品味是你的护城河（Figma 的 Dylan Field）

Hey everyone, welcome to the Latent Space podcast. This is Alessio, founder of Colonel Labs, and so happy to be at the Figma office today with Dylan Field. Welcome.

Thank you. Thanks for having me on the podcast, and welcome to the Figma office.

Yeah, you know, we almost couldn't choose where to do this because there are so many beautiful spaces in here, but we finally decided on this corner. I'm super excited to have you on today. I was reading through some of the history of Figma, and your initial mission was to close the gap between imagination and reality. If I heard that today, I would assume it was the slogan of one of the vibe-coding platforms. Maybe talk about what was the first “we should take AI seriously” moment, where you were like, okay, imagination to reality. In the first phase of Figma, that was helping designers bring what they had in their mind into a canvas, and now with Figma Make, you're obviously moving to a much broader audience. What was the journey to get there?

Yeah, I mean, if you go back far enough, AI showed up in different forms for Figma. I had the chance to be on the data science team at LinkedIn as an intern prior to working at Flipboard, getting more into design, and then starting Figma. We were doing a lot of classical machine-learning approaches, and I was absorbing that. There was plenty of discussion about agents back then with my mentor, Pete Skomoroch, and thinking through what it might look like if some of the ideas from the ’90s were to resurface. Those were just fun, geeky conversations that were pretty abstract because obviously the world wasn't there yet.

Then, back at Brown with Evan, my co-founder and our original CTO, who's no longer at Figma but is an absolute legend—we were talking a lot about some of the stuff we were starting to see with machine learning and computational-photography approaches to doing image editing and what could be accomplished with that. For example, there were papers being written about how to use Internet-scale data to complete scenes and make it so you could basically do the equivalent of content-aware fill, but instead of doing it in an algorithmic, deterministic way, do it based on the entire Internet. We thought that was a pretty fascinating concept. There was a professor at Brown who was doing some cool research in this area.

We were also getting very excited in the early days of Figma, before we even incorporated, about things like how to turn a 2D image into a 3D scene. So, more computational photography, plus alpha blending and some of these early techniques that got you about 85% of the way to something awesome, but not 100%. It wasn't until we really had deep learning that you could get to 100%.

All of these individual demos that we were able to work on—and by “we,” I mean mostly Evan; he's the real genius in the equation here—as we started to explore a bunch of these areas, it felt like there must be some way to make creation easier. That's why the vision was stated as “idea to reality,” and not “idea to X” as a subset of reality, because we thought you could actually do this for a lot of different areas, and I still do.

Fast-forwarding to today, with Figma Make, for example, we're really trying to make it so that you can go from an idea in your head to an actual shipped product as fast as possible. That might take the direction of an internal prototype to explore different ideas. It might be an internal app that you're using. This morning, I was doing some work on random data munging, but I was using Make for it, which is kind of fun, rather than writing a Python script.

I think it's very exciting to think about how far you can help people go and how you can make them both more productive, but also help them explore more of the option space of design with some of these techniques. Then, of course, we're also excited about what that means in Figma Design as well. How do you prompt to edit, prompt to do generation, and do it in a way that's consistent with everything else that's in your design system and the patterns you're using? How do we actually infer from what's already inside of Figma what you want to do, and really be expansive in the way that we understand your intent?

So you have a background in, obviously, math and computer science, and now you run Figma. You have this kind of duality of aesthetics and code. Were you first AI-pilled by image generation, the more creative things? Early on in the podcast, most people would say Midjourney was their favorite AI product, and another half of people would say GitHub Copilot. What was the first product that you fell in love with in AI?

Not a product, but my first AI-pilled moment was, I think, 2014 or so, maybe a little bit earlier. I was a Thiel Fellow, and in my class were a number of amazing people, one of whom was Chris Olah. Chris and I would go to these retreats together for the Thiel Fellowship every 3 to 6 months, and I remember one of them, Chris had been working on some cool 3D-generation stuff [?]. It was all a bit out there and not clear how it would be productized.

At some point, he was just like, “Deep-learning neural nets. This is the future of everything.” I remember him sitting down with me. We were at a wooden table outside in some Santa Cruz nature setting, and he was on the Wi-Fi, which was super slow, connecting to AWS. He said, “Look at this. I can go on AWS and spin this up, and I can train this tiny neural net to classify handwritten digits.”

I was like, “Chris, this is a solved computer-vision problem. Why are you excited?” And he was like, “No, you don't get it. It's a neural net, and there are hyperparameters. I can tweak them. I think I can actually make another neural net to figure out how to tweak the hyperparameters.” I was like, “That's all great, but this is a solved problem.” I lacked the vision at that point to see where it was going.

It started to get me to pay more attention. Then, watching his work when he was at Google and some of the great blog posts he was putting out, as well as listening in on more of the people around me and the conversations that were happening around AI and machine learning, got me more and more excited about where this might go. But I don't think I truly internalized scaling laws for quite a while longer and what that could mean.

I think GPT-3 was probably the first time that I was like, “Wow, the delta between this and past models is so great. Something exponential is definitely happening here. It's not just hype.” Then, plenty of conversations around that time with other AI figures that we both know well definitely started to make me think, okay, there's something very important to focus on here.

I think it's very different to be in the context of more deterministic software building than AI research. There are completely different motions in how you run those teams in those areas. So it definitely took us a lot longer than starting at “Okay, GPT-3, amazing” to get to the point where we're starting to ramp up and push the boundaries of what might be possible at Figma.

No, that's great. I would say Figma Make is one of the most impressive releases I've seen this year. I was playing around with it the last few days. I built a Figma clone in Figma Make, so you guys are cooked.

Exactly. Or are you so back? I don't know.

I know. I don't know. It's so hard to keep up. It depends on the day. Tomorrow, that might change.

But to me, there's this interesting triad in software engineering: you have the test, you have the spec, and you have the code. Usually, if you have 2 of the 3, you can generate the third. I'm curious how you think about the Figma model, so to speak—the almost Figma data model. You have Figma Design, which is where the visual work happens; then you have Figma Make, which is basically, in my mind, the bridge between the design and the code; and then you have the Figma MCP, which is how you bring that into code in a way that's not even UI-driven—the model is kind of doing the work for you.

Does it feel like it's changing, in a way, the tools that you need to build? How do you think about using AI for editing the design and whatnot? Do you feel like natural language is becoming more and more the interface, even in design, and that the work is going to be done that way? What are the pieces in your mind?

Yeah, lots to unpack there. I'll start with just: Is natural language the interface? Yes. Right now, I've said this before, but I really believe it: I think we'll look back on this era as the MS-DOS era of AI, and the prompting and natural language that everyone's doing today, I think, is just the start of how we're going to create interfaces to explore latent space.

I just cannot wait for an explosion of creativity there, because I kind of think of these models as almost like an n-dimensional compass that lets you explore this wild, unknown fog of war in latent space. You can push the models in different directions through natural language. But if you have a more constrained end state and you're able to dimensionality-reduce a bit so you can push in different ways, there should be other interfaces available than text. These might be more intuitive, but they also might be more fun to explore.

And I think sometimes constraints unlock creativity in ways people don't expect. So I'm excited for that. But right now, yes, natural language is where we're at. While I'm excited to push that forward and meet people where they are, I think that's usually a good model for product development before you get to the point where you've really refined it.

Going back to your triad, maybe we can start with the spec. I think the notion of a spec is evolving so much right now, and what should be in a PRD versus what should be in design versus what should be in code is much blurrier than it used to be.

It used to be that we had this very waterfally process of, “Oh yeah, we're going to go gather some requirements, and then we're going to go make a big doc, and then we're going to go make some designs, and we'll code it up until we feel it's ready. Maybe we repeat a few times.” But it was a process, and I think with Figma you can absolutely follow that process. We also recognize that roles and stages are blurring. As all that blurs, how do you actually support different ways of working?

You might want to make a prototype as part of, or in place of, a PRD. You might actually want to focus more on the design as a high-fidelity descriptor of what this could mean if the cost to make designs is lower. I think the more that you can expand that option space for people and bring them into a surface to align design and visual fidelity, that might be the place where you can align best.

There's also the question of, okay, how far can a spec get you, and why is a spec different from code? If code is the complete spec, in terms of it being the most determinate, clear way to show intent of what should happen in every edge case, how much of that can be inferred, I think, is an open question—one that we'll all be thinking a lot about soon.

If you think about the value stack overall, it feels to me that the better code generation gets, the more design matters, and the more that the human pushing on design matters too. Even if you have a good starting point from your design system, from AI generation, whether it be code or image, I think you need to push design forward—not just as an individual screen, but as a system—in order to actually compete, differentiate, and win.

It's been our thesis for a long time: design is a differentiator. But I think it's even more true in this world where we're at now, where the rate of software creation is going exponential and maybe even vertical. In that world, you have more software and more competition. So what wins? It's brand, it's point of view, it's taste, it's craft, it's design.

If that's the world we're headed for, which I'm very confident it is, then it's not enough just to use AI to generate an output. I think you have to push further than that and really get into the detail and the craft, in addition to utilizing AI to explore the option space faster so you can go as deep as possible in the direction you choose.

Yeah, you know, I only have the Pro plan, so I don't have Code Connect. But I'm curious how you think about that, because Code Connect—the whole idea, in my mind, was, “Hey, instead of having to make sure that the code stays in sync with the design, we kind of build this bridge between the two.”

But now, if you have the design, you can, in theory, regenerate the component every time anyway. So why add this additional layer that, before, was there but is not needed anymore? To me, that's the most interesting thing. I was like, what's going to end up being the source of truth, and what are the two-way bridges?

For example, right on Figma Make, I use the MCP to bring that code into Cursor, my actual codebase, but there's no way yet—I'm sure you'll do it—for the MCP to write back into the design and say, “We actually ended up implementing it this way.”

I'm curious where you feel like the center of gravity is going to be. Obviously, you're biased in a way, but as an engineer, I'm curious about your thoughts.

Yeah. Well, first I'll just explain Code Connect a bit more. To expand on what you already said, I think there are different situations that you might find yourself in. You might be going from 0 to 1, making a prototype of something that's rather disposable. You might be working on a personal project where you're not making something that's disposable; you're building on something that's existing, but the codebase is small. It's pretty clear to you what's going on, and there aren't a lot of patterns that exist.

Or you might be in a pretty large codebase where there are a lot of existing patterns, a lot of code, and you're trying to fit those patterns. Especially in that last example, as you get to these larger codebases and larger settings inside of companies, it's very important to be consistent with existing patterns in the code. But it's also important to create a design system where you're able to create consistency at scale for designers and make people more efficient, so they're not always recreating different buttons and so on.

There's the world of Figma and design, and there's the world of code. There are advantages to having a source of truth in Figma and a source of truth in code. In some cases, we look at libraries that customers make and they are one-to-one: the design components perfectly reflect the code components in the codebase. In other cases, you're working on the thing that's next in Figma, and that's not yet all built out in code. Both cases are important.

In the case where you are one-to-one and there's more of that connection between components in Figma and components in your codebase, you want to define a formal mapping. That way, you're able to give context via MCP, make it so that developers are easily able to implement a design on the front end, and Code Connect serves that goal.

We're investing a lot to make it easier to set up because right now it's too much of a pain, but we're also trying to get further in terms of how many people can use Code Connect. In terms of where the source of truth lies, I think there's a variety of ways it will probably play out in parallel.

I think that it's okay if, in some cases, code is the source of truth. You'll see us do a lot of work to make it so that you're able to bring your codebase's design system into something like Figma Make or Figma Design. But if you're wanting to rapidly iterate and express and try out different visual explorations, and if you think visually—if you're someone who's not necessarily maximally comfortable with code—I think a visual surface is very important as a place to explore.

I think there are different modes of thinking, and it might be the case—it's likely the case—that the visual metaphor is easier for a wider set of the population to grok than to go into code. I also think that as we move forward in time with more agents writing more parts of your codebase, you will also be less familiar with the code. So then you might want a different abstraction where you're able to work on things and basically plan out what your app should be, what your software should be. Figma can provide that.

Yeah, I almost think of Figma as the context repository for aesthetics. To me, as an engineer, the design product isn't even that helpful or useful in a way, as long as you can generate the components and I can do small tweaks.

I think one of the big tailwinds that Figma has is—pardon the pun—Tailwind CSS, bringing more of these classes, named classes as a way to define style, along with the way Figma variables and your system are set up. For me, once I saw Figma Make, I'm like, “Okay, now I get it.”

Before, when I had the blank Figma canvas, it was like, “I'm not talented enough to start from here.” But if you can build an initial thing through AI, then I'm good enough to tweak it, have that be the bridge, and then take that into Claude. I can take that into Cursor, or maybe I'll just stay in Figma Make forever and the prompts just go there.

Yeah, I think what you're saying is super important. The blank-canvas problem is real. We're always trying to figure out, for Figma Design, how do you make it less intimidating for someone to come in? How do you make this more approachable?

There's always tension between the power users of Figma Design, who would like every single power feature that you can imagine—why can't you make a feature for everything in the CSS spec? Of course, we can over time—and someone coming in for the first time, technical, nontechnical, whatever. Are they intimidated? Do they feel invited to go create something?

Regardless of our UI, the first thing that can block people is that blank canvas. Getting people from the place of “I have an intention” to actually putting something on that canvas is so important. Once you start getting people in that loop, then it's less intimidating and you have more that you want to explore.

One of the things I'm excited about that we actually just shipped today is a way to copy designs from Make into Figma Design. If you think about Figma Make as just an easier entry point for Figma Design—it's like the flight simulator to the airplane cockpit or something—then perhaps you're able to make it so that's an early entry point you go through. Then you actually can do more than just tweak components; you can visually manipulate a design.

And you might find that you actually can go faster that way than if you're doing it in code.

Yeah, I think—I don't know what your model internally is between diffs, but to me, it's easy to communicate what a diff in code is, but it's kind of hard to communicate design in a way that I can then put into an LLM to apply. So, I'm curious how you think about that. The design is obviously more than the sum of the components, right? If you just took any piece that's in Figma individually, it doesn't look like anything. When you put them together, it looks great. How do you see the way people communicate design also changing now that more of it needs to become language because of the interfaces?

Yeah, I think I might push back a bit on the “more of it becomes language” part, but maybe we can explore that later. It depends on exactly what you mean by that.

But in terms of the way that we represent diffs, you can go to version history in Figma and see sort of composites of diffs. You can make a new version at any point, and of course, internally, there is the journal of every single edit. I think there is an opportunity there, to your point, around how you basically use not just what's in Figma Design as a source of truth, but also the tool calls made via MCP and when they were made, to understand what context has changed since I last made a call.

So that's an opportunity that I think is a smart one to point out, but I also think it's interesting to think about the journal data and what can be possible in terms of thinking about what you might do next, how to help you be more efficient, and help you explore more ideas.

Yeah, my natural-language point is, if I give an LLM a prototype, it can understand from the code what type of changes I've made. I'm trying to figure out—and this is just because I'm not a Figma power user—if I gave an LLM a Figma JSON or diff, could it understand aesthetically what has changed, to then update maybe other docs or parts of my system?

Yeah, right now that's not there. I think it's a really good idea, and I think it also mirrors other ideas that are going to be important, too. For example, if you pull in Figma context into your IDE or your agent, whatever format it takes, let's say that the design is not perfectly implemented, because currently that's where we're at: a lot of times it's, “Wow, this is a great start, but there's more work to do.”

Well, what's the delta? How do I get to the point where it's perfectly implemented? And that requires some back and forth, too.

Yep. Yeah. The only thing I got wrong when I did Figma to Cursor was the border color. Everything else was right.

I'm glad to hear that. That's awesome.

I was very impressed.

And to be clear, some of the time it does just work.

Well, yeah. No, exactly.

And it's magical. By the way, I'll point out that's not a comment on, “Oh, front-end engineering is dead.” There's so much more to front-end engineering than just that translation step, right? That's kind of the most mechanical part. Actually thinking through all the states, all the intended behavior, and how to make the design truly come to life—that is a lot of the interesting work.

I'm excited for how interfaces will get much richer and much more interesting as we make it so that they're easier to go from design to code in the first place. That's just the first stage, right?

Yeah. When we had Greg Brockman, we talked about these purple and blue gradients kind of taking over the web because of all this training data. Do you feel a sense of responsibility, in a way, in Figma Make also, to set the new standard for what things should look like? And how much do you think about making that explicit for the user?

It's like, “Hey, you need to make some aesthetic choices early on.”

Yeah, I feel very strongly that across the Figma platform, we should do our best to find ways to help people explore more of the space of aesthetics rather than impose a personal viewpoint on aesthetics. That's a hard problem, but if we can accomplish that, I get very excited, because if you can actually open that space up more and figure out how it applies to software, to product design, then not only could we be in a place where we help you generate high-quality visual output, whether you're a trained designer or not, but also we could get to the place where you could be nudged into—or nudge yourself into—different directions that are underexplored relative to the design community and design history.

There's an ability to interpolate between different styles and different ideas, and AI can help you do that, but the designer can then take that so much further.

Yeah, I think regurgitating the median website is maybe where a lot of us are today, but where we need to get to is a place of really pulling out new styles.

The other thing I'd say is, you look back at the Flash era of the web—you know, we both grew up through that—and it was maybe not always high quality, but dynamic, exciting, fun. This was an era where experimentation was happening. Then, at some point, it was like, okay, Steve Jobs gets up onstage and goes, “Flash is dead.” Here's my new world, and—brief skeuomorphic phase, then Swiss minimalism—and we've all been Swiss minimalism for quite a long time now.

I just think that, again, going back to, “Okay, more software created than ever; what needs to happen?” Designers need to push us forward, and that's going to mean an exploration and explosion of creativity and so many different visual styles that will be explored, so much more dynamism in interfaces, and new patterns emerging, especially as you start thinking about all the screen targets we're going to have. Those are going to explode, too, and states—all the different surfaces that will be created—and designers will have to think through systematically. That's a big challenge, a big opportunity as well.

Yeah, I started Kernel Labs a month and a half ago, and James, our designer, I've been working with him on our initial landing page and our product design. He's almost like my Latent Space shepherd. Even if Figma was AGI, I still wouldn't necessarily know what to ask for.

I think that's really what people get wrong a lot of the time, which is that software is the same thing: you could give a software AGI to anybody, and it doesn't mean they could build great software. To me, that's the most exciting thing. But what I really like is the ability now to reuse this across surfaces in a way that wasn't possible before, because I can create, ideally in Figma now, my design system.

Now you have Make to generate all new types of products. You have MCP. You also have Slides, and you have different products on the other end of the spectrum. Again, going back to being the context repository of aesthetics, I know that as long as I have the context from Figma, whatever the AGI—or whatever I'm talking to—is going to generate, it's going to have some sort of rooting in what I think looks good.

Yes.

I think some people have that personally. I think some people should have some sort of personal Figma, almost, whereas when you're generating images on ChatGPT or Midjourney, you should have some sort of aesthetics to draw from. How do you think about that evolution, because you're going from a world where only designers work on your product to now it becomes a core part of a lot more constituents?

Yeah, I think in a world where design is the way you win, it's only natural that we need to get more people involved in the design process. That is not going to diminish the role of designers. In fact, I think it expands the role of designers, because then you have to shepherd people through the design process and help them go from—not even being aware of design, kind of blindly going through the world—to, “Oh, man, aesthetics matter,” to, “Oh, can we make it pop? Can we make it cool?”

Then people start to actually think about, “Well, wait a second. I'm looking at one screen. What's the actual experience here? What is the entire flow?” Then it's, “Okay, let's take mental models of how we can think about this experience and consider it in different ways. What are the potential different paths, metaphors, and experiences that we can create here, and what are the abstractions that matter?”

From there, it's like, “Okay, wait a second. This all exists in the context of our brand, the greater culture of the moment, the business constraints, and all sorts of other things you might be optimizing for.” I think more people coming into the design process can help add context as well.

There's no reason why someone who's outside of design or doesn't call themselves a designer—whatever they identify as: engineer, product manager, CEO—everyone should be able to come in and say, “Okay, here's an idea.” The idea hopefully could be parseable in high fidelity to the standards of at least a design system and consistent, so it's not distracting, because an idea should be evaluated on its own merits.

But then I think from there, the actual exploration and making that great—that is a hard design task still. So how do we lower the floor for everyone coming in, but also raise the ceiling and make it so designers can do even more and produce even greater work?

I think, obviously, it’s cliché: everything is changing. But I think there’s a fundamental shift in both how people perceive software. Sam tweeted; someone tweeted about the fast-fashion era of SaaS, but there’s also a negative connotation to fast fashion. In software, though, if you can get the software that you need at any moment, that’s not a cheap thing; that’s an expensive thing that can now be made cheap, and it’s still high value.

I’m curious how different products are going to drive that. Even though it feels like, “Hey, I just created this for you very quickly,” there was so much that went into that, which is maybe sometimes undervalued. So I’m curious if there’s something you think about where, in a way, people should come to Figma and do a lot of work, but maybe we can help you, from all the work you’ve done in the past, come to the right result much faster in the future.

Well, I think especially in the context of teams where they’ve done a lot of work in Figma, there are patterns that, with that team’s consent, you can tap into and figure out how to help improve outputs. But I do have a little skepticism about the fast-fashion interpretation.

I just think that, where we’re currently at, at least with the models—and of course, we’re on this trajectory, whether we’re on an S-curve, an exponential, or it’s an S-curve that’ll turn into an exponential, I don’t know. Maybe you’ve got a point of view—but I was kind of like, “Okay, I’m excited for the ride.” As long as I’m here today, my mental model and strategy is just this: your strategy should always be, “Okay, assume AI models get better and make sure that makes Figma better.” As long as I believe that’s true, I’m happy. If not, change strategy. That’s the algorithm.

In terms of that interpretation of where models are going, I don’t think the world is in a place today where the fast-fashion era is here. I also think that so much of designing software is doing it in a way where many people can use it. It’s rare that you’ve got an individual piece of software that is truly just for you.

I think it’s awesome that more people are exploring their ideas and creating software and tools for themselves. But then the next step, if they want to go further, is, “Okay, how do I make this good for other people, too?” In a setting where people are trying to learn software, most people learn it from other people. So now you’re in the same place you were before: you had a piece of software, you made it just for you, and you decided that the problem you solved actually applies to other people. Now you’ve got to have something that is probably consistent enough to actually share with others so they can learn it and it gets adopted.

I don’t know. I think it’s like, yes, more people will create software. That’s awesome. But I’m also not sure that software will just be disposable. If you look at the way that people work, whether it’s with Claude Code, Cursor, or Warp, so much right now is, as you said, you need to have some expertise about how software is built that lets you discretize the task, just like you would for an intern.

Maybe it goes beyond intern level, but still, I’m not saying, “Go build Figma, and your agent is just going to go figure out all the complexities of Figma.” I think that’s just not something I see happening in any near-term future, even as longer-running agents start to occur and we get better capabilities. That’s a long way out.

Now, maybe that’s a high bar, but you look at the actual workflows that happen with the very big SaaS applications. Consider Workday or Salesforce, for example. A lot of CIOs would love to go, “Yeah, I’ve vibe-coded Workday, and I just saved my company all this money,” or whatever. But you actually peek under the hood of Workday or Rippling, and these are very complex pieces of software that have accounted for every edge case that you can run into as you’re thinking about your HRIS and the platform of data that you can tap into, and then go build out from there into different workflows.

They’ve done that through, in Workday’s case, decades; and in Rippling’s case, not quite a decade, but also with a lot of prior knowledge about what needs there are in the market. It was very intentionally built. I’m skeptical that, without that knowledge of the workflows that people will encounter, people will actually make something that can scale.

I think you run into the same problems that you’ve run into all along, and then it’s a loop. Maybe that loop goes a little faster, but it’s not like it’s just going to replace software. I think that’s the bull case for software still being helpful post-AGI, whatever that means.

Because, in a way, you still need to prompt the AI. I think you’re going to end up having these interfaces that, again, just like you’re going to help people explore the latent space in design, there’s going to be a way for interfaces to compress the way that information gets passed through the system.

I think in data analysis you’re kind of seeing a lot of the applications going away, in a way, because the models are so good at it and it’s so natural to do on a conversational level. But again, sometimes it’s like, well, how do you give it the right data? How do you ask for the right type of charts? How do you ask for the right follow-up questions?

I think there’s a good question of: at what point is a piece of software, and at what point is it just, again, a layer that helps you project your interest into the model?

Yeah. And it’s interesting. If you look at data analysis, maybe break it up into a few things that have to go right. First of all, you need to have trust that the right queries are being written in the correct way, and that’s a lot of trust, because if you get that wrong, you have a bad, shaky foundation for the rest of your experience.

Then there’s, “Okay, what’s the next query?” I’d probably be more bullish about AI predicting your next query, or a follow-up, than I am about a 100% rate on the query being constructed correctly. If you were to show people some prompts around, “Here are example next queries you might want to run,” that might spark other ideas they have for follow-up questions that further their analysis.

But then there’s also how you display the data and the visualization itself. There are canonical visualizations we’re all used to, but I think visualization is fascinating, and we’ve only scraped the surface in terms of how we can visualize data.

Especially as you get to larger data sets and more complexity, and you’re really trying to communicate data to people, that is one of the most interesting design problems out there. How do you communicate how much of the federal government’s budget is spent where? People have tried so many times. I’ve never seen something that is clear and actually gives you any sense of scale that you can relate to.

How do you communicate the data inherent in biology to a layperson who hasn’t studied biology? Again, people have tried. It’s a very hard problem. Maybe you break it into subproblems, but there’s still so much to push on there.

Yeah, yeah, yeah. Exactly. There are almost two sides to it. For me, how do I communicate to the model what I need? And then there’s also the other side, which is that the models have so much imbued into them that we need to get out of them, and we don’t quite know how to do that.

One part is information communication; the other part is, once I prompt it, how do I get the response in a way that I can really parse? I don’t know. It’s been breaking my brain for the last few months, just thinking about what will feel like software and what will feel like a conversation.

I know that, obviously, OpenAI has this big goal of being your companion, and you have voice mode and whatnot. But at some point, you just need something that is beyond a component rendered inside a chat interface, and it’s hard to figure out.

Especially when you think about—I know you do a lot of angel investing—so I’m also curious how you think about startups and what kinds of products are now possible that maybe weren’t before, and what products people should stop pursuing because you think they’ll be part of the models.

First of all, I’m hesitant to give advice on that, because I think the idea that all software, or lots of software, will exist in a session with an LLM or any model is a little overblown.

There are so many different ways this could work out. It could be a back-and-forth where the model is the origin of a request and then you go elsewhere. The model could be embedded in software, but you still have a destination you go to first, and it’s not the model. We can think of many other ways, too. I think that’s maybe a shaky assumption.

Also, it’s often the case that you’ve got some space where a thousand people are starting a thousand companies, and everyone’s going, “Don’t go there; it’s too crowded.” But then one person comes up with some really new, clever idea and a totally different take, and they propel from there.

I never try to say, “Don’t do something,” to an entrepreneur, because somebody listening to this podcast is going to have some insight in a space that both of us think is really dumb to work in, but then they’ll be the next trillion-dollar company.

So, whoever that is, let me know.

Right, yeah, let us be a part of it, at least.

Let us both know. But yeah, I think there is some amount of memetics around people seeing other people doing something and following on, and you have to have a unique insight.

If you're starting a company or working on a product, it usually should be something that's unpopular, right?

And this is just cliché advice at this point, but I think there's something deep to internalize there about the contrarian nature. Going more Thiel language now, but I think he's basically right about this.

If you're investing in something, unless you're just going after momentum, which a lot of people do, you need to have some point of view that a lot of people would just blanket disagree with. It should be scary to you if you're investing in something and you tell your friend about it, saying, “Here's my point of view on why this is really cool,” and your friend is like, “Oh, yeah. I totally agree. Makes complete sense.” That should be a warning sign if you survey people and they're all saying the same thing about that.

Yeah. What have you learned about yourself during the Thiel Fellowship? What are things that you changed about how you approach—yeah, just life, thinking, learning?

I think, whether it be that interaction with Chris Olah, where I look back and go, “Man, I maybe dismissed that one too soon,” and then learned over time, thankfully. Or another example is, in 2013, there was a Bitcoin hype cycle. Bitcoin went to $1,000 or something, and half the Thiel Fellows at the time were really excited about Bitcoin. I'm just like, “These idiots. How do you short this thing?” That was my default reaction.

I think the overall meta-lesson that I've learned, over not just the Thiel Fellowship but just being around tech for a while now—I was paying attention even as a kid. I was in some commercials, like Microsoft and Tiny Toons, for example, and so then I'm starting to track, like, what's this dot-com bubble? And wait, why is that? When am I not getting residual checks anymore? That became real from a monetary standpoint.

Then, working in high school at O'Reilly Media was a great way to get exposure to some of the starts of cycles. The Thiel Fellowship taught me the meta-lesson that I've learned: don't look for reasons why things are not going to work. That's important, too, but it's not the place to start from.

The place to start from is, “What could this be? How big could this be? How important could this be for society?” Let yourself imagine and dream, and then go and think about all the ways it's not going to work so you can mitigate each one. But start with the dream. I think if you start there, it's just a better default position to go from.

Yeah, I'm really worried about the X algorithm and what it has done to optimism, because it's so easy to get likes just by being negative about things.

I do think the X one seems to have changed a little bit recently. Maybe Nikita's in there tweeting things.

I'm thankful for that. But yeah, I do think algorithmic feeds reward controversy, and being negative is a way to get controversy. But also, I don't know, I'm default optimistic. I feel like society just builds antibodies to different things over time.

I mean, remember when we were all worried about, “Oh my God, everyone's going to be a zombie playing FarmVille all day?” Well, here we are. Some people still play FarmVille, but—

I don't know anyone that does that all day long. Most of us kind of forgot about the social gaming era.

So, when I drive by the Zynga building, I'm always like, “I remember back in the day.” How do you use X? You were famously responding to product feedback on X on IPO day. What's your routine for staying on top of that?

I try to search for Figma a lot and see what people are saying, but I've also trained my algo feed to show me a lot of stuff that's relevant to Figma. There were ways—I don't know if they're still as powerful as signals, with whatever algorithm shifts have happened—but you kind of find out what signals matter, algorithmically, right?

“Not interested in this post” seems not to do anything. A like or a bookmark, I'm not sure how much that matters, but “Copy link” turns out to really matter as a signal, or at least it did. I wasn't always sharing the link, but I'd copy it whenever I saw something I wanted to signal-boost to my feed.

The more you learn as far as the algo feed, the better you can train it and the more useful you can make it. I think feedback across any surface—not just social media, but support, sales conversations, conversations with our community, gathering people together, and just talking with folks—is extremely useful for our team. Research, both qualitative and quantitative, is also extremely useful.

I think of intuition as a hypothesis generator. Then you have to test the hypothesis, so using feedback to be part of that test is important. I'd always rather give feedback to the Figma team by surfacing the voice of a user rather than saying, “I have this point of view.”

You should do the latter as well, but the former is my preferred method. I'd much rather champion user feedback, a user bug report, or a feature request, and then dive in with that person than just have it come from me.

The other thing is that I'm always looking for those visionary users who are a step ahead of everybody else. They know, intuitively, what is needed. When you can find them and separate them out, that signal is amazing to get.

I remember, in the early days of Figma, there was this one user test that I literally brought a bottle of wine to because Figma was so slow at that point. To complete the user test, I knew it would take hours, so it was kind of tough to administer. We went through the bottle of wine during the user test.

The person we were doing the user test with was a guy named Peyman, who was an amazing designer and was then working at Coursera. The next day, he followed up with this super-long document. It was 8 to 10 pages, and it laid out a lot of what ended up being our roadmap.

We didn't literally follow the document, but I look back and compare and contrast, and he's an example of someone who was a visionary user. There are a lot of people who will give you more local feedback, but some people can see the big vision, too.

That's always really exciting because it's validation for you and the team about where you should go, but also a source of new ideas and insights.

How did you think about hiring and building the team back then? I remember I was working at a YC company, and we were all on Sketch. We were all on Sketch, right? I think a lot of people were maybe like, “Why isn't Figma just going to be like Sketch?” and so on.

How did you figure out who were the right people to bring on the mission? I think the same thing is happening in AI, which is kind of like the meme of what should be built, and then maybe there are some more missionary people. What were some things that you think people should take on in early-stage recruiting, especially?

Early-stage recruiting is so hard. First of all, just don't give up.

Right?

That's my first piece of advice. Second, just think long term. There are folks that I talked with in the first year or two of Figma who didn't join until year 5 or year 6.

Those relationships are amazing. If you're consistent and just spend time with people you like, eventually it turns into something. They might join the company, or they might just be a friend outside the company—someone who inspires you—and that's great, too.

Taking the long view is super important. Of course, you need conversion today. You've got to hire. The reality of being early-stage and having a lot of risk is that you have a natural filtering function: only the true believers are going to get on board.

I'm a fan of just not selling too hard. Make sure people understand what's going to go on and what's going to happen, where you're pushing, and what you're going to do. But if someone needs to be sold that hard, it's usually a sign they're not going to stick around.

Yeah, I think you have to have a really good process. The best recruiting advice I got in the early days was when I told John Doerr one day that I was having a lot of problems with recruiting and wasn't very good at it.

He said, “Well, do you wake up in the morning, and is the first thing you think about recruiting?” I said, “Well, no. I'm thinking about coffee.” He said, “Okay, well, it's midmorning. Are you thinking about recruiting?” I said, “No, I'm probably thinking about what snack I'm going to have.”

He said, “Okay, it's lunch. Are you thinking about recruiting?” I said, “No, I'm probably thinking about what emails I have to do.” Then he said, “It's the last part of the day. You're about to go to bed. Are you thinking about recruiting?” I said, “No, definitely not. I'm tired.”

He said, “If you're thinking about recruiting in all these moments where you just have a moment to pause, and you're actioning on it, then it'll fix itself.” The way that can manifest as a process is that you basically make a spreadsheet of your funnel, and you obsessively look at it all the time and go, “Okay, how do I make sure that this funnel keeps going?”

Just like you would with sales, if you're a salesperson, you have to continue to feed the funnel and move people through it. If you're not doing that, you're not recruiting. You have to be very disciplined, which is something I have to push myself to do. I like to kind of be in the cloud.

I definitely like to have coffee.

How has that changed now? Now you have a public company that you run, obviously with Make. I'm sure you have to build a new team to lead that. How has that changed? Also, this is a great call for recruiting engineers who are listening. What are the types of people who succeed at Figma today?

One thing that's been interesting is that, yes, of course, we've hired amazing researchers who are pushing the boundaries and thinking in a way that's on its own life cycle, in terms of not being as tied to, “Okay, we have an explicit date we're trying to release on,” because that's just not how research works.

But what I found is that very good engineers who are more full-stack and oriented toward learning new things can be quite successful on AI products. There's some reorientation they might need to do. They might need to, of course, learn new skills, just like designers need to learn new skills.

In general, we're looking for smart, high-agency people who have product sense, who care about design, who see the world the way that we do, who want to learn skills, keep growing, and work on hard problems. I think that's the filter that's always been the case for Figma. If that resonates with people, then, yeah, please apply.

One question I had from Zach from Warp was: How do you position Figma Make in the universe of prompt-to-app and prompt-to-creation? How should people think about how you fit? Do they feel like competitors to you? Does Make just feel like an extension of the design team? What is the competitor universe for you?

You know, it depends on how you define it, right? I think engineers feel much more comfortable in IDEs than non-engineers. As we get to more agentic environments, perhaps that's a different vibe. But still, I think most people feel like an IDE or Claude Code, if they're not engineers, is not made for them.

With the platform we have and the visual metaphor, there's an opportunity for free-form exploration and ideation on an infinite canvas, being able to try out lots of things, and then also see the big picture of what the different paths are that I can go down. That is a metaphor that I think works for a lot more people.

In that sense, as we tie Make even further to more than just, “Okay, I can copy a state and put it in Figma Design and tweak it”—that's what we launched today—but there's so much further we can go. The further you go, the more you're able to bring more people into this sort of surface.

In that case, it's just like we're trying to race against ourselves. I think if you evaluate Make as it launched—we raced toward launch—is it prompt-to-code, with not a lot around that and not a lot of integration with the Figma platform? Then, sure, there's a million other tools and more coming every day that you can evaluate against. But I don't know if that's the right way for us to think about it.

I was listening to the Bret episode you did, and he had this one line that just stuck with me. He was talking about, “Do you want to resell coffee beans as roasted coffee beans, or do you want to go make the amazing, special latte?” Obviously, you want to make a special latte. Everyone does. But how can you make something that is unique and meets the needs of designers, and does that in a way where we can really bring an advantage to people?

I think there's so much we can do there. So many different tiers are coming that I'm just really excited about.

Yeah, nice. I know we don't have too much time left. I want to talk about some things outside of tech. You obviously have a CryptoPunk as your Twitter PFP. How do you—

It's actually a Chain Runner, not a CryptoPunk.

See, but it's okay. We'll go through NFTs later. How do you think about the digital world, especially when you're going to have this huge divide between scarcity and abundance? How do you feel about the future of these digital collectibles and communities, and how that fits into the universe of, “Hey, you can generate anything at any time”? Enzo Ferrari used to say, “A Ferrari can never be readily available to be desired.” I'm curious how you think about the distribution of things people will see on the internet between the super-niche, tailored, just-created-for-you things and these iconic cultural properties.

The paradox of digital scarcity is what made me excited about NFTs before they were called NFTs, back in 2017, whatever it was. I think not everyone has that collector gene, but some do. For those who do, whether it's digital items or whatever it is they're collecting, the items are scarce, and people will enjoy getting into whatever collector aspect they want there.

But in general, I've found myself wanting to distance myself a bit from the NFT space. It's kind of interesting. It actually has some parallels, I think, to AI. I got into it so early, and at that point it was this niche community on the internet of weird people who thought digital stuff could be scarce, and you might want to collect it and pay real money for it.

Yeah.

But it was not expensive, so anyone, for example, in the States, could be part of that. It was gated more on just, “Do you know about it, and do you get—”

The idea of it? Is that idea exciting to you, or does it repel you?

Then, over the next 3 to 4 years, it went from this very idealistic, “Let's think about what the future of creation and digital items and scarcity could be,” to get-rich-quick scams and just this overall vibe of flipping stuff and trying to make money. I don't know, the whole meta of it changed, and that's when I peaced out.

I realized at some point I was like, “Oh, I get excited about collecting some NFT project because I think that the art is cool. I think that the creators are awesome, and there's some intention behind the work that's unique.” And then, if I talk about it online, the project might do worse, right? It might attract people to buy it, and folks can apparently—I learned—do scams where they basically pump and dump and create trading behaviors that are no good.

So I stopped talking about things I'm excited about for the NFT space. The parallel that I think is kind of interesting is that, if you compare AI to that, there's been a long era of people who I think are very on-mission and thinking about the big picture, the risks, the opportunities, and the possibilities. That's meeting in this moment with the get-rich-quick energy. If you look on YouTube, there's a lot of people making videos about, “Okay, how do you use AI to make passive income?”

I'm not trying to dismiss that because, great, if you can make money using AI, that's great for you, and some people certainly will. But I think there's too much “Do it because it's going to make you some money” energy in the space right now. That makes me a little bit nervous, having been through that NFT cycle and seeing where it ended up.

You know, I own a card store in San Carlos, so I do like Magic: The Gathering and Pokémon. There's a similar thing happening where there's a lot of speculation just because everything—

Is Magic: The Gathering super cool now?

It is.

Oh yeah, I've been waiting for this moment.

Let's go. We'll do a Magic: The Gathering event.

Draft night. I'm down.

Nice.

We used to, in the early days of Figma, do draft nights.

Nice. What sets were coming out then, do you remember?

Yeah, there's a guy named Andrew on our team, and he wowed me so much with his expansive, encyclopedic knowledge of Magic: The Gathering that I was like, “Wait a second, can we move you from support to product education?” Then he killed it at product education because he has just as encyclopedic knowledge of—yeah, it was basically Magic: The Gathering draft night that gave me the confidence and insight of, “Oh wow, these skills are transferable.”

That's funny. Imagine the Magic: The Gathering career opportunity.

Exactly. You know, I should go around. I'm going to play the Regional Championship for the Americas in November. I should just go around and say, “Okay—”

You're going to the Regional Championship. You're, like, really hardcore.

I think that's the best way to disconnect because you have to be so focused on the game that you're not actually thinking about things.

But there's obviously the collectible side, but there's still, at the core, a community: let's come together at the store, hang out, and play games. I hope that's what we'll see more out of AI, which is enabling more of these small communities locally to have more entertainment and support themselves in a way that doesn't have to be, “Oh, is this going to make money? Is this going to be profitable?”

I think the more you can go from a mode of, “I go on the social media app of choice and mindlessly flip through my algorithmic feed,” to “I'm going and making things,” that's good. We want to move consumption behavior to creation behavior, and I think that will happen. I'm just a little nervous about the get-rich-quick vibes.

Right. Yeah.

Awesome, Dylan. We'll have you for draft night at the new Colonel Space [?].

Looking forward to it. But thanks so much for the time.

Thank you.

Thanks for having me.