Jenny Wen: She Went From FigJam to Anthropic. This Is the New Era of UX.

Jenny Wen led design on FigJam, which is one of the most playful, high-craft tools to hit the design world in arguably the last decade. Changed how teams whiteboard, how people think about collaboration. And then she went to Anthropic. Now she's designing Claude, the product, not just the model. And I was curious how you go from sticky notes and stamps to one of the most complex Highest responsibility domains in tech. Turns out she doesn't view them as different at all. FigJam was this complex multiplayer canvas disguised as something simple, and Claude is a large language model disguised as a chat box. Both require the same skill, taking something impossibly complicated under the hood and making it extremely obvious to the person using it. We get into a lot. Why designers at Anthropic are shipping production code, why she thinks UX designer as a role is outdated, and this philosophy that she keeps coming back to of automate toil, augment creativity. But when I asked her about burnout, she said something that I wasn't expecting. She's worked some of her longest hours on the projects that energized her the most. And her worst moments weren't when she was exhausted. They were when she wasn't shipping. This is State of Play. Let's get into it. You came from FigJam, right? And really led that effort and brought this super playful kind of high craft experience into a lot of people. I mean, it changed, I think, how a lot of teams work and now you're over in, in this space with Claude, and you're really working in, in kind of this complex, I'd say one of the most complex and high responsibility domains in tech. And, and I just want to kick this whole thing off because I have to ask you, how do you kind of balance between both of those forces?

Hmm. Interesting. Uh, which 2 forces specifically?

Well, coming from kind of, it feels like FigJam would be this really exploratory, playful place. How do teams work? How do we bring this? Maybe we can be a little bit thoughtful about how teams are going to engage and interact in this canvas, in this collaborative space. And now high trust is extremely important at Anthropic. Is there, are there differences? Is there like a perspective you bring from one to the other that's maybe a little different?

Yeah, it's interesting because you think about them as like the way you've stated it is like they're 2 really different things. I actually think that the 2 products They are actually very similar in a lot of ways. And I actually think take a lot of similar, like, acumen to build for. Because even though FigJam feels like this sort of like simple, fun, like you jump right into it kind of experience, there's like so much complex mechanics under the hood. Um, you know, you're dealing with this like multiplayer canvas and like the really the way that we designed it was how do we just give people the right options that simplify the experience so that way they can just like jump into it, even though under the hood it's basically the same. multiplayer canvas and mechanics as like the Figma design editor. Um, so that approach is just like, make this complex thing simple and sort of fun for people to use. Um, and I think actually this exact same philosophy exists for a product like Claude, right? Because you're dealing with this like really complex, again, also non-deterministic, um, thing, which is like the LLM, like the Claude models, right? But we are trying to present them in a way that is Simple and often playful and easy to use for people who a lot of the time might not actually understand what's going on under the hood with the LLM. So I actually think that design skill-wise, it hasn't been— I mean, I've learned a lot about LLMs and whatnot, but it sort of requires the same level of systems thinking to understand the underlying mechanics, and as well as the same level of craft to be able to execute it in a simple way to for, you know, a broad user base to be able to use, if that makes sense.

Yeah. Yeah. Well, and what's, what's pretty interesting too is, you know, coming from a lot of our traditional product backgrounds is, and I've got to imagine it might've been this way with FigJam, is how do we introduce the right amount of complexity and reduce some of the things that aren't necessary right away? We look at something like Claude, it seems like surfacing a lot of things under the hood. While it might feel kind of counterintuitive, I want to introduce AI into a product. And I want to balance that with human agency. And we're, we're thinking about this with one of our kind of more consumer-driven apps right now. What amount of complexity or how do you make the decision of what should be surfaced by default in a product that's, that's so AI powered?

Yeah. I mean, I think we're still trying to figure that out, honestly. Right. Like, um, give, like, for example, some, like, you know, an LLM can have these Basically capabilities like memory, like content, like a context window. And I think we have a lot of these discussions of like, oh, like how much of this should we reveal to people? And I think we're finding also that like, as people are becoming more and more comfortable with AI, they actually themselves are internalizing a lot of these concepts. Like they realize, you know, hey, I need to sometimes switch over from this chat to a new one because Claude's context window has been exceeded. And so they sort of understand like, okay, the LLM can only handle so much information at a time. And I think this is like an internal debate that we're having. And I think our stance is evolving too on just like how much do we reveal to people? And I think we are starting to push the other way where we're like, okay, we don't need to abstract everything away to people. And actually a large set of our users who are, you know, these like AI enthusiasts, people who are super early adopters, they actually, I think, want to have some of this information out there and for them to understand, because that actually makes their usage of the tool better and it lets them do more powerful things. Our sort of like philosophies and principles around how to design for AI is just like, I think the needle is moving and it will consistently move as we sort of see our audience change and become more and more, um, I think understanding of how LLMs work.

And how do you start to think about, um, you know, you talk about some of the principles over at Anthropic on how we introduce AI to the end user. How do you start to think about making trade-offs between we should automate this, we should get more agentic, or we should add intentional friction so that the human can remain in the loop on that? Yeah.

So our, the philosophy that we have internally is like we should augment, you know, the kind, we have these like 2 kind of modes of using AI, which is like augmentation, which is like working with the AI to do something, and then automation, which is like, you know, it does the task for you. And the way we think about it is like, we should automate toil, like the things that people hate doing with their jobs. Um, like, you know, filling up status updates and like, uh, you know, checking that these 2 documents are exactly the same that you're looking at and stuff like that. The stuff that like isn't creative and it's just like, you know, we call it toil. Um, and then we should have Claude augment, you know, the things that you love to do and, and help you be more creative and help you do more with your work and whatnot. Um, so that's the general philosophy. And, and I think we also believe that as the models get more agentic, it's more about like collaborating with the, the model as opposed to like giving it everything to do all at once. Um, especially as like, you know, right now the model is going to be as the worst it's going to be at sort of like doing some of these tasks and like throughout you using Claude to be, to do agentic things. Building that trust over time. So I think you'll see in earlier days, we'll ask you to verify a lot of stuff and sort of check and sort of have this transparency around what Claude can do, sort of like what extended thinking does today. But I think over time, like as you, the user, gain more trust with the model around like checking its outputs and making sure it's doing the right thing, because I think that stuff actually might get old to you after a while. Uh, you might just be like, cool, Claude can do this. I trust it. I don't really need to see what it's thinking all the time.

I recently had a great conversation with Ria Liu from Cursor, and we spoke about this metaphor of kind of the fog of war, right? Old games we would play and you explore the map and the shroud uncovers. And, you know, my question is about pathfinding in a product at scale like Claude with so much ambiguity. I'm sure ambiguity is the same from one place to the other, but I've got to imagine this territory is so new. What are some of the methods that you are finding your team using to best pathfind in this crazy ambiguous space?

I don't know if the methods are that different than they used to be, you know? I think the tools that we have are better for sure. You know, like a lot of it really is prototyping and prototyping a real thing. Like I think, When you are working with an LLM, it often doesn't help actually to make an interactive prototype that's just like clickable. Um, even like I, I used to make prototypes in Framer way back when, and I was like, oh, this is like, you know, a technical prototype, uh, because it can show all these states and it has like all these animations and interactions. Um, but I think a lot of times now you have to like make a prototype that is using the model, right? And, and so that way you can sort of like interact with it. Um, and see what its responses are and whatnot. And that often includes like prompting. And I think that often actually requires like working directly with an engineer. I think even some of the most technical designers, um, it, you can only get so far with the prototype that you can make because you're not just prototyping the UI and the interactions, but you're actually prototyping the capabilities built on top of the model.

I'm noticing that prototypes have started to replace a lot of things for us. I mean, we'll, we'll still write some PRDs, but they're a lot scrappier. And prototypes have started to become kind of the centerpiece of, of our, our handoff and our meetings. And I'm curious, uh, to hear from you, how much does prototyping play a role in, in like the early discovery of, of new ideas?

I think most of our new ideas come from prototypes. There have been some things that I have like mocked up in a flat way and, um, and done sort of like a clickable prototype for just to like prove a concept, but Um, actually a lot of ideas come from engineers who are making prototypes as well. They're usually, again, like they're prototypes of capabilities as opposed to prototypes of UI or interactions. Um, and those are the things that sort of like change the game for us.

Coming from FigJam and the collaborative piece and people kind of working together in that early discovery to organize and synthesize, um, in a canvas, in your mind, do you think prototyping becomes a sort of collaborative early stage artifact as well?

Yeah, I mean, I think so. When we were working on FigJam, a lot of the prototypes there were, they were a little bit more design driven, you know, because a lot of them were sort of interactions as opposed to like technical capabilities. Um, but a lot of that was just like, or, or a lot of the, the prototyping at, um, Anthropic tends to be Okay, like one person made a prototype here and then somebody else like riffs on it and makes another version of the prototype there. And it sort of like builds on each other over time. And those ideas like come back to each other, come back to us like time over time. And eventually it becomes something that we actually like pursue. Um, it's less so like, okay, we decide, you know, we are going to prototype this specific thing. It's a little bit more organic actually. It's like the prototypes gain heat over time in the organization and people sort of get nerd sniped into the prototypes. We do have a corner of the org that does do very explicit prototyping and their job is to just basically go through different hypotheses and prototype them out. But actually I think some of the more successful prototypes I've seen were just people across the org prototyping things in their own spare time and those things gaining momentum.

Yeah. Yeah. I love the permissionless approach of, of doing that. Is that a pretty, is that a pretty standard cultural thing at Anthropic to just kind of put it together and see?

Yeah, I think so. I think so. That feels like it's the most effective method for ideas to come together. And I also think that's just the way it happens. I feel like sometimes when you're like sort of told to, to think hard and to brainstorm about something, you actually don't have the best ideas. Like the best ideas just come. And they strike at the most random moments and then you just sort of gain energy around them. So yeah, I think it's both effective in getting the idea across, across the organization, but it's also effective, I think, in sometimes getting people to get the best ideas. Yeah.

And you've said similarly about process and I love that too with, you know, design process is just kind of this muddy, messy place. And sometimes we'll follow one linear approach and then sometimes we'll do a different thing. And I'm curious, you with such a robust background, Are there things that you are seeing in day-to-day workflows that are quite a bit different from, let's say, your time at Figma or Dropbox?

I mean, I do think the big thing is designers are implementing code, you know, through, especially through using Claude code. That has been the biggest difference, I think, from working at Anthropic versus like back when I worked at Figma, even though I'm guessing that at Figma now, they probably are doing stuff like this too, given the availability of the tooling. But it's crazy to think about like how quickly the sort of like development tooling for non-developers has progressed in the last, you know, 6 months or so. Even today, we were like reporting some bugs and some quality issues and like one of the designers was like, cool, let me just fix them. And that was cool to just like have. Yeah, and not have to like tag an engineer for, for them to do anything. And so our process is actually kind of similar to what it was like at Figma in that it was, it didn't really follow a process, but I think that's probably because what we were doing at Figma at that time was like kind of radical. Um, like we were sort of developing products in this sort of like Wild West kind of like fuck the process way before like other people were. And I think that's actually what created a lot of the magic of Figma and made us execute really well. is that we didn't follow that process. And I think so similarly at Anthropic, there isn't a prescribed process. A lot of it is intuition and just being like, oh, that is a very cool idea. There's something there. Let's try it out. Maybe one difference at Anthropic is given that we are working on top of a new technology and we work at this research lab where the technology is being invented, We often are just like taking the technology and trying to find the best fit of the problem for it. Um, and so as a designer, it often doesn't make sense for you to make this like grandiose, like vision deck and being like, cool, 2 years from now we are going to do this, this, this. Um, because even though we can sort of see where the model is going in like a year or 2, we just like don't know. Um, and so often like design needs to follow what the model is capable of. And design from there as opposed to starting from like a design vision first kind of thing. And I think that can feel like tough as a designer because you're like, oh, I want to be design led. You know, we should be designing it first and then the technology should follow. But I think that's just the reality of like working at a research lab where it's like the technology is emergent and you have to sort of like decide what to do with it.

That's a really interesting call out because I, you know, I think a lot of designers like to start with the ideal state and they're thinking too, I mean, that's how we're thinking. What does this, what does V4 look like? All right, let's go backwards from that. And you're saying that's typically how you'd operate, but in this case, you're so much more dependent upon where the models go that it's almost kind of redundant to do that.

Yeah. I mean, I think there's still a little bit of attention. There is still sometimes like, hey, we want the model in the next generation or so to be better at X use case. So you can definitely inform the training, but I think it's still a large portion of it is just like, cool, the model is good at this. These are the new capabilities. What can, how can we design for that?

So I want to ask you a question and I don't talk a lot yet, um, about the job market and the, the real like role of the designer in that context. And I feel like you, of a lot of people we've spoken to, would have some incredible insight or some ideas about this. So I have 4 teenage boys, different skill sets and things that they're really excited about. And some of them love to build. And they're not thinking about, I want to be, you know, a developer or a designer. They're not necessarily saying that out loud. They're just having a great time building. But I have a lot of people who, you know, that I talk to day to day who are like, I don't know, am I a designer? I don't know what skills to lean into, what makes me hireable. When you look at the market today and when you're evaluating talent to come onto the team, what are some of the things that you're really looking at?

Specifically for hiring at Anthropic, we've been looking for people who do have this experience and excitement about AI and where it's going. I think we just talk a lot about people who can help see the future with these models and understand where they're going and are Hacky and scrappy and like, like to try things with them. Um, I don't think that's necessarily like a, a, always a prerequisite, but I think that really helps. I think if I were personally like a, um, a hiring model, a hiring manager, like anywhere else as well, um, I would also look to, to see if folks are just like really flexible and sort of like able to like learn and like kind of question their process. Um, 'cause I do think so much of that is changing right now. It does make a big difference. Like someone who's like willing to like try out new technology try out new parts of the process, think about the flexibility of their role and not just be like, cool, I am a designer. I work on my mockups in Figma. It's like, our roles are very much expanding, contracting in some ways, expanding in others, and having this flexibility, but also growth mindset and ability to pick up new things and try them. I think it's going to be a skill that you want on a team, but also I think it's a skill that everybody needs to have right now, um, because so much is changing.

Big advocate of this idea of creating your own kind of personal playgrounds so that you can do that in kind of a space that you don't have someone to answer to. You can just kind of build some things out. And I love, I love that concept. Plus it allows me to kind of refresh a lot. Do you, do you think that we're going to see more and more generalists because the tooling is so broad now? Or do you think we're still going to kind of go down specialist route?

I hate making these predictions because I'm like, any, I feel like with AI especially, I'm like, anything could happen. In the next 6 months, anything can happen. I do think it's going to be more of a generalist world because I think even now, I feel like a lot of the designers at Anthropic, for example, are pretty technical, right? And they can write production code. And I actually think that's pretty advantageous for the folks that are really technical. I think in the new world, when we have a lot of these tools at our disposal, It's really just gonna be like, what is your ability to use a whole breadth of them and sort of like understand how they work and fit together? And I think it's like also, you know, if as a designer's sort of portfolio of things that they do also expands into like PM work, for example, like being able to do that and like drive initiatives and sort of like get people to execute on the team, like I think is also gonna be a really useful skill. 'Cause I'm sort of seeing that design role stretch into those 2 corners as well.

Are you, are you playing with any technology at the moment? Do you give yourself room to kind of have a playground? Do you have time for things like that lately?

Am I? I try to outside of work. Yeah. Like I try to like try out all the new tools, mostly like for sort of like competitive intelligence. Um, but also there are a lot of new tools. Like I feel like every day there's like a new like design or development or like agent tool. Um, that's coming up and I, I don't know if I have like time necessarily to try them all in my design process, but I'm like trying to keep an eye out.

Are, are there things in the technology you've seen either at Anthropic or, or just in the industry right now that are really exciting you? Um, that maybe, you know, there's, there's hype everywhere. We're seeing it all over the place. I can't get in every single day. There's something new, but is, are there like some things that have at least piqued your curiosity? Maybe you haven't played with them yet, but they've been, They've looked interesting.

Basically any sort of design generation tool that like spits out a can— on a canvas, I think is really interesting 'cause that feels like a big unlock. Um, and it's pretty different from, you know, like the Lovable or Bolt models where it sort of spits it out into like this linear web page thing.

You're talking like the Floras and the Weavy type models or?

Uh, yes. And like there's like Magic Path too.

Um, Magic Path Paper. Yeah.

Yeah. Those are really interesting. I do think it's like, The ceiling or the, not even the ceiling, the floor for creating like a really good functional, like canvas tool is like very high. Um, so I think it's going to be a while before one of these is like really good enough for, for like a designer to use full time, but I'm excited about them because that feels like a, it's like a missing gap in our like designer's toolkit.

There's another side of this and, and I love talking to you about this because I mean, it just already seems like there's like a joy in a lot of this stuff that you do. But there's gotta be this level of intense competition right now, this level of just high stakes, high performance. And I'm curious if that's true, and maybe it's not, but if that's true, how do you keep kind of the joy and the curiosity alive when you've got all those pressures around you?

I mean, I find joy in winning, you know what I mean? Like, and I find joy in competition and things that move quickly and Um, I find it very hard to, uh, perform without pressure. So that, and maybe that's just the way I'm wired, is like, I sort of like it when it feels like there's a space and everyone's moving quickly and you sort of have to like look over your shoulder. And it's, and it's also very cool to see just like a lot of great work come out from like different teams and think about like, oh yeah, they did it really well. Or like, oh, I think we can do it better. And how might we do that? Like, I sort of like the challenge of being like, Cool. Like this other team did it really well, but like, I, I have the confidence and I believe in us as a team that we can do it even better.

Speed is this like thing that I've had such a complex relationship with throughout my entire career. Um, I have learned to make very fast decisions and it's like, it's almost like I've had to build this strange mental model around that's okay. That's the, you know, that's only, that's the, the 80%, like it's going to impact the 80% of the product and that's cool. These days though, the mandate for my team is, you know, we're really trying to move quickly across, like every week there's some new initiative, it feels like. And the question here is when you motivate a team, you're, you're intrinsically motivated by winning. That may be the case with some people on the team and there, there may be other things that people are looking to get fulfillment from. How do you keep the drumbeat around speed? How much is speed part of the culture over there?

Yeah. Um, I mean, speed is a part of the culture. For sure. Uh, 'cause we are sort of, we are faced with so much competition, but also like the, the speed of the, like the models advancements are just also really fast. Like if you think about, you know, how many model releases there already have been this year, I, I don't know how many there have been, but I feel like at the beginning of the year we were, we were like dealing with Claude 3.1 or something, and now we're at like 4.1. And so it's like we've already launched a bunch of stuff and like the technology's always moving. How do you motivate people? I mean, at Anthropic, it's very mission-driven, right? It's all about delivering safe AI and increasing the number of safe tokens there are in the world. And I actually think it's not hard to motivate people around that because that's usually the reason they joined. Because there are other labs you can join, right? And most people are motivated by that. And so I think a lot of the time it's bringing back the reasons why we're doing something to That, you know, uh, like the reason we want to win as a product wing, right, is so that way we can make revenue to support building safe AI, um, to build these safe models, to do the sort of research in order to improve AI safety. And so it's like folks are not motivated by, you know, winning from a product perspective. It's more so motivating folks around the mission and, and being the company that sort of shepherds safe AI into the world.

When I think about, you know, we're incorporating some, uh, AI patterns into one of our products and You start to think about how do you layer on AI? Well, we'll, you know, maybe we have a, we have a couple of patterns and mental models for thinking about this. Now we can look at a lot of people. Microsoft put out something great a few years ago on how they were treating interaction patterns around AI. And then everybody else has continued to build these libraries. It used to be to me that in at least desktop interfaces, we were probably reaching some sort of ceiling. of patterns. Um, and at that point we're just kind of, you know, reusing things and adding little tweaks here. Do you think that there is a, we've already reached a ceiling or we've opened new territory when it comes to interacting with like large language models on desktop or mobile devices? Or do you think like that's just completely wide open again?

I think it's still wide open. Like I think we have, we are at a place with chat where we're like, okay, it actually is like pretty effective for a lot of things and it's like not going away. I do think in whatever future there is, there's some part of our interface with AI that is still chat. It is just the most natural, the most flexible, the most powerful way for us to communicate with it. But I think it's like the model's capabilities are continuing to get better and it will be able to be capable of more things. And we still have to figure out the right patterns for that. And we have to figure out how that's interacts with chat, like what part, what, which parts of interacting with the model should happen in sort of like more guided UIs and which parts should remain in chat. Um, and I think it's like really interesting to watch the industry sort of figure these things out together, um, and, and see different products evolve. And so I don't think we have reached some like local maximum or anything yet.

One of the trends we talk about, you know, what's happening all over Twitter, Is the hype cycles are continuous and loud. And what's interesting to me is I'd say Anthropic's actually one of the folks who doesn't really subscribe to overhyping around the models and the things going on. Not to say you guys don't do marketing. And I'm curious from you because what I'm hearing from a lot of people, and I feel like I'm this third party who fends off the cynicism. I'm like, yes, the promise of AI is a little janky sometimes, and you really have to go into it with a certain frame of mind. But what's not helping someone like me is the overhype of things and the underdelivery of what's coming. And people feel like they've been burned over and over. I'm curious to hear from you. How do you think about hype and how do you think that kind of influences the mission at Anthropic?

Yeah, I will say, I have noticed that we, like, I'm not marketing, but I also have noticed that we don't like, overhype our abilities. And we're also very cautious to not claim that we're good at something before we're good at it. For example, we often just say things are in beta or research preview for often longer than I think we should. But I appreciate that. I sort of appreciate that we are not overclaiming what we are doing. And I think what that's doing is creating trust with our users, especially because a lot of our users are sort of power users and early adopters, and they They can suss out like, is this fake or not? So I appreciate that. I appreciate that sort of like working in openness and like trying these tools out in open. But I think, you know, hype cycles will die down. And at the end of the day, it's like, where do we see users going? And that sort of is going to be the thing that tells the truth about what's working and what's not.

Yeah. Yeah. I, you know, I have to tell you, speaking of hype cycles, I am, I am just like this. regular consumer of new tools and technology. I just love trying it. Let me give it a good day. And I keep finding myself when it comes to the AI coding, AI-powered coding tools, Claude Code is just, in my opinion, hasn't— there's nothing that's beat it yet. Keep finding myself coming back to it. But I am curious because one of the challenges around the actual development of software has been, well, we can kind of lean into Tailwinds, we can kind of lean into ShadCN, and we're getting a lot of the Things that look similar there. Screenshots, I think, seem to help a lot in these aspects of bringing that. But the front end, the UI is still, I think, a lot to be desired in some of these. And what do you think, how do you think that's going to start to evolve? Do you think it is just a paradigm of like canvas tools helping us tweak that? Or do you think there's something more there on how we bring like good interactive experiences to life through AI?

Yeah, I mean, I think a lot of it is the model needs to be trained to be better at it. I think they will get better at it, probably within the next few months to a year. I think we'll see the quality of, and not like, I think all the different models that sort of do this work, I think we'll see the quality of the UI being produced be much, much better, 'cause it's already gotten a lot better in the last year or so.

Yeah, and then there's the question of kind of collaboration. I see people spin up just agents and agents that are just coding and collaborating with each other. Meanwhile, and maybe this is a skill issue, it probably is. I have a hard time if I'm going to go deep into, you know, we'll use the term vibe coding with like Claude Code for a second. And I'm working in a code base with another engineer somewhere. How do you think about, especially coming from with FigJam, how do you think about 2 people communicating with an AI to code, but collaborating on the same project together? Do you think there's like this big—

Yeah.

difference between how we use version control in a manual environment versus how we might collaborate with, uh, you know, augmented by AI, or is it pretty much the same?

I think we're starting to see that it can be kind of the same, you know, like especially as the models get more capable of like doing long-running tasks. Um, I think we're sort of seeing the, like, Claude Code, for example, show up in all the surfaces that you're already working in and like, be able to do stuff from GitHub like it's, you know, your coworker that you're assigning it stuff to do. I wonder though, like when there is a world when you are just running a fleet of agents and not necessarily like working with people, like maybe there's just one project and it's just agents working on that and it's not people, like there probably might, there might be like a different interface because like, you know, right now, like putting something in a GitHub PR, that's like really readable to like you and me, but maybe it's like actually not the most readable thing for an agent.

Yeah.

So I don't know. I'm guessing that some of these, um, surfaces might actually evolve to better lend themselves to agents.

Where you're at today and you think back to, uh, I had said it earlier, you know, something I've never said to myself is I want to become a designer. Did you ever say that I'm going to be a designer when you first started?

Um, I think so. I did. Like I actually, uh, in high school, actually, I wanted to become like a print designer, like somebody who designs like the layouts of magazines. Um, and I actually remember reading Seventeen magazine when I was a teenager, and they had a, uh, a sort of article about what it was like to be a layout designer of Seventeen magazine. And I thought that was like so fricking cool. And that's when I decided I wanted to be a layout designer of a magazine. And then I eventually like applied to go to graphic design school, but actually didn't get into the schools that I wanted to go to. And then Did not go to graphic design school and then eventually became a designer.

Yeah. Well, and your background's so interesting and it's clear through talking with you, you have a very competitive spirit. You're a very forward-thinking person, optimistic about the future. Where do you think a lot of this started, like in your adolescence? Where, what was kind of, where was there a moment that was maybe the first time you were really competitive about something or brought something to life and said, I want to do this at scale?

Yeah, that's so deep. I'm like, I don't know where this comes from. Where have I been competitive? I think I've always just really liked being good at things. I was honestly like a lazy person. I was like a lazy teenager and a lazy kid. Um, I think it was mostly just because like I was never challenged, you know? Like I was just, I think I like, I like, I think I knew I was like pretty smart, but then I think school was like pretty easy for me. And so I was just lazy, you know? I would just, I would just show up, do what's expected of me, and like not really do anything else. I actually think I only really learned how to apply myself when I went to university, um, and I went to engineering school specifically because I was like, oh, everyone else here is like really fucking smart and they might be, they, most of them are smarter than me. Um, and basically the program that I went to at the University of Waterloo, you have to do internships basically for half the time that you're in school and you basically graduate with like 2 years of work experience. But that also puts you in this pretty intense cycle because you're doing school and you're doing it for 4 months at a time. And school is pretty competitive too. People fail out all the time. I think only 60% of our class made it to graduation. And then you're also trying to get a job during that semester that you're studying super hard in. And that was the first time in my life where I was like, oh my God, I might fail, I might not make it, and I actually have to try really hard. And I think through that period of time, doing all those hard things at one time, I was like, oh cool, like I'm actually capable of a lot. Like I can do this if I apply myself and work really hard. I can graduate, I can get a job. Um, and I think that just gave me the sort of confidence in myself that like, oh, I can, whatever it is. I can learn it and I can get through it. So I think that really stoked, like, not just a competitiveness, but also just like, cool, like, I know how to apply myself and I believe in myself kind of thing.

You know, that resonates with me so much because, you know, I've been in a lot of situations where I've just kind of had to have some grit and push through. And it doesn't mean it's been, it's not been without its pain, but it almost seems like anytime I suggest that that's something that's worked for me. I do get sometimes pushback from people and say, oh, that's, you know, you're, you're talking about burnout and you're not approaching it with a certain sense of care. And in my mind, I think, you know, and even right now, the kinds of, if you're trying to achieve something larger than yourself, there are going to be a lot of times where you have to make that choice. What do you say to designers and people who, you know, are feeling like they're moving mountains? And maybe not getting the rest that they think they need, but they're close. What do you say to people who are experiencing kind of the burnout?

Yeah, I mean, I think I would ask people and question like, why are they experiencing burnout? Because I do think that like one thing we often equate burnout to is just like time spent working. And I do think that's like actually one way to get burnout. It's like, oh, if you're just like working all the time and you are not getting rest, That is a way you can get burnout. And I have gotten burnout that way, and it's like, cool, you should take some rest. But I think another big way that folks get burnout is when the work that they are doing does not fulfill the goals that they have. And I find this a lot where it's like, oh, if I feel like I'm working on something and it's not going anywhere, or it's not challenging or exciting to me, I can put in medium amount of effort, but I probably feel worse that I do than when I'm like working on something really hard that I really care about. Like some of like my most energizing moments at work, I've like worked really long hours, to be honest, but then I, you know, put something out there and like ship something great. Like, for example, when we were shipping FigJam, I was like working all the time, but then I was like super excited about what we were doing. But then, so my worst moments at work have just been like, hey, everything's pretty slow. We're not shipping. I'm not working that hard, to be honest, but I don't feel good. And I'm not necessarily burnt out, but I'm just not feeling great. So I would question, hey, why are you experiencing, what is it that you're experiencing burnout around? Is it time spent? With that, I think that is kind of hard to solve if your job just demands a lot from you and there's a lot of time. But maybe it's actually more a question of, you don't feel like you're spending time on the right things and you disagree with the way that the organization works or the things that the organization is asking from you, then that feels like potentially, okay, cool. What do you care about? What energizes you and how can you find something or align to something that gives you energy, even if it is a lot of time spent?

So you're at that tail end of Waterloo. You're finding a job or you are at that tail end of like FigJam just launched, right? And you mentioned rest. How do you rest?

I love to just be a potato for some period of time. 'Cause I do think it's like really nice to just like let the brain rest, to just lie down. Like that's a big thing in recovery. And that usually, I think takes, it depends what it is, but it, that can, that itself could be a full week or 2, you know? And then after that, it's often doing a lot of the things that I've put off in the past. It's like, oh yeah, I have been meaning to update my portfolio website or something, and then getting to that and feeling like, cool, I did something and it was useful and I'm glad I finally got to it and I cleared my plate. I finally cleaned this one part of my house. I did it. Um, and then I feel like the next stage then is like, okay, what are the new things that I can start to pick up? Um, so it feels like, yeah, these are like stages of recovery. It's like sort of like declare bankruptcy on your brain of all the things that have been loading it. And then the other thing is like, you know, do the things you've been putting off. Um, and then finally do net new things. I think those are my stages of recovery.

I, I love that. That's actually such a fantastic way to frame that. That's, uh, very cool. Um, My startup background, created a company back in like 2014 and 2015, and it just happened to catch wildfire pretty quickly. And I look back at that time when we sold that company, and I've always wondered, I've had like, this thing has nagged at me for a while. I'm like, oh my gosh, will I ever have an opportunity to work at something that is so widely received or well received? And at scale, or was, or was that it? You know, do you, have you ever experienced kind of a moment of pause about that when you look back at some of the things you've been able to achieve and, and then you find yourself now at Anthropic, like FigJam, where you're like, oh my gosh, was that, was that the height?

Yeah. Yeah. Yeah. I mean, I do sort of feel that way about, uh, some of like the earlier FigJam days actually. Um, and it's, I don't know if like I have also found that yet here, uh, cause I think that was specifically like Oh, I like basically started and like founded this product with like a small group of people. And there was honestly just like such great energy on that team for like the few years that I was working on it where like things just sort of like felt magical and they fell into place. Everybody worked really well together. Everyone was like really fast too. And so like I would be like, hey, like there's like this one bug and then it'd be like immediately fixed. And it was so magical and it felt like everyone was at the top of their game too, where they were just making things that were just consistently delightful and we were riffing on each other's work. And I think those early days of, especially a product too, because there's also no tech debt and everyone's like, anything's possible. So I do think back to that time as a very special period of time that, yeah, I don't think I've experienced that again yet. And I— I do worry about that. I'm like, oh man, is it ever going to be like that again?

Yeah. Yeah. That's, you know, it's interesting because I have such a complex relationship with creating things. Like a lot of times, you know, designers or people will get attached to kind of the identity they might have. I have such an attachment to the things I create and it's, uh, I guess I'm working through that.

How do you, what is your relationship with, with like I don't know if I have like this super strong attachment to things that I've created. Um, I think what I love is like continuously improving things. I get less attached to like a specific solution or something, but I really like, like putting something out there and then like immediately getting feedback about it, whether it's like positive or negative. Um, just the feeling that like, oh cool, like I know how people are using this now, how they're reacting to it, and I just have a complete clarity on like what to do next. I remember being like kind of like obviously stressed out before we launched it, but then like after we launched it, there was just like clarity. It's like, cool, we know people, people have reacted to it. We know what their feedback is. Now we know what to do next. Um, and so I think just like, yeah, for me it's less about Um, being really tied to the things that I've made or certain solutions, it's like, cool. Like, how do we keep making this thing better? Like that, that's the thing I'm always thinking about.

And are there people in the industry and they, or, or even adjacent who you kind of look to or really inspire you? I always like to find the people who inspire me and ask who is inspiring them.

I don't love to like have idols, you know?

Um, sure.

But I think currently I am really excited about Like a flock of basically younger designers, because I think they're sort of showing us like what is possible when you don't have an existing sort of like career and all these things that you carry around with you. And I feel like there's this like hybrid of, uh, you know, designer, design engineer, kind of like new designer that's coming up because they just like didn't have the definition of like what a product designer is and they're just like making stuff with AI and like, and so. There's a flock of those folks that are coming up on my Twitter feed that I'm really excited about.

The same for me as well. I say the youths, the youths are out here.

I love the youths.

You know, no preconceived notions about how things should be done. And it's incredibly inspiring even watching my son connect. He has kind of like a windsurf to Unity workflow and I'm like, try this or try that. And he's just doing it in ways I wouldn't have even thought of. And it's interesting to compare that to people who I say, hey, you know, consider playing with the tool over here, doing this. And sometimes the kind of salty response is, oh, I'm expected to now know how to do this and to do that and to do this. And it's like, well, I think the difference now is that the tooling kind of increases our capacity in the day. We have more things we can do now because the tooling makes those kind of rote things easier. What do you say to people who, you know, are watching the youths, uh, go through these, uh, number of skills and just kind of putting things together, but feeling like, oh no, like having to learn engineering is a whole nother vertical now for me.

So I think that that's what's giving young people or like new grads, like an advantage, right? Is like, they're just sort of willing to learn these like new things. Like I remember being a junior designer and just being like, cool, like I suck at visual design. I need to learn visual design and just be like, cool, like I need to do this. Otherwise I'm going to like never make it. And I, so I think it's like we, given that there's new technology, we all are sort of almost in this like new grad phase again where we have to remember like, oh yeah, we have to learn new things with our careers. And yes, it is hard and yes, we have to find time for it, but I think that's sort of the reality of, of where we are at right now. Yeah.

And how much do you think, especially with the design field now, there's a, I'm seeing a lot of debate about folks who spent most of their time maybe on the researching user experience, not necessarily the visual output side, starting to wonder how much of that should I start indexing on visual output, my, you know, quote unquote craft, things like that. And especially coming from a research company, I'm curious, how do you see that evolving in the field of design?

I mean, I think it's what we are grasping onto right now because the models are not good at some of those like really fine visual details. And the model is like pretty good at like broad UX decisions. I think it's going to get better and better at some of these visual details, um, and, and making stuff that actually like looks good and has like shows a high level of taste. So I actually like don't know how long that's gonna last. I think there will always be some amount of like, can you prompt and sort of choose and, and make and get the model to like do this in the right ways at the right moments?

How much do you think design as a, as a field is going to index, uh, less on the research side and more on the visual output side, which seems to be a debate happening now?

Yeah. Yeah. What do you mean by the research side? Like the sort of UX side?

Yeah, just as the role. I think job titles is what I'm thinking about when I think about a UX designer versus like a product designer and how much of the day-to-day is gonna be output-driven activities.

The UX, like designer, like non-visual designer role feels, it does sort of feel outdated at this current moment. Um, I think, cuz that, that role almost feels like, I mean, that role has always almost like felt like PM to me in some ways. Um, and so like Maybe that's kind of what it will graduate into. But I think also given if the models do get better at some of these highly high visual fidelity outputs, then maybe this UX designer person can then not be the person creating that anymore. They can leverage the models. But then I think what then it depends on is does this UX designer person actually have the sort of taste and ability to be like, cool, these are the moments when visual quality really matters, and I will sort of push the model on this. Because I think that having taste on something versus being able to execute on it is 2 different things. They're usually pretty closely bundled together. I do think people who can execute on things with a high visual quality usually can also— they have taste, and kind of vice versa too. So that sort of depends. But I do think that the sort of UX designer role is a, it's a little bit, I think we don't know, really know what it is.

I think it's very interesting that you actually kind of juxtaposed like taste versus the ability to execute on something. And I want to hear from you, how would you define the trait of being able to execute on something?

The ability to, I think, actually create a bunch of like different options. You know, that just like look good and feel good and like to be able to do that like probably pretty quickly and to be able to do that in a way that like meets the goals of the project and like does it pretty intentionally. And then I think the person with taste is like, once you've made these like 12 options, right? Like you can actually really understand the level of detail, understand like the decisions that matter in visual quality, and you get to, and you could choose the right one that Is worth doing.

I really like the distinction. It is subtle. It is a bit nuanced, but I think that is a really smart distinction that deserves more attention maybe. Otherwise, if you're just someone with great taste and no ability to execute, you're just kind of a critic.

Yeah. Yeah. Yeah. Yeah. I mean, I think there, yeah, I think there's, I think there's, there is some merit in like having great taste and not necessarily having to be the person that executes all the time, but Um, I think often you'll find that somebody who can execute also usually has great taste.

Jenny said something that reframes how I think about the design role right now. She said having taste versus being able to execute are 2 completely different things. They're usually bundled together, but they don't have to be. And in a world where AI can increasingly execute, the question becomes, and it's kind of uncomfortable, do you actually have good taste? Or are you just pushing pixels around? I think a lot of designers, myself included, are having to sit with that. I think a lot of designers, myself included, are having to sit with that. And she described the culture at Figma while she was designing FigJam as a sort of Wild West, fuck the process, her words. And she brought that same energy to Anthropic. No grandiose vision decks or these 2-year roadmaps, just prototypes that gain heat or ideas that strike at random. And designers and engineers who are both building the same things. If you want to figure out where design is headed, not the hype on Twitter, this is who you should be following. She is quite literally in the room where it's happening. I'll see you next time.