask it for a bunch of variations. So we'll use, let's a conductor. Hey, conductor.

This is using GPT 5.5. Can you make three new variations of my selection in paper focusing on the craft and alignment and contrast choices to elevate the visuals? You can do a lot of prompting, but I often find like simple prompting is fine, honestly. What's really powerful about this, a couple of things. One, I love this concept of just spitting out a bunch of different variations. It's the best way to get inspiration.

And sometimes, you know, you have a sense of what you want, but you don't really know. And this is a great way to look at a bunch of things and go, I like that from this one. I like this from this one. And combine it and make so much more progress, so much faster rather than being stuck in that creative rut. And then the other thing is really cool is you're able to just say, hey, the thing that is my selection on the canvas and paper, and it knows what that is, and now it's able to go to work on that, which is really powerful.

I think I had the wrong selection, though. I grabbed the YC card. Let me restart it there. One more time using my selection. Oh, I see what happens. It changed it with the ADHD content. Oh, actually, you know what? This is a good way to show another thing we can do.

So we can leave a comment on this frame and say, please make three new various. of this content, explore the layout for visual impact. And instead, what I'll do, so I left a comment in paper. And I'll switch over to the agent now and say, please address my comments in paper. There's only one. And this will be another way that you can kind of like target. And so you can imagine having a big design and leaving all of your feedback

and then just setting your agent on it, letting it spin, you know. But we see a lot of folks are leaving these running like overnight. They're setting up loops of an agent, generating hundreds of variations. And it's actually a challenge for us to be like agent scale with our servers and everything. They're doing a lot more work than I thought they were going to be doing. And it's so cool. And now they come in and they curate.

You know, they come in, oh, that's a good idea. That's a good idea. Let me combine these things together. And it's kind of this new curation design process. I think is really interesting. It's so cool to think about this world where you just go in, you leave a bunch of comments on the work. You go to bed.

You wake up in the morning. And they're all either addressed or lots of iterations for you to choose from are just sitting there waiting for you. Exactly. I think it's really cool. It's a new world. I find that it'll do things that I know, but I wouldn't have thought of because I was too busy. So it kind of like, it lets me scale myself a little bit and have more time in my day because I can just set it on these tasks that are maybe more boilerplate or I just don't have time to do four things at once. And so let me set the agents on it to go address it. So it's, this is 5.5. It is working through ideas now. And you'll see it iterated. It can take screenshots. So it'll notice things like

that alignment isn't, it's running right into each other. It'll fix itself as it goes through. and yeah that's cool okay so it's actually spit back a variation here but the alignment's a little screwed up there and it's still working so it's not done once it gives you something it's it's an iterative thing that's happening here right it checks its work yeah and it can do agents are getting better at this too all the time for instance i can see like the first shot it ran the text over the cards it'll usually catch that and correct it or if it doesn't you can just you know tell it hey correct the text overflow um me do this Yes, we'll set to fit.

And we'll pop back in Conductor. Please address any text overflow issues. The other thing I noticed, too, is the models are always getting better, and they have different strengths and weaknesses. And so I often use fast models for demos like this, and they tend to make more mistakes. And so, you know, if you're doing this for real, you're probably using something that's right at the cutting edge,

and those deliver better performance. Yeah, what are your go-to models these days? So Fable is a wonderful, you know, I think that's probably the best, visual taste right now. It's very slow. That's why I don't use it for demos. I use cursor's a composer 2.5. Actually a lot. It's a very good model for the speed. Quality is very good, very precise. 5.5 is still usually a pretty good balance. So I jump around. I think if I need pure taste, it's the fable and the anthropic models in general. If I'm going for precision or I have

like, hey, I have a hundred art boards and I need a task to like go down the entire list, I might reach for the composer. It just does a really precise job. I find cursor is a really good harness. It's not too lazy. It'll often spin up subagents to get jobs done. So I jump around a bit. So here it gave me kind of like three different ideas. Yeah.

And I think from here what I would usually do, spend a little more time on exactly what I actually want out of it and then give it more direction. But I might start branching. I might be like, oh, I like this. I kind of like the list of this. That's actually very easy to read. I kind of like, the card layout of this actually,

like featuring one card and having some other cards to the side is nice. So I might just start branching. I think the power of this is you can ask five agents to go generate five more of each, you know, and kind of find one that really fits what you're going for. I think one of the things that's interesting is it gives you a lot of inspiration and ideas and then you can start branching and going down a path. But it's still the models that's doing the design. And it still has some of the hallmarks of the things that you see models do frequently for better or for worse.

And so how do you, when you're designing and you've got this. inspiration, you're like, okay, this list is interesting. I hadn't really thought of doing that. Maybe I want to go with that. How do you then make it more your own and get out of kind of the standard slop that a lot of them will spit out? Yeah. I have some like very specific rules I can share with you. One is that models love bold. They love bold styles and really like try to pull your font weights back. You know, it's so tempting to bold things. Just try to pull them back as light as you can get and it'll just magically look better designed. Another thing is too many too many sizes. Models love having five, six, seven, eight different font sizes. Try to work with three. If you get

down to three, so let's just try it real quick on this one. If we bring this down to, um, uh, let's see, like a, like a 500, even that's maybe heavier than needs to be. Um, same here. Like these don't actually need to be bold. This is such a tell of a, of a, of a, this one up here, the care organized black is so heavy. Let's just pull that back. And suddenly it looks more designed already. Right. Right. It looks cleaner. It looks cleaner. It looks cleaner and more approachable black here too. Let's just pull that back. These are really simple tweaks that like you can make. The contrast here is pretty good. A lot of times model screw up contrast. Yeah. So I would, that's another step. You can just make sure it's like is like, is the text

readable? Right. These, these little two, three, four, these are a little too contrasty for my taste. I would probably pull these back. They're supposed to be a supporting. I could just delete them. I mean, those are the kind of things where the models like to add numbers or little icons or things like that there. And yeah, it's really cool that you can just go through and delete it. and it just snaps everything else up rather than leaving a blank space there. Totally, totally. Well, that's one of the powers of this tool in general

is paper is a design tool first. It's a human design tool first that happens to be really, really good with agents. And so that means you can customize whatever you need to do. You're not stuck with the track you're on. You don't prompt again. You can start reaching in and changing things. So yeah, 100% agree.

Maybe get rid of that one too. The first thing I do is try to get down to like three font sizes and just pull all that bold way back. And you'll immediately get something that feels more design. and intentional. Yeah. Yeah.

And then what do you do from here? Great. You've got the new section. You're excited about it. How do you then wrap it back in? Well, so there's a couple ways. You can right click and, let me actually grab here. You can copy it as React.

So that's kind of the long way though. So I can get this out as code and paste it into my IDE. You can also just ask the agent to pull this back. So, you know, I have the paper code base open. But you can just say, you know, please build my selection in paper into a new into a new component in the codebase, which we will not actually ship in paper. It can read everything that we just did.

And it knows my code base really well because it's the agent that I use. And so it can actually pull in this design. It'll match the coding styles that we have and the conventions that we have and build it into the code base. And if it already existed, usually you would already have this as a component.

It would just update it. And it's really, really fast. So we see a lot of people are just kind of like, this is the whole loop now. Like there isn't even a handoff anymore. It's just like one person making these changes and iterating and sharing. which is really cool.

Yeah, that's really cool. Yeah, that's what's so powerful is, you know, it's easier than ever for designers to actually build and ship things in the way that they see it. Yes. And to actually get that last mile and get something live and iterate on it and make it better. And it's tools like this that I think are giving people superpowers, which is, it's a whole new world where one person can do the work of an entire team. Oh, absolutely.

Yeah, absolutely. And I think even in teams we're seeing a lot of this too. So teams are speaking up using this, but I think the core of it is agentic workflows. If you're using agents to accomplish your work, you need tools that speak to the agents really well. And so that's what we're focusing on. YC's next batch is now taking applications. Got a startup in you, apply at Ycombinator.com slash apply.

It's never too early, and filling out the app will level up your idea. Okay, back to the video. Sure, do one more. Let's do it. Here's another one. All right, we got SITEX, accelerate field infrastructure operations, plan the process, execute in the field, and turn every action into one living knowledge base that you can monitor, automate, and just ask. Okay, I mean, right away, the first thing I noticed is the same super bold headline, just like you were talking about. We've got some of those glows, the colors, feel vibe coded, a lot of purple again.

Yeah. You know, it tends to be another tell. the animation here looks cool. It's kind of like showing how the product works. I love that. That gets my attention. Yeah. Also, I had no idea what it was until we got to the animation.

Yeah, I still am not quite sure. But yeah, it's something that looks like a canband board, I guess. Yeah. And I'm getting the sense that I can have my information in such a way that I can monitor it more easily. And then I think I would, my feedback here for design is actually make it more clear what I'm getting out of it as a user. Well, it's interesting because then, they've got this image down here, this video that's playing. It's showing construction trucks and somebody with a hard hat on working on their phone. And you go back up and it's like the field

operations or field infrastructure operations. And it's like, okay, this is something with managing operations involving construction or heavy machinery or something like that. But it's not obvious above the fold here. Because that that biggest clue right there is right below the fold. And I would love to see, too, like what is it actually? doing because I have AI enabled combine boards already. So what is what is making it special for construction? I would love to see that as as part of the value problem of. But to your point on some of the fundamentals, like very, you know, the bold is here. This looks like it may be an AI created some of it. Here's another great section. So here's one that would be cleaned up a lot

by having fewer textiles. So you see like, you know, one size, two sizes, three sizes, four sizes. That's probably another one too. If you take that down to three, it's a three. It's going to look so much better. So let's just try that real quick. I'll even ask the agent to do it for us. And again, I use paper snapshot to grab it from the live site, paste it into paper as editable layers. And then we'll just say, let me just name this one. Here's another way you can do it. You can say, what is the Sytex? This one will give to cursor. We'll say, hey cursor. Please clean up the Sytex frame using at most three font sizes and no bold, everything regular or below. low in wait.

And so you named that frame Sytex and then you just gave it that context clue to the agent. Exactly. Just different ways to kind of tell it what you want it to work on. Yeah. You know, my favorite is actually leaving comments. And the thing with the selection, I noticed earlier as I was jumping around, you tend to change your selection a lot. And I'll accidentally, like, deselect. So I love leaving comments.

I think it's just the best way to do things. But here we go. So it's starting to work on this one and it's going to align, just, you know, bring back some of that vibe-coded stuff. Now, in paper, we actually give instructions to the model not to do these common mistakes. And sometimes people go, like, what's your secret sauce? Like, how is it better? What does it look better? And it's just like, honestly, all we're doing is these basic rules of typography and contrast.

And we had the more senior designers on the team, like, distill their knowledge into the instructions. That's great. For the model. Yeah, even that is very powerful, though. Just putting those guard rails up around what the models are doing can just save a lot of time and energy. Yeah, like an expert, you know, kind of guiding the, model a little bit is really helpful. In real life, I might have just done this by hand because it would be faster, but if you had an entire site that you wanted to go through, for instance, using the agent

is really useful. Or if you wanted to pull this back into your code base when you're done, the agent can do that for you. This already looks like a huge improvement. Yeah, I think very simple changes, right? All we did was pull back the weight of the font and we consolidated how many sizes there are. To me, it looks much more designed and more intentional. And so if we go pull the original again, And like, you know, here's kind of the, we'll pull up paper and do a side by side. Now, and this is subjective, right? And I know somebody probably spent some time on this design. And I think they nailed the content.

But by pulling some of these AI tells out of it, to me, the one on the right feels more trustworthy. It feels more intentional and more designed. It's interesting. You talk about some of those AI tells. Yeah. What are some of the most common ones you've seen? What are the clues when you look at a page or you look at a design and you go,

somebody just spit out whatever the model. spit out and called it good and moved on. One, I don't think you should do that. And it's so tempting, especially I'm sure if you're like a YC founder and you have zero time and you're trying to sell your first, you know, make your first sales. It's probably very tempting to use whatever cloud design spits out. I don't think you should do it because I don't think it helps you stand out from the crowd. I think it makes you look like one of a million other projects and it makes you look

less intentional and less like, less like you care. And so I think design is this great differentiator. And if you look at every great company of the last 10, 20 years, like basically all of them have exceptional design. You know, and I think that there's, I can't think of an example that doesn't, you know. And so I think if you want to be one of those really great companies, you have to put the carry in on design and make sure that you stand out. So AI tells, you know, we talked about the font weight, the font sizes. Those are huge ones to me. YC's next batch is now taking applications. Got a startup in you, apply at Ycombinator.com

It's never too early and filling out the app will level up your idea. Okay, back to the video. You've probably seen the side, the little side swoop of color, like a two pixels of color on every card. Overusing cards in general, just 20 cards all over the place. Now, these are very easy things to correct, too. Just ask the agent, ask paper to create 20 variations of that layout, and then pick your favorite one, because it's going to look a lot more thoughtful than just a bunch of cards all over the place. What are some of your, do you have other examples of? Purple. Oh, purple, yeah. A lot of gradients, a lot of icons and all caps, like, time. Oh, the tiny headers.

Oh, the tiny headers. Yeah. You know what? A lot of it turns out to be widgets for widgets sake. You know, it's not something that's actually trying to communicate something important or valuable to help the user understand what the design is trying to communicate. Totally. It's just there to fill up space is what a lot of it feels like.

It's like an insecure designer. The agents are very insecure. Yeah. You want to fill it out and there's like lots of little ones and two-dues. Yes, exactly. Like, why do you need that? Well, I don't know. It looks like it anchors the corners on all those cards better or something.

But like, does it actually communicate anything? Do those numbers mean something? No, they don't. Well, we should just get rid of them then. Totally, totally. I think a lot of design is deleting. You know, it's like, it's kind of nice to overbuild and then pull it back. But a lot of avoiding the agent look is pulling back.

It's hard when, you know, the model spits something out. And then you've got to go in and edit it. And you have to say, oh, that headline and describe it and all this stuff versus when you can just go in and select it, hit the delete button. Yes. And everything snaps into place. It feels like you have so much more control. and the effort required to get it to a more thoughtful place is way, way, way, way less.

And that, I think, is incredibly powerful that doesn't force people to compromise on speed, but they can still get something that looks really, really great and stands out and doesn't feel like the models just spit it out. Sometimes we think about this as the best tool for the job, and sometimes it's prompting and sometimes it's dragging. Yeah. And you don't want to use the wrong one.

That's no good, right? So every vibe coded app has light mode and dark mode. And it's the kind of thing, all MVPs, all first, you know, it's like, nobody would ever build that as like the first feature that they want to launch. Right. It's just that you get it for free and the models love to do it. And so that's why everybody has that. But rarely is it the thing that, you know, is worth spending time on or including in your design. And so that's one of the big tells as well.

And is it actually good. You know, I think a bad dark mode is another big role, right? where it's like it's literally using black instead of, you know, or just reversing the colors or things like this. Yeah. So, so Wes had a great tweet recently showing the four, the four-husband of the apocalypse,

which is the, yeah, you, I mean, we talked about this. Yeah. That's probably just delete that. If you have that, you get rid of that, you don't need it. This one, the all- uppercase, little kickers. Yeah. Probably don't need that.

With the extra letter spacing in between. Totally. Yeah. Yeah. This was like me when I was learning design when I was 14 years old. I was like, yeah, space the stuff out. So it looks great. Yeah, this is a good one.

the badges with the extra little icons. Yeah, a lot of pills, a lot of icons. Yeah. And, and, you know, and of course, the purple gradient. Yeah. Yeah. You know, linear made this style great. And then I think it got encoded into the models for the next, you know. Yeah. That's one of the things that's interesting is in isolation, these things are not bad. It's just when they're overused and you see it all the time, to your point, it doesn't stand out. Yeah. And it looks sloppy.

Design is always evolving. The standards are always evolving. The baseline is always evolving. the baseline is always evolving, and you always have to be, stay ahead of that. Otherwise, you know, you just, you look like an average thing, and the average thing is not exceptional, and that's not something people are excited to use. And I think it's hard to build a great company if you have a mediocre presentation to the world. Even impact who you hire and who's attracted to your company for recruiting and a talent you can attract.

So it's worth being intentional. All right. Let's take a look at Moreta here. Pay like a local via QR codes, a global wallet that unlocks, local QR payments across Asia and Latin America. Okay, so we've got a bunch of flags, which I think indicates lots of countries. It's a payments thing, and I get a little nervous, a payments thing that looks vibe-coded. Yeah, again, the credibility of your brand is really important, especially if you're dealing with finances.

Here's another example of like, pull up the content, you know, get your good content above the fold. I kind of like the energy of this site, even though some of it is a little, it's a little too wild. It's a little too all over the place. It's sloppy. But I can tell it has this like, it's trying to suggest this, this global feeling, which I think is really cool. Before we, oh, I love this, how it works. Yeah, this is cool. Get that up too. You know, I had to scroll pretty far to get there. And this just looks a lot cleaner and more professional also. It does. Which makes me trust it more. Absolutely. So before, I ran this one through paper real quick before we came on, just to, all I asked it was actually to pull back. and just create a foundation for this site

that is a little more professional and a little more trustworthy. And we used 5.5 on this, and this is kind of what it came up with. And I actually think, in this case, it lost a little of the excitement of the original design. But that's where you're in a design tool

and now you can come in and bring that back yourself. But I do think this gives you a little bit more of a foundation. Again, way too much text. Yes, so much text. Pull back on some of these things. But I think it gives you a little bit more of a foundation and again, you can come in and delete

some of the vibe-coded stuff, less is more. That's what I would try to do with this site is, is keep the excitement, keep this kind of like international feeling. I like that. It's unique. But you can't go so far, you know, maybe this text gradient is a little too wild and doesn't feel super trustworthy. But I do, I do like the value prop. And I understand what it is for the most part pretty quickly, which I like. I like that they've got a QR code in the bottom right. Which is part of the product, right? It's like. Yeah. Yeah. Which is cool. That's something that I haven't seen very much before. It makes me one of

interact with it. Yeah. It's inviting. Yeah. The overall feeling, the vibe is if we can make it a little more trustworthy. Yeah. But I like the excitement. I like the energy. I like the invitation that I have. Yeah. It's almost too easy to miss it down in that corner, though, too, because, you know, you're so used to seeing the chat with us box down and you just assume there's a bunch of things that you should ignore unless you have a problem. Like, that's where you look. And I don't know, maybe they should bring that QR code up to the main area and say, try it now. Just be the in the hero. Yeah. Yeah. I do. I feel like this hero could be working a lot harder for them. Yeah. Because When I start getting down here, I'm like, yeah, this is good. This is solid. It seems to get more

trustworthy the further you go. Yeah, it is interesting your point that, you know, at the end of the day, the models are the ones that are generating the design when you use them to generate the design. And you're kind of at the mercy for how, like what they've been trained on and how they like to design. Yes. Yeah. And so what's important is using the right models and those models getting better at design, which will inevitably happen over time. And then the next thing that you need is just more control over it quickly. Yeah. And so to have that control, to be able to edit it and make it more in your own vision is actually

the really powerful missing piece here that I think is a lot of what paper is providing right here. The agents can speed you up. They can save you a lot of work and they can do tasks like translation or resizing for you. But we're a big believers in the human element of design and how important that is. And so we're trying to create a tool that lets humans, we're work really fast with agents. Do you think the agents can learn taste in the same way than a human can?

They'll get better at things like letter space. The little tactical things will get better over time. And I think what I'm encouraged by is a model like Fable seems much more thoughtful. It feels like it actually put some thought into its decisions compared to previous models. And so I think that is a piece of design, is that thoughtfulness. Now, do I ever think it's going to replace a human making these design decisions? No, no way. Because a lot of design is actually a very human task of, like, decision-making in an organization.

It's showing people comps and being, and people can be like, oh, not like that. I don't like that. The role of design is, you see the pixels at the end, but the role of design in an org is, like, decisions and stakeholders and bringing in requirements and problem space. And those are not things that agents have any skill at, really. You know, who knows over time. They're going to get smarter and smarter and smarter. But I just think humans are just such an essential piece of building stuff that stands out from the crowd. Yeah, it's interesting. I was having a very similar conversation with someone who leads design at a multi-billion dollar design forward company about there's a thought process that goes through a human's head where they're reacting to something and a criteria they're evaluating.

And it's very implicit for us as humans right now. And the question is, can you verbalize what those things are? Like what is the process of you deciding whether something is good or not or looks good or not? Very subjective. Hits the bar or not. Can you literally go through and quantify all of that and teach that to a model to the point where it has tastes like a human would? And then the question becomes, well, if all the models were able to do that, does everything just look the same again? And then you always need the humans to be driving to finding the next thing that's actually going to be new and stands out and pushes the limits more.

We see that even within human pre-models as trends take over. Everyone kind of looks like linear for a couple years and then somebody else figures out something, right? I do think the models will get good at tactics. They'll get better at not making everything bold, you know, and I think that's great and that'll be really, that'll be really nice. But just the thought, the product, there's so much of design, again, it's like, was this showing value proper or not? That stuff's more important. This site, for instance, that I think is a little too wild. Maybe it's really important for them to be really fun and exciting.

Maybe that's what their market needs, you know? And so maybe it's a higher order thing for them to be very exciting. And can a model make that decision? Can it, you know, maybe with enough information you could start to get there. But I think we're pretty far away from that. Yeah, maybe 10 years, something like that. Yeah. Maybe I don't know.

Maybe I'm too pessimistic. Maybe it'll come sooner. But I still see humans as just this essential piece of the puzzle. Yeah, we certainly hope so. Yeah. Let's hope so. I'm curious. How are you guys using agents internally at paper?

In some ways, we use them a lot. And then in other ways, we're kind of like dinosaurs in some ways. Like, we don't have agents in our. our Slack. And we don't have, we use agents for coding quite a bit, but we still read our code. We read all of our code, every single line. Every line. Usually multiple humans reads every single line. Why is that? Well, one is design tools are really hard to build. You know, you can't vibe code a design tool at the level of a Figma today. Maybe you can eventually. And so it takes a lot of really precise, really careful work. It's very multidimensional in terms of the systems. And so

we just have to be really careful. And we want to make sure our experience, you know, designers need 120 FPS. It needs to be fast. It needs to work every single time. And the agents just aren't there for that yet. The other thing is a small elite team, right? So 12 people keep the talent bar extremely high. I think when you're building companies, that's how you should try to do it these days and use the agents to accelerate the people. But if you have less people, you have less communication burden, you have less of these other costs to go through. One example of how we use agents, though, is Agu, our brand designer. He's a brilliant brand designer, great at visuals. He codes a little bit, but like I don't think he'd call himself like a programmer. And he made our

entire website in about a week before this last launch, including shipping it, including all of the animations, and he designed everything in paper. And this is the kind of design. You wouldn't want to do this with prompting, right? You need this, he has shaders in the background, subtly coming through these motifs. He did everything in paper, and then he just prompted, and this was, you know, six months ago now, so different models. He just prompted the animations on top of his paper file. And it's like, you know, take my paper designs, turn it into a code base. This is the next JS site and put it together for me. No one else looked at this.

So Agu did this end to end in a week. Because he's a designer, he puts in these Easter eggs. There's little like fun extras. And it's all just interactive. It's interactive. And this is something he would not have done. He wouldn't have had the time. I'm sure he could have, you know, skill-wise, he could have got there.

But like he wouldn't have the time to do this. So I think that's really, really cool. And so I think a lot of times it shows up more, not in our core product where we're actually pretty human-based still for precision reasons. But in all the other stuff, all the videos we need to make, all the marketing site, all of that other stuff where we can, you know, have agents help us help people go faster. Yeah. So we do, yeah, quite a bit of that. And then, of course, the coding agents, you know, we do use cursor extensively.

We use cloud code extensively. We use bug bot for our PR reviews, which has been really helpful. It's a little too pedantic sometimes, but it still catches bugs, which is good. There's a little bit of a fomo right now with agents where it's like, oh, every company is shipping without looking at their code. Are you going too slow, you know? We are kind of taking the approach that quality software. where it takes time. And we have competitors that have every single feature,

but they don't have the attention or people don't care. And for whatever reason, whatever that is, is like people want paper to be better. They don't want the competitors that have every single feature to be better. And I think that's because we care. And we're putting in that precision, that care about these things. So even if we're moving a little slower,

people still care more about it. And so I think that quality matters still, thankfully. My sense from talking to a lot of designers is that you guys are almost building kind of a ground swell, movement, like a cult movement of people that are now using paper as the new cool thing that a lot of designers are using and it's the new modern way to work. Was that intentional? And what are the things that you're doing to try to get more users, to grow, to get some mind share? Like, you seem to be one of the only companies. Figma obviously has built this huge behemoth and gone public and,

you know, all the Fortune 500 uses them and on all this stuff. But it feels like you've started to to crack some of that, certainly within certain user types. And how have you gone about doing that? And what's been most successful for you there? Yeah, a huge part of that is like being very values aligned with the market. We didn't have a product for the first year, and we still got to about 25,000 Twitter followers just by talking about our values. I'd go on podcasts, be like, I really like design.

Do you like design? Then we get into the topic and talk about typography or whatever. Being values aligned with your market, they will forgive so much about your product, missing things. they'll wait for them to come out, they'll root for you, if they can tell you're authentic. You know, and if you're building design software, it really matters. It really matters. Maybe not for all software, but for design software, usually best product, most authentic product wins. And so I think for us, it's just, you know, again, company of 12 designer and engineers and just live our values and people really resonate with that. And by the way, the product is

really good too. I think that part's important. You've got to get there. But, you know, I think that's part of it. And then there's market clock, every company. Hopefully we have this problem someday, every company eventually goes up market and you just stop being as cool when you do that, you know, you're selling enterprises. You're not, you're not underground at the cool events, you know, anymore. So I'm really honored that we are part of the community. And to your point, like, I think you saw, you know, Ramp has been releasing the data of actual Ramp subscriptions on their credit cards. And if they released an AI, you can chat with about the Ramp data. And if you ask it about design tools, it's, you know, Figma obviously, Sketch is there. And then

papers is right there with Sketch. And so we're the first company since Figma's come out to, like, make a dent in that market at all, you know, and it's growing very quickly, which is really encouraging. It's really exciting because you don't know when you start a company, like how it's going to go. You have no idea. Like I said, there was months where I was like, nobody's going to care about this. And to now have like so many cool people. People are way cooler than me that care a lot about paper and are like wearing our merch and stuff. It's like so much fun to see. And honestly, it's the best part. That's very cool. And this is not your first company. This is your second company, right? And it seems like even before you had a product, you have gotten yourself out there and your vision

and your values, and you become the ambassador, and that's what people were buying in to before you had a product, and certainly in the early days of having a product, and that's what's been successful as you, and getting that out there. And how is that different from, you know, you built another company before this and, like, contrast the two. Well, so my first company Modules was working on designer developer handoff, and we built Redix UI along the way. It was like one of our side projects that just kept going, and, you know, Redix became this

default React component library. one was this was pre-AI. So, you know, everything was harder. Designer, developer, handoff was harder. Two was we really listened to both sides at once and we got very confused because designers want engineers to change and engineers want designers to change. And like they will tell you the other team will use your software. Like you can't build software that way.

And so we learned some hard lessons about making something people actually want. And then I think I've been able to apply at paper. But we did make great. I mean, the Radix has amazing product market fit. We just didn't see a way to build like a venture scale business around it. And so we thought, let's just make it for. and give it away and then, you know, come back with something next time. And so that's been cool.

I mean, that worked out really well. We sold this company to Work OS, which is an amazing, often identity company. We use WorkOS for paper. And then a couple of years after that, I just wanted to start back into creative tooling again. And are there any lessons that you took from that experience that you're applying this time? You can way over complicate product building. You know, the make something people want is a great phrase. And really, a lot of times with your product, it's like, what does my company need next?

What would be like the best thing that possibly could happen in my company right now? At the beginning for us, that was like being part of the conversation, having Twitter followers. Well, let's go do things that get Twitter followers. Let's go on a podcast. Let's talk to users and then represent their values outwardly. So, you know, a lot of times it's actually very simple. And I think we make it too complicated and we're trying to have these brilliant strategies for our companies. And it's like, what do you need next and just do the most straightforward thing you can possibly do to get there?

And, you know, and don't be scared to do things that sound hard. you know, obviously you got to do them. So that's the philosophy we've taken bottlenecks first. Huge, huge philosophy I really believe in. Like for us right now, the bottleneck is early adopters are using paper. The people that are more, you know, maybe not an early adopter type are very used to what they're doing now. And they're missing features like comments and components.

And so now our bottleneck is actually features again. And so now it's like, well, that's what we got to go do next. And it release that bottleneck to the business. Another just like a tactical thing is just like talking to users every day. It's so, so important. I had a rule. The first year of the company, I had to talk to a designer every single day. And I did it all the way through the year.

And that was so useful because I learned just how, like, when you start making the company and the product, you just have a different perspective. You can't help that. You have a curse of knowledge. And so just staying rooted in user conversations is really important. Yeah. It gets harder and harder to do that as the company grows in scales also because there's so many other things that are vying for your time and pulling you in different directions

and that seem important or hair and fire in the moment. Yeah. But really it's those core fundamental things of build your product, talk to users. Yes. And, you know, exercise or whatever you need to do so you don't go crazy doing the first few things. But that's the core of it for the life of the company, right? And that's it. And so it's cool to hear that you've done that from the very first day. And I'm sure it is something that has been instilled culturally in your team now. Yeah. And that everybody is talking to users at this point. Oh, yeah. Yeah. When people on board, it's like, yeah, you have to be in the Discord talking to people. the slack. It's really, really important. And not everyone's comfortable. Maybe they've come from a bigger company where they were told they can't talk to users. And it's like, no, get in there.

You can't screw it up. And you just learn so much. And I think it guides your decision making. Decisions are really important in a company. You can kind of look at successful companies of like, you can trace their decisions backwards in time of like, oh, that was an inflection point. That decision was really good. And so I think having an informed employee base, like, I can't make every decision in the company. Even at 12 people, there's no way I could do that. And so if you can have your employees be really informed about users and what they actually want, you'll have good decisions being made all over, you know, all over the place. And yeah, it's really important. What do you think is next for design? Well, I think a lot of design isn't changing. Again, in

orgs, it's like the function of design is still really important as a way to explore problem space, competitive analysis, executive stakeholders, things like this. We're not seeing that change at all. And so, you know, that function of design, I think, is sticking around. If anything, I think we're to see more because there are more companies and there's more software to be made. So we're seeing just more and more design need in the world. I do think the tooling is going to accelerate because I think we just need to be able to keep the quality bar up at a pace that is, you know, keeping up with the engineering with the product teams. The software will ship whether the designer gets a chance to make it better or not. This is kind of like a reality of the competitive

landscape. So how do we help make sure that the designer can be part of the conversation and keep up in that shipping process. But, you know, I think in general, sometimes I think there's less change coming than people think, actually. Because I think that a lot of design is a very human problem. And the pixels at the end are like the output of the process. Yeah. It's the medium at the end of the day, right?

Exactly. Yeah. And what's next for paper? Well, so we are, like I said, we're building out some table stakes, components, comments, bottleneck reducers. I'm really excited. Something I've learned from talking to users is people are building all these prototypes all over the place. And they're, they're ending up on Vercel or their HTML files,

and people don't know how to leave feedback onto them, how to even find which one someone's talking about. So what we're going to do is we're going to put, basically, you can like, eye frame things onto the paper canvas as a place to keep your prototypes kind of like organized. And then people can leave comments as they're used to doing on the prototypes. So when people ask us if we're doing prototyping, it's like, well, kind of, we're going to help you organize your prototypes and have feedback cycles. And I think this is going to be a really big unlock for companies that are, you know, building playgrounds for their designers to vibe code in, but then it's kind of like this messy result and they need some tool to help them organize that.

So that's a big thing we're going to be working on the next, you know, six months or so as we're building out the table stakes that people expect to. Amazing. Steven, thank you so much for coming on and sharing paper with us and all of your design insight that I know has been hard earned over the years and the companies that you've built and the incredible places you've worked. So thank you. Yeah.

Thank you so much for having me. That does it for this episode of Design Review and we'll see you on the next one.