Welcome to Bos to the Future, a podcast I started so that we could get a little deeper on a few topics. Often in our industry conversations can be very surface deep and very broad. I am thrilled to be joined today with Dylan Field, the CEO and co-founder of Figma, a company he started when he was 19 years old. Now 13 years later, it is the de facto design standard in our industry. You have a hugely impressive bio. I'm not going to get into it because people can go look it up. What do you think is the most important thing for people to know about you?
I really like the pursuit of craft and just finding in different places and trying to figure out what the future holds, understanding where I get things right and where I get things wrong. So yeah, that's why I was excited to come on too.
Well, I'm so glad you did. The audience will get from that a thing that everyone about you will know, which is you're incredibly humble in your approach to the world. It shows in the work and the curiosity you approach everything with. We were just before this talking about data centers in space and how they're going to manage thermal energy. Every conversation I've ever had with you shows tremendous curiosity and optimism about where things can go if we design them carefully. When you started Figma, the industry considered the space you were in to have been, if not solved, at least fully colonized. There was a set of tools, Adobe was established. Often when you see big shifts from an incumbent to an upstart, it's accompanied by a structural shift. What do you think it was? How did you find a niche that very few people saw and grow from that? This is a prelude to a question about AI. What did you see that everyone else didn't see at the time that allowed you to get to where things are today?
Yeah. I'm curious where the prelude goes. We'll get there next. I think it started with my co-founder actually, so I have to give him all the credit. He had an insight about WebGL, the ability to use the GPU on your computer in your browser. WebGPU is the successor technology. Evan was really deep on WebGL very early and showed that you could actually use it for real things. But showing that you could use it for a real big production application was different. We had to test that out. Then it was a question of where to go with it. Ironically, I'd been interning at Flipboard on the design team and the summer before as an engineer data analyst, building recommendation engines and doing some old school ML. With Flipboard, we were using a combination of tools, mostly Adobe Fireworks, which was originally Macromedia Fireworks. It got acquired. I left Flipboard and started working on Figma. We explored combinational photography but it could only get 85% of the way there until deep learning. Eventually we said we have to ship something for real. We focused on photo editing in the browser. Then we realized people would take photos on their phone, so why do it in the browser if it doesn't support RAW? Then Adobe killed Fireworks, so that was a starting point. We thought we'd build other creative tools in the Adobe platform world, but we ended up building into the product design and development workflow. We had two things wrong: we underestimated how big the market could get. We had an instinct that design was growing, but couldn't back it up with data. The Bureau of Labor Statistics had only 250,000 designers, which seemed small. But that instinct turned out to be really right. Design hiring grew enormously over the next decade. The instinct of WebGL led us to the cloud, and the instinct of collaboration came from our experience with multiplayer games, online communities, Google Docs, and Google Wave. We felt multiplayer could make sense, even if we weren't sure exactly how it would be used. That unlocked a lot of possibility.
Really having an insight about WebGL. A WebGL ability to use the GPU on your computer in your browser. WebGPU is the successor technology. Evan was really deep on WebGL very early, had shown that you could actually use it for real things. Now, showing that you could use it for a real big production application, that was different. We had to test that out. Then it was a question of where to go with it. I was advocating for drones in 2011, but Evan shut that down fast. He was right; drones would have been too early. For WebGL, the main question was where to go. I'd been interning at Flipboard, building recommendation engines. With Flipboard, we used Adobe Fireworks. Adobe killed Fireworks, and that became a starting point.
Two good ideas, by the way.
Yeah, I think drones would have actually been too early. For WebGL, the main question was where to go. I'd been interning at Flipboard the summer before on the design team and the summer before that as an engineer data analyst, building recommendation engines and doing some old school ML. With Flipboard, we were using a combination of tools, mostly Adobe Fireworks, which was originally Macromedia Fireworks. It got acquired. I left Flipboard and started working on Figma. We explored combinational photography but couldn't get all the way without deep learning. Eventually we said we have to ship something for real. We focused on photo editing in the browser. Then Adobe killed Fireworks, so that was a good starting point. We thought we'd build all these other creative tools, but we ended up building into the product design and development workflow. We had two things wrong: we underestimated how big the market could get. The Bureau of Labor Statistics had 250,000 designers, which seemed small. But design hiring grew enormously. The instinct of WebGL led us to the cloud, and the instinct of collaboration from multiplayer games and Google Docs made us think multiplayer could make sense.
With Flipboard, we were using Adobe Fireworks. That got acquired. I left Flipboard and started working on Figma. We explored combinational photography but it could only get 85% of the way there until deep learning. Eventually we said we have to ship something for real. We focused on photo editing in the browser. Then Adobe killed Fireworks, so that was a starting point. We thought we'd build all these other creative tools, but we ended up building into the product design and development workflow. We had two things wrong: we underestimated how big the market could get. The Bureau of Labor Statistics had 250,000 designers, which seemed small. But design hiring grew enormously. The instinct of WebGL led us to the cloud, and the instinct of collaboration from multiplayer games and Google Docs made us think multiplayer could make sense.
That's right. That's right.
And got acquired in. I left Flipboard, started working on Figma. We started to explore combinational photography, but it was basically 85% of the way there until we had deep learning and couldn't get all the way. Eventually we said we have to ship something for real. We focused on photo editing in the browser. Then Adobe killed Fireworks, so that was a good starting point. We thought we'd build other creative tools, but we ended up building into the product design and development workflow. So I think we had two things wrong: first, we underestimated how big the market could get. We had an instinct that design was growing, but couldn't back it up with data. The Bureau of Labor Statistics had 250,000 designers. That's not a big market. We had a sense that people were hiring more, but it was just small local examples. That instinct turned out to be really right. The way design hiring grew over the next decade was wild.
At some point we thought, wait, everyone's going to take photos on their phone, why do this in the browser if it's not supportive of RAW? Then Adobe actually killed Fireworks. At that point I was like, this is a good starting point. We can build out from there. At the time we thought it'd be all these other creative tools, very much in the Adobe platform world. What actually happened is we ended up building into the product design and development workflow.
And then Adobe actually killed Fireworks, right? At that point I was like, this is a good starting point. We can build out from there. At the time we were thinking it'd be all these other creative tools, very much in the Adobe platform world. What actually ended up happening is we ended up building into the product design and development workflow.
And at that point I was like, well, this is a good starting point. We can go build out from there into other places. At the time it was interesting because we were thinking it'd be all these other creative tools, very much in the Adobe platform world. What actually ended up happening is we ended up building into the product design development workflow.
And so I think it was like we had two things wrong. First, we underestimated how big the market could get. We had an instinct that design and the number of designers and the amount of software being designed and the value of design were all growing at a potential rate, and we couldn't back it up with data though. It was like the Bureau of Labor Statistics had 250,000 designers in the world. That's not a big market. We had a sense that people were hiring more, but it was just small local examples. Anyway, that instinct turned out to be not just right, but really right. The way that design hiring grew over the next decade was wild.
Yeah. You didn't hire McKenzie to come in and check. We were already freaking out about our $3,000 rent in Palo Alto.
Anyway, that instinct turned out to be not just right, but really right. And also the instinct of WebGL led us to cloud and collaboration. The native primitives from multiplayer games, online communities, Google Docs, and Wave made us think multiplayer could make sense. We didn't know exactly how it'd be used, but it was native to the browser. That also unlocked a lot of possibility.
There's so much there. Three things stand out: how many pieces of the future you and Evan were thinking through that would come to pass. Computational photography had a tremendous rise. Drones had a huge arc. Timing is everything. You pointed out Wave. Many people tried but the market wasn't ready. There was an explosion of design. When we started at Meta, Aaron Sitig and Cleo Quervo were our first designers for a long time. It was a small, highly leveraged team. You wanted to get time with them. At some point everyone figured out we should hire more. When we got to the point where we had so many designers, the collaborative workflow became a problem. You had people like Alex Ro going in, and Figma didn't exist yet. Shout out to Dreamweaver, my first one before Fireworks. We started building internal collab tools like Pixel Cloud. We heard about it from inside Facebook. Timing plays an important role. You guys executed beloved tools for a tough community. Being internet native was a huge part.
So you're using experience from multiplayer gaming and online movements to think about tools development. I was playing Neopets as a kid, and the idea of a community online made intuitive sense. Then we went to market and everyone was skeptical. Only one or two out of a hundred people leaned in and said it would be awesome. Everyone else was like, why would I want that? It was against the mental model. The first two folks who gave us a heck yes were Tim Vanam and Belinda Pino because they were doing pair design at Dropbox. They were the only ones. We launched and people were averse. I remember comments like, 'Is this the future of design? I'm changing careers.' 'A camel is a horse designed by committee.' But one engineer said if you're getting a lot of love and a lot of hate, that's good. Then a designer trolled us by setting up a multiplayer party in a Figma doc, and everyone joined making fun of us. It went on for three days. We were just trying to keep the servers from melting. It was a weird moment where people were making fun of the thing they were using, but they were proving the use case. I think that's really positive even though the tone was super patronizing. The worst response is no response. Having a clear point of view is important. When you have a point of view, you'll have people who don't like it, but that's because it's not consensus. Otherwise you wouldn't create disproportionate value.
Exactly. Right. So to some degree you had macro trends, internet native, you identified key technology like WebGL, but you were absolutely right about macro shifts. Software was eating the world, and someone has to design that software. Also, moving to the cloud, you have AWS, app stores, developer tooling getting better. It's all about how to make great software. Consumer experiences like Facebook and Gmail were raising expectations.
Well, and also if it's moving to the cloud, you have AWS. Totally. And if you don't have box software, you have app stores. Developer tooling is getting better. It's all about making great software. Ironically, given there were very few designers at Facebook at the time, experiences like Facebook and Gmail were starting to come into play. Expectations were changing. I'd go home and use these great products and wonder why all software isn't great. That was driving the change too.
Yeah. This goes back to Doug Engelbart. He wanted interfaces to be incredibly powerful and train humans for dozens of hours. We didn't do that. We chose to bring capability to people. There are counter examples like Dvorak keyboards and Wacom tablets that are hard to learn but give efficiency. You're a Vim wizard. When I first opened Visual Studio in senior year, I was outraged that I could have been tab completing. That's the transition. We're poised for another shift with AI. AI has a last mile capability to bring functionality to consumers without the arcane complexities. The meme about formatting in Word – everyone has experienced that. AI stands to unlock full capability without elaborate training.
Yeah. I'm a Vim wizard, although I don't know why it was so important that I learned Vim. The first time I opened Visual Studio in senior year, I was outraged that I had been using Vim. That's exactly the transition. We're poised for another shift. AI has a last mile capability to bring functionality to consumers. Even with today's technology, it's incredible what can be done with prompting. If you're really good at prompting, you can do incredible stuff without fine-tuning. When you get into applications, you get to a world with more efficiency on laborious tasks, but also people can create more. The critical mindset is that it can't be a crutch. I had dinner with a family and the son used AI to write his homework, while the daughter used it to drill her on problem sets. That encapsulates the dichotomy. If you use it as a crutch, you're not using it to max potential. Use it as a source of inspiration to think further and learn more.
Yes. You get a stratification where some people get the minimum viable bar faster, but for others it raises the ceiling. That's interesting. The market for designers grew, giving you a bigger customer base, and new customers are more likely to adopt new tools. There's also an interdisciplinary trend. At Meta, we had design-oriented engineers. Over time, designers started coding and coders started designing. LLMs help with that because it's easier to get inside an iterative design cycle with the help of a model.
Approached working places like Figma. There are some other new kind of fun companies like Korea and a few others that do different pieces. I think it's pretty interesting. I'm curious for you: is this a threat? Is it an opportunity? Is the bigger market a lot more people playing the game and that's cool? There's also kind of a muddling of who the market was before.
Yeah. How do you guys look at the growth of people for whom these tools are relevant?
Yeah. I think we both saw AI coming.
Uh, you know, well before ChatGPT, I was in my head.
And I'm like, okay, is this a 10x, 100x for Figma? Is it a 110th, 100th? Like which direction is it headed? And I think where I got to eventually is that the framework has to be okay. As the models get better and as what you can do with the models gets better, you have to get better too.
And that's the basic rule.
And so if there's some place in the business where that's not true or could not be true, if model capabilities improve, you have to fix it.
Yeah. And so, I think though that it's really exciting if people can get their ideas out. I mean, that's like the entire premise we started with.
That's right. That's the point of Figma, right?
Yeah. Fred Victor was a huge inspiration. One of the things he said was that when an idea goes undiscovered or falls by the wayside because someone can't bring it to life, that's a moral wrong. I really thought that was an interesting propagation.
Now we're in a world where you can go from imagination to something being real very quickly. And you can have your idea evaluated on its merits rather than on the visual style. That does not mean your output is going to be what you want to ship.
If anything, the role of designer is greater because the thesis for the last decade has been design matters more. That was the thesis before AI, that's the thesis now with AI too. You go look at something like paper where you put that stake in the ground, this is the future. The ideas percolated, that caused your products to stand out. You had a point of view, impeccable visual craft and execution. It was dynamic.
You had a point of view, impeccable visual craft and execution. It was dynamic. I see that across many of the products you pushed over time. That attention to detail and making sure it's a great experience.
And I think the investment in that craft, design, point of view, and brand only increases right now because that's how you stand out. It's a really important moment to be leaning into that.
A lot of creatives have been very spooked by the models getting better. But I think what will happen as we look back 10 years from now is that model outputs are really more of a new medium. You can shape that medium like clay, create processes to transform and mutate outputs to get to something incredible. The first prompt should not be the final destination.
We recently acquired a product called Weevi, now called Figma Weave. It creates a node-based workflow to chain together different models and outputs to get to a final destination. That's what's needed for the future of creativity.
There's a bunch of interesting pieces here. The beauty of founder companies like yours is you have a value system that is the product. It's hard to have a thing be a threat because it's either useful or not. If it's useful, we're going to use it. You're able to embrace new technology because you have a value system in empowering people. AI is a tool that helps people achieve their vision. Fighting it is futile.
There is a tremendous care for the craft that people are worried will be lost with AI. Craft is not a consensus.
Craft is very specific. We love the little details that upon close inspection reveal thoughtful design. I do expect AI to keep getting better and more interesting.
My prediction is that as AI models can show higher craft, it will elicit more thought. The editorial role is key: we edit and decide among controversial high-craft directions.
This is one of my constructs for the history of technology. We keep thinking we know what we are as a species and then inventing things that do that thing and finding out we are something else. It's a dialogue. Modern art was a repercussion of photography. It didn't eliminate painters; they continued to thrive, and people went in new directions.
And you know, if you think about art right now, I think the reactions to AI will lead to new art movements using natural materials.
There's always a reaction to what's going on. In design, we've been in Swiss minimalism for so long. Should all software be Swiss minimalism? That seems wrong. Think back to the flash era of rampant creativity. What if you had high craft and an explosion of creativity?
Yeah, I've seen more maximalist software design online. AI art will become its own media type. There's a reaction and people rebalance. We seek the craft at the edge.
What are the things you see as unchanged? Where is there no shortcut, where AI isn't coming to save you?
I'll build up to it. What is unchanging from a human perspective? We've had millennia of evolution. We still seek social connection. People will be looking to make an impact. The pursuit of craft and creating things is intrinsic to being human. That doesn't stop.
There's a global news structure that can make people feel like they're not making a difference. But at the end of the day, we're eating, sleeping, creating family units. We like to be entertained. We're crafting society. The inertia of society is real.
We like to be entertained. We're finding out how to spend our time.
