As software gets easier to build, design becomes more important. And there's going to be more designers hired. Because at the time it felt like this tiny market. I think that so far AI is very much in the tool category. You know, we talk a lot about lowering the floor, raising the ceiling. Making it so that more people can participate in the design process, but also raising the ceiling on what you can do. The more time has gone on, the more that I'm confident in the designer's role and believe that's going to be one of the critical roles in building software in the future.
Today, we have a real treat. We have Dylan Field, the co-founder and CEO of Figma, which millions of people in the world use for collaborative design. Dylan, thanks for joining us.
Thanks for having me, Yuri.
So, we're in the middle of this crazy AI revolution, and design itself is also changing quite a lot. How do you view AI changing the face of design right now?
It's definitely an incredible time that we're living through. Ever since we started Figma, the exponential curve we've been on is that there's just more software being created than ever before. But it's kind of looked like this before, and now it seems to be going vertical.
I think it's super exciting. And my point of view is that design has the opportunity to differentiate software at this age of AI. We'll see whole new workflows emerge, where designers, engineers, product will work together in different ways than maybe they're doing today, or at least different cadences. And it's been pretty neat to have a front row seat to that, and with our platform, figure out how we can empower everyone to use Figma to be able to be on that journey of design. If I think back to just even the earliest days of Figma, some of the metrics that have held constant, one is that we've always had this really strong international base of users. So, even from the start, over 80% of our weekly active users were outside the United States. The second one is, from the start, it's been like one-third designers consistently. And the other two-thirds are non-designers, all coming to the platform end, making things, viewing things, giving feedback. I'm quite excited about how we can activate those users more, make it so they're able to make more contributions to the product process.
That sounds right. Does the capability go up, and are you doing more, or will AI take these things over? What's your view so far?
Oh, I think that so far AI is very much in the tool category.
And whether it's designers, developers, others, people are using AI to just do more and to explore more. So the way that I see it is that, you know, we talk a lot about lowering the floor, raising the ceiling, making it so that more people can participate in the design process, but also raising the ceiling on what you can do. And I always had this image in my head of like the idea maze you go down, all these different branching paths that you explore as you design and ideate your way through that product journey. My point of view is you get more breadth now in terms of the things that you can try, but it still requires a lot of depth to fully explore it. We're sort of in this uncanny valley still where, you know, you can do prompt app, but prompt app doesn't give you an app that is well designed yet.
So, you know, given that, like, how do you see this playing out?
It's kind of interesting to like look at the language people use around this stuff. You know, people talk about like, you're locked in, or I'm cooking, or they're cooking.
And vibe coding, exactly. And for a while I was just like, you know, whatever, it's new language. And then at some point I realized that maybe the reason people are using this language in particular is that's kind of the feeling you get as you're developing something really fast. Like there's always this call and response you have with tools as you're trying to get the thoughts out of your head onto the screen and you're iterating quickly. And the faster you make that feedback loop, I think the more you can get into that flow state. And also the more fun you can have, actually. And, you know, for Figma, one of our values as a company is play. We're always trying to make it so that whether it's as a culture or as a product, Figma is fun to use and is a tool that just like lets you creatively express. I think that the opportunity here is how do you create even more of that feeling, even if you're not an expert at Figma, and have that rapid ideation loop going at all times.
But to the point you made, it's like, you also want to give people a way to not just get started and prototype rapidly, but also get to the finish line.
And I think that's where the disconnect is. And I think that that disconnect is not just for design, but also for code. Like you can go and get locked in for three hours and make something cool. Will it be the design you want? Probably not. But also it will take some time to untangle the spaghetti mess of your code base as it currently stands when you're going from start to prototype, and then get to something that's extensible and you can build off of. The design part is very interesting because you can get something that works very quickly, but maybe it doesn't work well yet. There's sort of this increasingly lost art, I'm worried, that founders are not focused on design enough.
Do you find that in some of your angel investing yet?
What I'm seeing is that people seem to, year over year, and this has been true for a long time, care more about design than ever. Which is great. I might have some selection bias. Because obviously people are excited to talk to me as an angel or my capacity at Figma. But overall my sense is that people just care much more about design, also understand the value of design. It's not just about can you make it work? There's a lot of people that can make it work. And as people are able to make stuff faster, it's how it works that actually matters.
Can you talk about how you're thinking about the evolution of the actual models right now? Because there's sort of two regimes, right? Image diffusion models like Midjourney and our friends, Suhel at Playground, they're doing image diffusion models. And then on the flip side there's code gen. And the code gen methods, the large language models themselves, they can't generate SVG, for instance. Do you think that one or the other is going to work? Or is it both in tandem and then way out in left field is the multimodal models coming from OpenAI and others? Is it just a matter of let's just wait a little bit and those will do it? Or is it some other combination?
Well, I mean, it almost goes back to the question of what is design when you're asking the question of why aren't models better at design? Because they still kind of suck. There's so many definitions of design, but one I like is art as it applies to problem solving.
I think maybe the reason why these models are not great at design yet, it's like on the art side you've got diffusion, on the problem solving side you've got LMs. And it's not clear that people have figured out how to marry techniques together yet and figure out how to go approach these problems in the right ways. And I think also just like as you think about what a designer does, they're bringing in all this context, way more than just, okay, here's this like two-liner problem I want to solve. They're doing a lot of research with the user like you talked about in order to go into that IconicPie and figure it out.
Yeah, they're in the IDMAs trying to figure out from first principles, what is it that this person needs or wants? What are they feeling at any given moment?
Yeah, and how do you take that all in? Also, the current cultural moment you're in, the flow you're in, the part of the brand experience you're in, how that all ties together, how it all maps. And then I just, yeah, I don't yet see models that have been able to figure out how to like bring all those different components in. They're lacking the empathy piece of being a designer. Maybe that's too, I don't know. For now, I don't know. Maybe not forever. Yeah, I'm curious to see like how they'll evolve. But I also, the more time has gone on, the more that I'm confident in a designer's role and believe that that's going to be one of the critical roles in building software in the future. And I think people that call themselves developers today, they might call themselves designers, you know, in the near future.
That doesn't mean they're not writing code. It just means that they're going to be thinking more holistically about the product experience along with their colleagues.
Yeah, that sounds right. There's this sense that the large language models are going to somehow become super intelligence through just getting smarter and smarter. But it doesn't seem like there's a lot of agency. Whereas what a designer, a large percentage of what a designer does is actually using their own agency to like create a thing on a canvas, basically.
Agency and also judgment.
And I think that it's super interesting to me because I think in the Grok 3 announcement, for example, one of them was saying that a lot of the training is done on math problems and CS problems and that kind of transfers over to everything else.
But then I think about the people I know as humans that are like the best mathematicians I know or the best engineers. And like sometimes they're really good at having great judgment about product experiences and sometimes they're really not. And it seems like it's almost a different muscle, at least for humans.
And so I'm curious if that transfers over or not.
Yeah. I mean, my hope is that humans basically retain the agency bit. And I think we were hanging out with some AI researchers recently and they were sort of disappointed that the models are not, you know, but they don't demonstrate as much human-like agency, you know, even after all these innovations with, you know, climbing the scaling laws, it just hasn't popped up. The thing they said was the models are more like hyper-intelligent toasters. But I think that that's the opportunity here. It's like, you know, the human designer has to understand people and understand their needs and goals and their reward functions and then create UI, create user experiences that match, like, you know, their jobs to be done. From there, it's about how do we equip them with the right tools.
Remove tedium, make it so that they're able to have that rapid feedback loop.
Do you think that people are going to do more and more, you know, chat-based interfaces or is it more visual design or, you know, is it even going to be like terminal and prompting? Which way, you know, what are you seeing right now in terms of the cutting edge of design?
I'm seeing pretty much everything right now.
So way more variety then.
It feels like we're kind of exploding out the possibility space of how you actually can work and that's been my prediction of what's going to happen for a long time. And I think it's actually go much further out from here for a while in terms of people trying different patterns with their tools and different things. I think it's a really good thing. I think like prompting alone is sort of like we're in the telnet days of AI. You're able just to have, I think, so much more ability to work with tools through different interfaces. That doesn't mean that prompting is not important, but I think there's a lot more to try out there. And I'm excited to see people doing that. And also explore different design patterns and things like, you know, AR, VR, XR. You know, there's different windows where I think we're going to see entirely different dynamic interfaces start to pop up. Fundamentally, you know, I love just how humans interact with machines. We don't have BCIs yet, but one day that'll be like the final frontier. You know, until we get there, like there's a lot of other stuff to explore.
I guess the really wild thing is once you have CodeGen, in theory, that might be something we see. Like sort of custom designed vertical software.
You mean basically custom designed for each user?
This is one I'm a little skeptical of. Doesn't mean it's not going to happen or people won't try it. But I think that I remember like seeing Snapchat for the first time. And then, you know, years later, like starting to use Snapchat more. That's an example of software that someone's going to teach you. It's like a consumer product and yet it's actually quite complex.
You're not going to just like figure it all out if you land in Snapchat. You're going to learn it because you watched your friend use Snapchat. I think there's a lot of software that's like that. And the more you shift the interface around, the less someone's able to just natively learn and transfer over that learning to someone else.
So I think there's like a trade-off there that's actually quite important. Snapchat and Minecraft are two of the more crazy examples of like, you know how in design, low bar, high ceiling is like a maxim. You just want anyone to be able to clear that low bar and then you want people to stick with you all the way through the high ceiling. They were like remarkable examples of like, actually the low bar is optional.
Let's take a step back. And I mean, I love your story of starting Figma. How did you come up with the idea?
It was a process. I guess going all the way back, we started talking about what would become Figma, Evan and I, in late 2011. I was at Brown at the time, Evan's my TA. We were kind of like asking the question of why now first? What are the technologies that are shifting and will create all sorts of opportunities? And like late 2011, that was for us, drones and WebGL.
And I was actually pushing a little bit harder on drones. And Evan very quickly vetoed that. He's like, look, I think drones will be very important, but they're going to be regulated and or defense. And I don't really want to do either of those. And also I like software more. Once you have the hardware component, the run debug loop, it's really slow. He's like, I don't want to deal with that.
Yeah. So that was kind of his call. And so it was okay, WebGL, which Evan had a lot of expertise in particular. He had done a lot of prototyping and making sort of toy applications with WebGL at the time. And we started to think about, okay, what do you do with that? So then I was like, okay, is it gaming or is it tools and software? And we kind of looked at that and said, okay, well, gaming is super hard. It's like a hit based business. You got to make something awesome. It's got to really resonate. Like, let's not do gaming. Let's do tools and software. And then I went and left for a gap semester.
What I thought was a gap semester.
Yeah, totally. To work at Flipboard. Evan was finishing up in sort of spring 2012. We started to just kind of like tinker and explore and think about this together. Looked at stuff like 2D to 3D scene creation. I got the Thiel Fellowship. Then we started working full time August 2012. My internship ended. Evan moved to the West Coast. And explored a lot of different stuff.
I think you made like meme creation.
We did. Meme generation was one of them.
Probably the low point of Figma. I was like very on it in terms of my thesis was correct. Memes to the moon. I was like literally looking at the graphs for memes and meme generation and just looking at the search terms and stuff. And I'm like, oh, man, there's no good meme generator out there. We can make the best meme generator. And we spent a week. We made it. It was like this moment of, but is this really what we want to do?
And I think Evan was ready to quit and I was ready to go back to school. And I was like, okay, this is a clear sign that this is not it. This is not it.
But Evan, being the technical genius that he is, he made the text rendering in our meme generator state of the art in the canvas. And then later on, when we eventually got to Figma, he would get us to kind of take that out and reuse it for our V1.
And so it's kind of like this hilarious moment of our first text rendering in Figma came from the meme generator.
Yeah. Do you remember, I guess, I mean, a lot of people were still using, I guess, even Photoshop or Illustrator for this type of high fidelity design.
Yeah. And it was a awesome tool in many ways, but also very buggy. And at some point it was killed. So I was looking at, we were looking at the photography space as well, photo editing, built a really cool photo editor. But it was just not something that made sense because we're doing the browser. And at some point we just kind of like picked our heads up and went, wait a second, like the best camera is the one that you have in your pocket.
It's on the phone. Like, why would we be making this browser based editor for the photos you're taking on your phone? It'll just be an app. And we looked at the app market for photo editing and was like, no, this is totally crowded and this can be a commodity.
So then we said, okay, can we do something else there? Storage or search or interesting machine learning stuff, computational photography. And it felt like either it would go to the platforms or it'd be sort of too early. Well, the computational photography stuff we're looking at was like, in retrospect, quite a bit too early.
You know, more than a dozen years too early.
And then we, you know, saw fireworks get killed and went, wait a second, maybe there's an opportunity there. And started to form the thesis of as software gets easier to build, design becomes more important.
