Dylan Field: Scaling Figma and the Future of Design

Designers need to be founders. We need to have folks that are designers step into the founder role and start companies. It feels intuitively like we're in the MS DOS era of AI right now. If you look back 10 years from now, everyone's going to go, can you believe that we just had this chat box?

Yeah. Today we are many different places. We're hybrid, 1,700 people now which is wild. I have to pinch myself on that number. We have eight products now. We just doubled our product lineup at our last config. So very excited to hear feedback if you got any on things like Figma, Make, Sites, Draw, Buzz. But it's been a very exciting time. Lots of work we're doing as we explore all the things that we can do to help our audience.

Yeah. So, in the early days of Figma, well, before it was even Figma, Evan and I were at Brown together. He was my TA and we were asking ourselves the question of why now? What's changing the world? The two answers that we came up with that we also felt deep conviction in, one was drones and quadcopters. The other one was WebGL. And Evan after about a month or so said, hey, not into drones for all sorts of various reasons. That was kind of the one I was pushing for more at the time also except WebGL of course. And then I was like, great, WebGL it is. So WebGL, I think everybody probably here knows, it's a way to use the GPU in your computer and the browser. WebGPU is its successor. And we started going really deep on what are all the things that we can build. Two main paths were games or tools pretty fast. We said okay, not games, let's go tools. And then it was a deep exploration with many twists and turns as we explored all sorts of tools that we could build. We really started in earnest August 2012, whereas we started talking about it more December of 2011. So it took a while to get to the point where we started. And then from there, I would say it was at least June or July of 2013 before we went all in on okay, let's build Figma as it is today. And even then there was still a bit of a narrowing path to get to the product that exists now.

No, it was definitely the hope was startup, and a startup that could scale at the same time. My downside case was I get to work with Evan who I considered then and consider now to be a hero. He's like the smartest guy I know. If you have any doubt about this statement, just look up his GitHub. He's an amazing man and an absolute genius. And I figured worst case scenario, I spent a few years working with Evan. I learn a lot. Then I go back to school, same place I'm at now, can't hurt. Upside case, we go build a cool company. All the problems that we were thinking about working on were very interesting to me. And so I didn't really see any risk to the scenario. And also it helped that I had the Thiel Fellowship. I would have done it without it, but having $100K over two years, I know now inflation etc. probably sounds like less than it was then, but to have actual cash and not have to dig into savings or go into debt, huge deal. Not just because of the cash element, but also because it gives you time. If we had stopped six months in and that was our point where we made a call, Figma would not be here today. So I think if you're a founder already going or you're thinking about founding, you got to give yourself time somehow. That's really important.

Well, first of all, just working with Evan was super fun. We're kind of thinking through ideas by building them. It felt every week like we're inventing the future in some way. At some point, I went, 'Memes are going to go to the moon.' And I convinced Evan, hey, let's go build a meme generator. This is 2012 time frame. And we built a great meme generator. I think it would have been the best one in the market. And my thesis was right, by the way. Look at the exponential curve of memes since 2012. We would have made some money there. At the same time, after a week of that, I think both of us were ready to quit. I was asking myself, why'd I drop out of Brown for this? That was probably a low point at the start. But other than that, there's the constant existential nature of asking yourself, what are we doing? What's the big goal here? When you're in that phase of really trying to discover what to work on. But I think if you've got a co-founder, you got a collaborator, you're not just alone. Hopefully your highs and their highs, your lows and their highs cancel out somehow and you can feed off each other to keep going. That really helps.

The first users of Figma, a lot of it was cold emailing and people in network. Folks that I had either interned with at Flipboard, LinkedIn, O'Reilly Media. From that there were people I could reach out to. They could tell me others to talk to. But also I just looked online: who are the designers that I think could be really helpful to us and I respect their work. If they answer my email and they let me buy them a coffee, it'll be a personal moment for me because they're my hero. And a lot of them replied. It's kind of wild that people reply to cold emails, but they do. So I went there and it turns out designers give great feedback. It wasn't just meeting them and them saying, 'Yeah, your product sucks.' They'd be like, 'Here's exactly why it's not great, and here's what you can do better. Here's what it would take for me to use this.' The more I engaged and we worked through that, the better the product got. I'd follow up with them and eventually they started converting. Some took a while. Later on we went on tour. The venture firms that invest in us invested in other companies too. I had them make introductions to the companies. For an entire summer, I met with five, six, seven companies a week at least, sitting down with them, sometimes several a day, saying, 'Here's a demo. Will you use it? If not, why not?' Very low conversion rate. In that entire summer, maybe two of them went in for it and actually started using Figma. One was Notion, the other was the company that became Kota, then called Krypton. It's interesting they're both cloud-based document tools with very similar philosophies to us. Then you launch it and people start using it more. There's a lot of folks out there that resonate with the message. So it was a slow arc over time, but the constant was feedback, getting feedback to the team, making sure we understood what problems we need to solve.

First of all, I definitely echo the point of launch as soon as you can. If you take anything away from this, it's don't do what I did. Get your product out faster and charge money faster to see if you actually can make money. Unless you have some genius galaxy brain consumer thing you're doing, in which case figure it out yourself. I don't know what to tell you. I think feedback is essential and you should launch as quickly as you can. For me, the feedback was very clear: it's not ready. That made it so we didn't feel comfortable launching yet. But looking back, we did have the capital. I should have scaled the team faster so we could move faster and get it out quicker. That's something I now looking back have learned. When a team at Figma comes to me with an epic roadmap that they think is perfection, the first question I always ask is how do we slim that down? How do we make it more bite-sized and test this earlier with our users? So it's absolutely the case that I try to push people internally towards a one month or three month cadence at most. If someone comes to me with a 9-month, 12-month, two-year cadence, it's like, what are you doing?

Yeah, constraints can actually really help. But I also think the startup cycle is something along the lines of if you're the leader of a startup, you need to be identifying what you're doing the most of, figuring out how to get someone else to help you do that, or maybe in the future it's AI, who knows? Then from there, how do you go find that person? If you don't have enough resources, how do you get the resources? That's a cycle you're always in. It turns out that having constraints breeds creativity. It breeds interesting ways to solve problems. So yeah, I think they're useful.

The point at which I started to believe that actually this might be real was way later than our users did. People were telling me, 'This is amazing. I'm really excited. Here's my 12-page doc on all the things that I want you to do for Figma.' I should have known then, even though our product was really bad, that there was something there. But in reality, it took until Microsoft told us, 'Hey, this is spreading like wildfire, and we're asking ourselves, should we shut it down or should we keep going? The reason we're asking ourselves that is because you're not charging us. Maybe you should actually charge for the product.' That was the moment that I was like, 'Oh, I think something might be working. We should probably charge people.' That was like five years in. So, yeah, don't do that. Also listen for when people are pulling the product out of you. Everyone talks about product market fit, but product market pull is really important. You'll see signs of it when people are highly engaged, when they are obsessive about what you're doing, when they see the future of the vision that you're planting. That is a sign that you should really double down. Many people interpreted that as, if only we had all these things they're asking for, then we might have product market fit. Guess we got to grind for a long time and who knows if it'll work. The right mindset is, oh my god, they actually care enough to give us this feedback. This is huge. I think people misinterpret that too much.

I think maybe it's just like childhood for me. When I was growing up, I was a child actor. Not like a child actor that got into anything really cool that you know about, like commercials and some TV. But as part of that, you audition constantly and you constantly get rejected. For me that was not a big deal. I was used to rejection and I had fun with the process. So for me it's just maybe a different mental equation than others. But if you're not there yet, seek rejection. It's got interesting data in it. Don't you want to know the data?

I think that in some ways it's new, in some ways it's not new. Design has been growing in importance exponentially over the past decade. At Figma we see it up close every day. More designers being hired. Design going from lipstick on a pig, make it pretty at the end of the process to let's deeply think about how it works every step along the way. That's been a mindset shift that's been ongoing. But now, in this age of AI, if you really believe that development gets easier and it's more simple to create software, it's faster to create software, then what is your differentiator? It's design, it's craft, it's attention to detail, it's point of view. What we're seeing is recognition of that. Airbnb literally said our differentiator is design. I think Brian said that. There are lots of takes on OpenAI and this more than $6 billion transaction. Some people are like this is the stupidest thing in the world. Other people are hailing it as absolute genius. I guess my mental model is there are some people out there who, when they do something you don't understand, it's easy to go into attack mode and dismiss it. But over enough time, sometimes you see patterns and you're like, 'I've consistently not understood what this person's saying over the course of years.' Years later, I go back to it and I'm like, 'Oh, what I said in response to what they did was just wrong.' Then you do this mental flip of, assume there's something to learn from whatever they're doing. Assume you're missing something. I look at something like OpenAI and some part I understand: design is differentiator. Some parts I don't understand: that's a really big transaction. But Sam is one of those people that he's right about a lot of stuff. So I would encourage you, if you just dismissed it outright, to ask yourself what you might be missing.

If you look historically at the products we've launched for Figma, the pattern is we notice behavior happening in Figma Design. We take it out of Figma Design and make it its own product. Therefore, Figma Design is able to be what Figma Design wants to be, a product design tool. Whether it's FigJam, a whiteboarding brainstorming tool, the first new product we launched that we can make a dedicated space for and make it be everything it needs to be, or it's Slides where we saw 5% of files created in Figma Design are slides. So great, pull that out, make a slide tool because there's all the stuff you need for slides that if you put it in Figma Design, now you've got a complicated UI and one plus one is not three, it's more like 1.5. A lot of the things you saw launch at Config are in that category. Draw, for example, which is a way to do more vector tasks, we made a separate mode so that users can go deeper because again, if you believe craft is differentiator, more people want to be more expressive. How do we enable our customers and designers everywhere to do that on the Figma platform? Buzz same thing: you have all these people that want to create mass exports and figure out ways to create production graphics. So if you got a brand team and they've created templates, how do you make it so that you can empower a marketing team to go use those templates and do mass creation of assets? That's a core workflow we see all the time, but we didn't want to make Figma Design more complicated or dumb it down. So instead you make a new surface. Sites, we see people designing websites all the time in Figma Design, but they have to go somewhere else to actually build the site and get it out there. So how do we do that so they can ship it? And then Make, we're so excited about. This is a tool that lets you go from prompt to app. It's already changed a lot of how we do work at Figma in terms of quickly prototyping and being able to get to the point where you throw ideas away faster. With Figma Make, there's so much more that we want to explore and are excited to explore there.

I think that for us it's all about speed of iteration, speed of testing ideas. Tools like Make really help with that. It helps to have ways to rapidly prototype and figure out what's going to work and what's not going to work and make that as low cost as possible. There are tools I can't talk about and things we're developing that have been pretty instrumental to how our development process is changing. Can't wait to talk about them, but not today. When you go back to the way that design and development are blurring more, there's a lot going on. I think product is also blurring with design and development and potentially even parts of research. All this is becoming less distinct and it's all coming together more. I think this is happening before AI, but it's happening even more with AI. AI empowers generalist behavior. I think the models today are better at the earlier phases of development than they are at late stage code bases. If you have an established codebase, you're going to get less out of AI development tools as they currently exist than if you're at the very start. So everything is better suited for prototyping and zero to one than from one to 100 at this moment. But in a week this could change.

I think that it feels intuitively like we're in the MS DOS era of AI right now. If you look back 10 years from now, everyone's going to go, can you believe that we just had this chat box? Yet the problem of how do you show users all the things that are possible to do with these models is a very hard challenge. The experiments that have worked there are interesting. Look at Midjourney: they started off in Discord where you can rapidly see all the other things that people are doing, and that was a way to show people what's possible. Or even Meta's new AI app. There's been a lot of press about the public aspect of people sharing things that are quite private, but the flip side is you actually learn what you can do. I think that's underexplored in the media. So there's this problem people have not solved of how to expose capabilities of these models. There's so much to develop and work through there. On top of that, everything will be more contextual: AI as you blend it into different applications. That's a really interesting layer to think about. On top of that, we're going to have so many new surfaces. The surfaces that will exist are not just your phone and laptop and tablet. It's going to be glasses. We're going to see much more in terms of different types of displays that exist throughout your life. The surfaces are going to multiply. AI will have context. All of it will be a layer you have to intersperse, and that is a really interesting challenge for design: how do you reconcile all that, keep it consistent, and actually navigate that broad spectrum that people expect you to show up on?

I think that a lot of researchers are trained in an academic environment and come at problems as abstractions and try to think very generally. If you're doing pure math, keep going, that is definitely the way to approach it. If you're doing more applied research, for example in AI, I do think that thinking like a designer can be helpful and working with designers can be helpful. We found that embedding designers into our research teams, because obviously we're doing a lot of work on how to make better AI tools for designers, has been critical. Researchers need that intuition of how designers think. Without that close collaboration, it really doesn't work. You might say, well, that's nice but you're building for designers. My response would be it's the case that designers have this mindset of building for an audience. Maybe it's a general audience, maybe it's a specific audience. That audience has a problem or a set of problems they're trying to solve. That sort of thinking is very useful to bring into the research context. Also qualitative research needs to pair with more deep AI research. The more you can surface through qualitative methods what people are actually trying to do and how they perceive and think, the more you can advance. So my push for anyone from a research background would be go get in the field, go talk to people. You'll learn from it and it'll make you go faster. Some of the ways that designers have learned and some of the tools that designers have are likely useful for you.