#15 AI Won’t Fix Bad UX: Dan Winer on SaaS Bloat, Design Systems and Product Strategy

In this episode Mark speaks with Dan Weiner, Director of Product Design at Kit, the creator focused email marketing and automation platform formerly known as ConvertKit. Dan brings a rare mix of design leadership, front end development experience, and deep B2B product thinking to a conversation that goes far beyond surface level AI hype. 

This conversation starts with a familiar problem in SaaS: why so many products still feel bloated, fragmented, and harder to use as they mature. but goes deeper into how AI is reshaping design practice. 

Dan argues that teams without strong UX foundations will struggle most because AI can accelerate delivery without improving direction. He also makes the case that design systems now need to be thought of differently. They are no longer just a bridge between design and engineering. Increasingly, they are part of how agents understand, generate, and build interfaces. 

Mark and Dan also explore the changing role of designers inside modern product teams: when Figma still matters, where AI prototyping is already useful, why design engineers are becoming more important, and what design leaders now owe their teams in a fast changing landscape. Dan’s view is grounded but optimistic. Design is not disappearing, but the kind of designer companies will need may be changing fast. 

In this episode, you’ll learn:

00:00 - Coming up on this episode with Dan Weiner

03:31 - The SaaS Bloat Trap: Why Platforms Keep Getting More Confusing
04:19 - Enterprise Dictatorship: How High-Ticket Demands Quietly Ruin Self-Serve UX
07:05 - The Hardest Choice in Product Strategy: Deciding What NOT to Build
09:22 - Beyond the Wrapper: How Kit Actually Uses AI to Drive Creator Value Faster
13:27 - Workflow Over Hype: Defining What Makes AI Genuinely Valuable
14:18 - The Acceleration Trap: Why Weak UX Plus AI Equals Moving Faster in the Wrong Direction 00:15:58 - The Unsexy Reality: Why Deliverability and Trust Still Beat AI Trends
22:19 - From Months to Weeks: Speeding Up UX Research & Prototyping
30:50 - The Paradigm Shift: Why Design Systems Are Now for Agents, Not Just Humans
38:42 - The Leader’s Mandate: Preparing Design Teams for an Unpredictable Future

Dan Weiner
Dan Weiner is Director of Product Design at Kit, where he leads product design across a platform built for creators to grow their audiences, automate email workflows, and build sustainable businesses. His background spans graphic design, front end development, product design, and B2B software, giving him a practical perspective on how AI, UX, and product strategy are colliding inside modern SaaS teams. 

Support the show

Transcript

00:00:00,100 --> 00:00:02,140 [Mark Swaine]
[upbeat music] 

00:00:03,200 --> 00:00:14,020 [Mark Swaine]
This is the UX Institute Podcast, and I'm Mark Swain, founder of UXI. In this podcast, I interview UX and product leaders from around the globe. 

00:00:14,020 --> 00:00:27,200 [Dan Weiner]
Design systems are the way that agents communicate. That's the way that agents build your interfaces, and if you've been sleeping on that, then you're not ready for this next phase of, of building interfaces. 

00:00:27,200 --> 00:01:45,380 [Mark Swaine]
This week on the AI Product and Design Podcast, I sat down with Dan Wiener, director of product design at Kit, for a conversation on SaaS, UX, AI, and the future of product teams. What started as a discussion about bloated SaaS products turned into a deeper look at how companies lose clarity, how enterprise demands shape UX, and why the hardest decision is often what not to build. We also explored why so many SaaS products still feel fragmented today, the tension between enterprise growth and self-service simplicity, how AI can accelerate the wrong outcomes when UX foundations are weak, the rise of agent-optimized design systems, and how AI is reshaping design teams, workflows, and expectations. One line that stuck with me was that design systems are no longer just how designers and engineers communicate. They're becoming how agents communicate. This isn't a surface level AI conversation. It's about the reality of building products while software production accelerates faster than UX maturity, operational thinking, and organizational change. If you work in UX, product, SaaS, startups, or AI native workflows, I think you'll get a lot from this episode. Dan, welcome to the pod. 

00:01:45,380 --> 00:01:47,220 [Dan Weiner]
Thanks for inviting me, Mark. I'm looking forward to this. 

00:01:47,220 --> 00:01:53,580 [Mark Swaine]
Thought what might be valuable is to give our listeners a little bit of a background to your role at Kit today. 

00:01:53,580 --> 00:02:53,720 [Dan Weiner]
So I am the director of product design at Kit. If you haven't heard of it, it's a platform for content creators where they can, uh, build landing pages, send out emails, set up automations that send out sequences of emails, things like that. So very exciting for me to work on a platform that I actually use myself in a space that I'm very familiar with, which is great 'cause I've been working as a designer since 2007, product design probably, uh, little bit, couple of years after that. My background is really quite diverse in the sense that I've worked in graphic design, printing design for magazines. I've worked as a front end developer. I've worked a little bit in product, most of the time as a, as a designer, either as an IC or design leader, and really specialized in B2B, so u- designing a lot of products that I never actually use myself. And then at Kit, where I've been for the last year and a half, is the first time I've worked somewhere in a product that, you know, I can really resonate with and use myself, so it's really exciting. 

00:02:53,720 --> 00:02:59,520 [Mark Swaine]
Yeah. That's nice. So you're in the space, you're using the tool yourself, you know what's working, not working. 

00:02:59,520 --> 00:02:59,609 [Dan Weiner]
Yeah. 

00:02:59,609 --> 00:03:01,960 [Mark Swaine]
You can drive some thinking there. That's nice. That's nice. So 

00:03:03,220 --> 00:03:31,380 [Mark Swaine]
there's so much that we could get into, and I say this with a lot of really great people that I bring onto the pod. I guess we're both very alike in terms of the industry and that we grew up in and where we are today in terms of SaaS, AI, and everything in between. But the evolution of SaaS since 2013, '14 upwards to where we are today, even today, a decade later or more, they still feel bloated, confusing. How has that lasted this long, and what should we be doing differently today in SaaS? 

00:03:31,380 --> 00:03:40,710 [Dan Weiner]
It's a topic I love and something I'm always struggling against. If we go back to successful SaaS business and then how it started, 

00:03:41,780 --> 00:05:56,500 [Dan Weiner]
oftentimes you look at these businesses and you think that they're product led, and in some cases they might be, in some cases it might be a big mix where most of the customer count that they have is product led, and then maybe most of the a- the revenue or, like, half the revenue or something is sales led because you have these big, big customers that, uh, contribute to a lot of the, a lot of the revenue, especially in B2B, and then you have the mass of customers that generate maybe half the revenue or even less than half. So you have this, this kind of strange effect there, where a lot of the features that these people are, uh, product designers and engineers and product people are focused on building have been requested by those very high value customers. And so you end up building for the high value customers whilst also at the same time trying to be product led and trying to onboard the rest of the customers. So you have this very strange dynamic where you have maybe, like, 5, 10% of customers making up half the revenue, and so teams bending over backwards to satisfy, uh, whatever they request. And then what that means is for the rest of the customers, you know, the bulk of the customers, they're having to self onboard in a platform that has this huge disparate feature set that's been adapted to the unique needs of, of a handful of high value customers. That's one part of it. But if you think back to, like, when these products start off, even if they're not in that situation when they start off, there's this, like, desperation to find something that works. You know, like, build a bunch of stuff, see what clicks, see what people resonate with, see, see what it is that can land that first, uh, sale. And so you end up kind of just building a lot of stuff and then figuring out your identity as you go along and seeing what works. But then it's very hard to get rid of the stuff that it maybe generated a few sales, but it's not the core value anymore, but yet there's a bunch of people now relying on that, and then, you know, to undo all of that means taking things away from people, risking backlash, risking them churning. Yeah, it's like these messy beginning stages, and also the fact that a lot of product led companies, you know, it's actually really just a hybrid of sales led when it comes to high value. 

00:05:56,500 --> 00:06:33,828 [Mark Swaine]
Really good. That story alone spreads across probably the birth of so many SaaS products over the pastdecade. I see it in portfolio companies I mentor where enterprise-level SaaS products are being entertained primarily by maybe three to five implementations for specific customers who are driving those features only, and where, in some cases, startups can eventually burn out of cash, they can burn through cash faster entertaining that and not really thinking about the self-serve parallel bulk that could possibly drive revenue faster, quicker, right? 

00:06:33,828 --> 00:07:05,108 [Dan Weiner]
Yeah, and you get into this dynamic, which I've been in with a lot of companies, where they acknowledge that entertaining the high-value customers is, is very high risk because, you know, you, you start to fragment your product, and you make it, you know, little bit worse for everybody. But at the same time, nobody wants to say no to them, so you're in this catch-22, this impossible situation where you kind of want to ignore them for the long-term security of your company, but at the same time, for the short-term benefit, there's nobody saying, "Oh yeah, don't worry about that, you know, that great deal. Let's just ignore it." 

00:07:05,108 --> 00:07:05,428 [Mark Swaine]
That's it. 

00:07:05,428 --> 00:08:13,428 [Dan Weiner]
And, and then the, the other part of your que- question I think was like, well, what can we do about it? And something that is always in my head is this idea, I got it from some book, it's not my quote, the core principle of a good strategy is understanding what you're gonna say no to. One of the most valuable things about strategy is really understanding who you're not going to serve, what you're not gonna build, and so it's all of this what you're not gonna do, like all of those negative aspects, is really one of the biggest parts of a good strategy. And that really resonates with me 'cause I think that's the bit that is missing a lot of the time from SaaS businesses, is like we generate all of these ideas, and now building stuff is getting faster, and so that part is something I think the teams don't struggle with. But the part that they do struggle with is, like, we're not gonna compete with X company. We're not gonna serve this type of customer. We're not gonna optimize the platform for this type of customer. We're, we're not gonna build these sorts of features. We're not gonna chase this opportunity. And if you are very clear about that, then I think the issues that we've been talking about are less likely to occur, or you, you can find a path to fix them. 

00:08:13,428 --> 00:08:18,488 [Mark Swaine]
Very good. So at Kit today, so you have a team of designers that you manage? 

00:08:18,488 --> 00:08:24,947 [Dan Weiner]
I've got four product designers, and I'm currently hiring a design engineer, so soon to be five designers on the team. 

00:08:24,948 --> 00:08:28,428 [Mark Swaine]
So at a high level, what are your designers focusing on today? 

00:08:28,428 --> 00:09:03,048 [Dan Weiner]
At the moment, we have a lot of focus on improving what's already there because what we offer is quite varied and complex. So one of our core focuses at the moment is really making sure that the journey through the app between the features and how they connect together is intuitive, more so than building new features. So we have a squad that is working on, on a new feature, and it's something that's net new. But mainly we're trying to improve what's there and make it more efficient and really focus on how things are interconnected and the journey through the app whilst using multiple features. 

00:09:03,048 --> 00:09:22,048 [Mark Swaine]
With that then, does that involve onboarding areas, key workflow areas for multiple email sends? Are you guys getting into deeper conversation around the creator industry as a whole and what's happening with AI right now? Are designers pulling that kind of thinking in in terms of where a particular feature or workflow could go in the app? 

00:09:22,048 --> 00:10:10,318 [Dan Weiner]
Yeah, when I joined, as well as doing design leadership, I was in charge of putting together an activation squad and thinking a little bit about the onboarding and the opportunities there. We're now kind of promoting that to a more of a full-time squad 'cause we were, we're kind of running experiments and borrowing resources from other teams. And so we've just hired a dedicated PM for that growth squad, and we're gonna really have more of a dedicated focus on activation and product growth. So in that area, there's a lot of exciting stuff that we're planning using AI. And then with the creator tools and AI space in general, what we're really actively avoiding is just slapping AI, like, wrappers into the app. So, you know, like, you're composing an email and basically just offer to do it for you by using- 

00:10:10,318 --> 00:10:10,318 [Mark Swaine]
Yeah 

00:10:10,318 --> 00:11:11,168 [Dan Weiner]
... ChatGPT the same that you could just do if you were, you know, using ChatGPT or Claude yourself, and really think about what's the kind of unique value we can offer using AI. So we are using AI to help people automatically generate different subject lines, things like that, where they can test them and land on the best fit and do all of that within the app, things like that. And then, like I was saying, with activation, I think is where there's a lot of really interesting opportunities where you can have a really, like a, a personalized onboarding experience by answering the questions that you already answer, but instead of us having a set list of, like, onboarding checklists that we can offer you, and the variety of onboarding checklists could be infinite because we're basically building those checklists on the fly with everything that we've learned from support and customer growth teams about what works best for our creators, depending on, you know, who they are, what industry, what they're trying to achieve. 

00:11:11,168 --> 00:11:39,888 [Mark Swaine]
And so have you found through team conversations, both with PMs and on your side, on the design side, that product priorities are starting to shift because of AI? Or has it always been a calm, parallel workflow in terms of where AI's gonna become valuable within the app and what workflows? Or, you know, is there pressure, more speed? Is there more conversations around, "We gotta get more AI in for the sake of it," or is it measured? 

00:11:39,888 --> 00:13:27,688 [Dan Weiner]
I think we're doing it in a sensible way where it's not, like, it's not reactive, and it's not, "Hey, we've gotta add some AI in. Where can we add the AI?" It's more like there are these very valuable moments for creators. You know, there's a moment when they've, let's say they've imported their subscribers. That's a real sign of trust. And to get them to that point, is AI gonna help accelerate, build trust, make it easier for them? Or for them to send their first broadcast or something where I think there's a lot of potential is setting up an automation. So in Kit-The sort of golden path in terms of conversion and seeing value from the platform is that you set up a Kit landing page, you use that landing page as a trigger for an automation. So basically, when somebody subscribes via the form on that landing page, they'll be entered into this automation where they'll receive a sequence of emails, let's say like one per week for the next four weeks or something. If you set all of that up, you really see the value of Kit, like that's kind of Kit's differentiator. You've got your audience, you set up your lead magnet, your landing page, and, you know, while you're asleep, people are getting emails sent out to them and, you know, acting on those emails, and you can kind of like scale your business beyond the regular thing you might have been doing, you know, manually sending out an email or posting on, on social media. So if we can get people to that stage quicker, they can see the value quicker and... But it's very complicated to set all that up. So for example, if we can get people just to describe what it is they're trying to achieve and start to build some of those connections between the different features for them, that's a great example of, we know what a valuable outcome is, can AI help us accelerate towards that outcome or not? Rather than, you know, starting with the AI and thinking, "How can we use AI somewhere?" 

00:13:27,688 --> 00:14:18,528 [Mark Swaine]
I guess everyone in SaaS, either startup or scale-up Y-series onwards are in a rush panicking, "Where do we use AI? We have to be AI-led." [laughs] Some workflow in here has to be repurposed or wrappers used for whatever sake to say that AI has been integrated. But to your point, which is really valuable, which is more of the conversations I'm in today, is where is AI actually valuable in the workflow? Where is it going to enhance or speed up? Even if it's semi-agent orchestration around a particular workflow that could be implemented, can the users use it? Do they know how to set it up? When does it become valuable? Is it worth their time? Is there more time setting up agent orchestration for those workflows rather than just a simpler AI implementation to support them and guide them? 

00:14:18,528 --> 00:15:32,607 [Dan Weiner]
Yeah. And I think if there's teams that are struggling with this, you know, it, it could be they just don't have anybody on the team who has the familiarity or imagination of how to use AI. But probably that's not likely, 'cause I think we're all getting a lot of exposure and we can kind of see the, the possibilities. I think the problem is more fundamental, and this is teams that have ignored fundamental UX practices, so they don't have a clear list of what are the aha moments for your users. What are the moments that the users really perceive value from your product? What are the customer journeys that are high friction for your users, but once they get to the end, are worth it? If you have a clear map of those journeys and you have a clear understanding of what users are trying to achieve when they come to your platform, but you know where they're struggling, where there are barriers in their way, then you have that map for how you can use AI to accelerate towards value. If you don't have that, 'cause you don't have that fundamental UX practice at your company, basically you don't know what your users want, so you can use AI to accelerate towards something, but it could just be random stuff, and you could just be using AI to go faster down the wrong path. 

00:15:32,608 --> 00:15:58,648 [Mark Swaine]
Do you or the team at Kit feel, you know, with the indie creator economy today, that requests or workflows are evolving fast? Is there change coming fast and hard around how users are going to manage email lists? What level of control do they still want, or are they willing to depart to AI? Are you feeling signals in the industry that way? Are other platforms popping up doing similar stuff or doing things differently that are interesting? 

00:15:58,648 --> 00:16:04,548 [Dan Weiner]
Yeah, I think there's something about email which is a little curious in this aspect, where 

00:16:05,668 --> 00:16:44,468 [Dan Weiner]
if you come to Kit, what you get is the best deliverability rate on the market, and that is built up over time. It's time and trust and experience and, like, slowly iterating. When we get a new customer, we don't just, like, send all of their emails out from a Kit domain. We slowly transfer them over, we monitor, you know, do 5% of their list at a time, things like that. So there's, there's like these practices that when you go online and you see people, you know, using OpenClaw to, like, set up their business from scratch, zero to one, like, "I made $20,000 in a, in a week," et cetera. 

00:16:44,468 --> 00:16:44,508 [Mark Swaine]
[laughs] 

00:16:44,508 --> 00:16:58,088 [Dan Weiner]
That's great when you're going from zero to one, when trust doesn't matter, security doesn't matter, you don't have a reputation. I think this is like a common thing all over the internet now, where designers, product people are looking at these stories online and they're saying- 

00:16:58,088 --> 00:16:58,098 [Mark Swaine]
Totally [laughs] 

00:16:58,098 --> 00:17:45,228 [Dan Weiner]
... "Wow, that's amazing." But then you're thinking about your own business, some of the things that actually lend value to your business are a little bit more difficult to just experiment with AI with, or there's, like, security risks, or you really need to be very sure about the outputs and you can't have, you know, outputs that are 90% of the time right, the other 10% are, you know, just random mishaps. So I guess that's a bit of context for to, like, hedge the answer where, yes, things, things are moving fast, but there are certain things that are not replicable. You can't just, like, spin up a lovable app and, you know, send out a million emails a day and expect to have a good, good deliverability and a, you know, good sender reputation. 

00:17:45,228 --> 00:18:06,468 [Mark Swaine]
No, it's a really good point. It's kind of funny you raise about the, the creator videos. Even if, you know, you jump into YouTube every day, there's lists of new people, "Hey, I've just built an app in two days or nine days and I'm at 20K MMR after three months," and et cetera, et cetera. There's an endless feed of this stuff. 

00:18:06,468 --> 00:18:10,958 [Dan Weiner]
To be honest, I love this stuff and I'm kind of jealous, but [laughs] 

00:18:10,958 --> 00:18:37,480 [Mark Swaine]
Totally get it, 'cause I'm feeding into it. But then to your point, there is a harsh reality around this stuffIn terms of how it's built, whether the momentum is there within that particular industry or area, and how you can align to something like that, and do you have the skill sets as quickly? You know, I'm hearing some people, coders, they're up and cloud code till 3:00 AM trying to build stuff, and you hear a lot of this stuff at the moment. I'm like, "You have to have a normal life, too." [laughs] 

00:18:37,480 --> 00:18:45,220 [Dan Weiner]
I listened to an interview with the, uh, the OpenClaw founder, and he was spending up to 20K a month in tokens sometimes to- 

00:18:45,220 --> 00:18:45,450 [Mark Swaine]
Right, right 

00:18:45,450 --> 00:18:46,310 [Dan Weiner]
... to build OpenClaw. 

00:18:46,310 --> 00:18:46,960 [Mark Swaine]
Right. 

00:18:46,960 --> 00:18:55,980 [Dan Weiner]
It's not like, you know, just, you know, use your ChatGPT subscription and become a millionaire overnight. There, there's a little [laughs] backstory to it all. 

00:18:55,980 --> 00:19:14,300 [Mark Swaine]
It's funny you say that. One of the CTOs I work with, there's a prototype we're gonna build together of a particular product we have. It's a Chrome extension idea. One of the things, the back and forth was token cost, token cost, shit, how are we gonna get there? It's a really good point to bring up because it seems to be bubbling up to the top more and more. 

00:19:14,300 --> 00:19:52,240 [Dan Weiner]
I imagine that the cost of AI is going to increase at the same time as adoption increases and, you know, people all over the company will be starting to use Claude Code and Cursor and, and these tools. And so I think there will be sort of a reckoning where we look at how much are we spending on AI, how much are we getting back, and I'm sure that, you know, we will see that there's productivity gains and, you know, using AI is... I'm, I'm not saying, like, we're gonna just stop using AI. I think it's here to stay and it's gonna get better and better and it's gonna add more value, but I think it's going to start to get folded into the business like other tool costs, you know- 

00:19:52,240 --> 00:19:52,880 [Mark Swaine]
Yeah, yeah 

00:19:52,880 --> 00:21:08,100 [Dan Weiner]
... where if I'm saying, "Hey, I wanna switch," we have started to use a user research tool. With any tool, you have to make a case for why do we wanna spend this much money per year on this tool? What's the benefit gonna be? Et cetera, et cetera. And I think with AI a- around the world it's not like that right now. It's more like, hey, we've got to experiment, we've got to be curious, we've got to try stuff out, we've got to see what workflows we can accelerate. And then I think that conversation's gonna shift in the future where it's more like exactly what are we gonna use this tool for? How is it gonna accelerate? Uh, where are we gonna invest money? Where are we gonna hire, like, dedicated people to unlock certain workflows for the rest of the team? I'm really interested in how do we work on our code base so that it is agent optimized so that a designer can come along and create a pull request that engineers look at and it just feels right. And I think for a lot of companies with a big legacy code base, they're not at that stage yet. So you do like a zero to one project and it's, it's fantastic, and the results are, you know, just... it can feel magical. And then on a big code base that maybe doesn't even fit into the context window and it's got a lot of old code that you don't want the agent to look at, but it ends up replicating and you don't necessarily have, uh, a design system that's in sync with Figma and all of this stuff- 

00:21:08,100 --> 00:21:08,530 [Mark Swaine]
Yeah 

00:21:08,530 --> 00:21:22,080 [Dan Weiner]
... you end up with the, uh, non-engineers trying to use these tools and not getting the results they want, and then engineers that are maybe skeptical looking at these pull requests and feeling validated like, "Oh yeah, I knew this was, was a bad idea." 

00:21:22,080 --> 00:21:23,290 [Mark Swaine]
[laughs] Yeah. 

00:21:23,290 --> 00:21:44,500 [Dan Weiner]
Uh, but there's all of this, like, foundational work to be done to optimize for AI, which, you know, I, and I think we're moving now into that stage where it's like, okay, being curious, experimenting was great, but now how do we actually do the difficult work of, like, laying the foundation so that we can benefit from this going forward as a, as, like, a serious business tool? [upbeat music] 

00:21:44,500 --> 00:22:19,280 [Mark Swaine]
Visit uxinstitute.com to sign up to our new newsletter. UXI will be sharing best-in-class UX research practices, as well as in-depth interview content so you can excel at your career in UX. I'd like to get into a little bit more about the workflow practice with the design team at Kit and how they spend their day. Are they using current AI tools to support UX research? Are they using any prototyping tools to spit out quick prototypes and flows, you know, typically out of Replit or Lovable or whatever it might be? Has much of that c- come into the workflow? 

00:22:19,280 --> 00:24:31,644 [Dan Weiner]
At the moment we're at the stage where everybody's sort of experimenting and finding things that work for them, and there's no sort of, like, unified thing that all the designers are doing that they're finding valuable. So for example, recently one of the designers used a script that Claude Code generated that allows them to export some users from Mixpanel. So s- let's say, you know, users that bounced at a certain step. We can export those email addresses, they can write a little script that e- emails them from his own Gmail to get some feedback, book some calls. The follow-up emails and the call of course is, is very much a, a human one-to-one connection, but that initial email, uh, is something where they're just sending it from their personal email through a script and then pulling all of the responses into a spreadsheet using AI, analyzing, clustering response types together, and understanding, "Oh look, 34% of users said they bounced because of this reason." So it's stuff like this where I think previously you would probably have to use a whole collection of tools or you would have to, like, go through and, like, tag responses manually or something. So what I love about that is it's not replacing research, it's traditional UX research, but accelerating the timelines, and that's something that we've always complained about, right? Like, we wanna do the research, but it takes so long and, like, people are waiting and stakeholders want results and it's... Now you can do what used to be, you know, a couple of months, you can do it in a week, which I think is, is amazing. And then on the other side we have designers using things like Figma Make to prototype. I've been playing around a lot with Magic Path, importing things from Figma into Magic Path. Or an example is when we have legacy screens where we can't find the Figma files or Figma files don't exist or... I can just import them using the Magic Path, uh, browser into Magic Path, and then you can edit them there. That's been great.I haven't had the same success with browser to Figma, but because I think Magic Path is building in React, it seems to work really well. 

00:24:31,644 --> 00:24:56,304 [Mark Swaine]
It is. It's building components in React, and, um, I've experimented a bit with Magic Path as well, and it's working really well for singular component and library building and everything in between. It's, it's, it's really good. Are you finding the design team is spending more or less time in Figma for new journeys, new flows? Are they pumping them out in the likes of a Lovable or similar tools and bringing them into Figma and trying to expand on them there to speed up everything or...? 

00:24:56,304 --> 00:24:59,764 [Dan Weiner]
No, it's still mostly in Figma, and the reason for that is- 

00:24:59,764 --> 00:25:00,293 [Mark Swaine]
Okay 

00:25:00,293 --> 00:27:32,854 [Dan Weiner]
... when I was saying, like, y- there's this difficult work to, like, lay the foundations, connect everything together, we still want things to look the way they're supposed to look in production. So if you go off and you're like, Lovable or Replit or Cursor or Magic Path or whatever, and just letting the system kind of generate screens for you, you... There's, like, this little gap, right? It gets you, like, 90% there, the intention is there, but it's not... If the engineer's like, "Did you really want, you know, to break the normal spacing rules and have, like, this 48 pixel gap here?" Then you'd have to be commenting everything, single thing, and say like, "No, don't... Like, actually don't follow the prototype here," or, "Don't follow it here," 'cause it's not quite right. So what we've found is most of the time is spent in Figma, because Figma is the thing that still gives us that high fidelity output that is exactly what we want in production. But as somebody that's spent a long, long time working in front end development, the irony of this isn't lost on me, the fact that you basically create a bunch of rectangles and ti- and text in Figma to then have a engineer look at it and recreate a bunch of rectangles and text in React. You know? It's like, well, I think because Figma took a decision for speed, uh, a long time ago to use WebGL to render everything, and so now we have, like, you know, render your rectangles in WebGL, and then copy that, render them in React. And so it feels like we shouldn't be doing this anymore. But still, that's the majority of, of the, where the work's happening. There is something that I'm really pushing my designers to do, and I would, you know, encourage every designer to do it, which is to get into the habit of also using AI to prototype things that you position just as, like, what things could be. Like, what about this idea, or, you know, this is interesting, but not the thing that we're gonna necessarily build. And, like, just keep spinning things up every week like that to just have a conversation, uh, about, like, what the future could be. This is something I've always been pushing for. It's just easier now with AI. The... One of the jobs of a designer is to start conversations. A lot of the time, designers, they feel like their role is, like, providing blueprints for engineers, and it's like that is an important part of the role. You need to help get things built, but you also need to be pointing towards the future, like helping people envision what the future could be, and even if it's like saying, "Oh no, that's not it, because this, this, this," that's helpful. Now we know, like, what people don't want to have in the future. So I think AI is amazing for that. 

00:27:32,854 --> 00:28:15,844 [Mark Swaine]
I think it's great the way you framed all of that. It's such a great distinction that you made, because that's where my head is at most of the time today, is that I can spit out onboarding flows, dashboard screens, whatever I want. I can benchmark and I can get Replit or Lovable or whatever to Cursor to make it the way I want and look it pretty good, and I can spit all that out. But at, to your point, which is the workflow piece that I don't think is going away anytime soon, um, is that the foundational production ready implementation of those journeys or flows or dashboards, whatever it is, they still have to be cleaned and readied in Figma. Would you agree foundationally? 

00:28:15,844 --> 00:29:26,584 [Dan Weiner]
At the moment, yes. So I mentioned I'm hiring a design engineer, and a big part of that role description is to bridge the gap between design and engineering and see how we can use AI to do that. So I want to have somebody on the team that is a designer, is a coder, is very into using AI to accelerate workflows, and to be really s- spending a big part of their time thinking about how we can bridge that gap. 'Cause I think that it doesn't happen accidentally. If you don't have somebody focused on it, then we'll always have that gap. So it, again, I think it's, we're in this stage now of, like, the very gritty, like, hard work to actually get some of these tools to provide business value. We're, we've been at the stage for a while of, like, exploring curiosity. You know, is this interesting? Is it not? And I think now we're at the stage, like, yes, this... At least in established companies, where we know it's interesting, we know there's potential, but we have to accept now that if we want to use these tools, we need to invest dedicated resources to actually laying the foundations for these tools to be, to be useful. 

00:29:26,584 --> 00:30:12,784 [Mark Swaine]
That's the best way I've heard it put in, uh, quite a while. There's so much talk daily, weekly, in terms of model updates and capabilities, et cetera. Even building out website landing pages in CloudCode, it's doing some beautiful stuff, right? But at the same time, the foundation's not been laid for the team internally. They need a central point, a workflow point, and this is where, unfortunately for people like us, sometimes you tend to live in the future. You see all the possibilities and opportunities, but you constantly have to pull yourself back to the daily workflow and say, "Is that relevant now? What is valuable?" I can take it 70, 80% of the way concept-wise by prompting. I can do a pretty good job to get a sign-off, but I need to actually prep it correctly and readied. 

00:30:12,844 --> 00:30:50,374 [Dan Weiner]
At Kit, we meet up in person, uh, twice a year. Last summer when we met up, we had a week, we call it Create Week, and you get to basically experiment for a week to try and create something interesting. Uh, and it doesn't have to be something that you plan to put on the roadmap. And for that Create Week, me and the, the group that I was in, we were trying to turn-Uh, a Figma component into, like, production-ready code. And that was interesting for me, and changed entirely the way I think about design systems, because if you use the Figma frame link or the Figma MCP with Cursor, you will get something that looks pretty close, but 

00:30:51,444 --> 00:31:07,893 [Dan Weiner]
Cursor doesn't necessarily know how to write the code. You want that color, like what color variable is it? So ended up making a lot of, like, very kind of brittle connections, you know, in, in the agent file to say, you know, "If this, then do that. If this, then do that." 

00:31:07,893 --> 00:31:07,924 [Mark Swaine]
Yeah. 

00:31:07,924 --> 00:32:49,944 [Dan Weiner]
Because there wasn't really m- like, we weren't using all the Tailwind variables in Figma. The components that we had in Figma weren't... didn't necessarily have exactly the same names as the ones in React, because these are two teams that they've been taking care of their own business, not thinking about, how does my work compare to the front-end engineer's work? And to be honest, not, not really investing that much in the design system because at a, like, a very busy startup with a, a big complex app and a small team, you're not necessarily thinking that the most im- important thing is, like, keeping this design system up to date and making sure that it, the components, the naming is one-to-one between design and code. That feels like over-engineering, like, you know, too much tidying your house when it's already tidy kind of thing. And then after that week, I've been thinking about design systems completely wrong. Design systems are not the way that designers and engineers communicate anymore. Design systems are the way that agents communicate. That's the way that agents build your interfaces, and if you've been sleeping on that, then you're not ready for this next phase of, of building interfaces. For me, it's like it... since that, it's just been a huge, uh, thing that's on my mind that obviously, like, now I'm hiring for somebody who is really thinking in that space that can help us accelerate those workflows. So think back to your original question, like, yes, right now, you know, I think designers have to be mostly in Figma 'cause we still need to produce things that are a clear blueprint for engineers. But I, I feel like at some point you could just say to Claude Code, you know, "Here's my design system in Figma," 

00:32:51,004 --> 00:33:19,304 [Dan Weiner]
and describe the journey, have a bunch of, like, skills for journey, customer journeys, and maybe you'll get something out the other end that works. And I think there's gonna be these tools, I mean, there already, already are tools like Paper, but I would love to see tools that, like, generate customer journeys on canvases so you can see multiple journeys side by side. But I think we're gonna move into an era where AI will generate journeys for you and help you visualize them, and, and I think it's gonna be pretty exciting. 

00:33:19,304 --> 00:33:35,984 [Mark Swaine]
Is there any major areas of impact that you've implemented recently in Kit for major changes around particular workflows that design has really driven huge impact, and that it got back to the org as a whole, y- you know, how design is considered to be so important at Kit? 

00:33:35,984 --> 00:35:45,804 [Dan Weiner]
Yeah. Uh, I mean, one s- comes to mind is that we have a new feature that is gonna come out soon for bigger creators that was pretty much entirely designed and built by designer PM, uh, together. So it started off as a Replit app. It's getting more and more serious, and it's gonna be eventually, like, a premium Kit app. So that's been a really interesting example where a designer was involved from the start with doing the research, interviewing customers, designing and building the thing, testing out the built app with customers manually. You know, there was no data sync, just manually importing their data, getting them to play around with this. So doing, doing the interviews where they're playing around with it, leaving them with the app, with access to the app, coming back and interviewing them later, iterating on it. And so that, I think, is a, is a kind of a very great example of how AI can empower a designer to do their normal UX foundational practice, but also build this thing in a couple of weeks, and then go back to the customer and test it out with the customer, and iterate. And, and that is an example where most of the work was not done in Figma. And then generally, like within Kit, the way that we have the squad set up is that the designer, engineering manager, and product manager sort of together as a trio lead that area of the app. So for instance, we have a, a builders team, and they're in charge of all, all things publishing, so publishing a landing page, sending out emails. And the designer for that area has a lot of say over the vision for what we're gonna build, helping prioritize. So I would say that designers are empowered in, in their squads to really be, like, leaders of that area of the product, uh, from a design sense. So it's a, it's a really empowering way to work. Uh, and obviously it's one that we can benefit from as a, as a small company. 

00:35:45,804 --> 00:36:05,884 [Mark Swaine]
Just a couple of quick final quick-fire questions, if you don't mind. So the current era that we're going through AI-wise, do you feel after this cycle winds down a bit, some norms are set, that this has empowered designers more or it has commoditized design more? 

00:36:05,884 --> 00:36:11,904 [Dan Weiner]
It's a good question. I think I'll probably give you a different answer like every week for the last [laughs] year or something. 

00:36:11,904 --> 00:36:11,984 [Mark Swaine]
[laughs] 

00:36:11,984 --> 00:37:26,676 [Dan Weiner]
I think more and more my feeling is that it's not gonna be too bad for designers. It seems like design hiring is on the up. I think that what people are realizing is what I mentioned before, like AI can accelerate your way to the destination, but if you don't really know-what your customers want. It's not that helpful. And now not just the visual design, but the connectedness of the experience, the visual design, all of that is a real giveaway for how much sort of real craft and design experience has been put into the experience. So I think we're gonna see that building the thing is really easy, but to actually make it feel unique, feel like it connects with the user, speaks to them in their voice, is intentional, and not just a sort of like off-the-shelf app, that's gonna be really difficult. I think design is in a good place, but I think what's difficult is companies are gonna be looking to hire a lot of designers. I think there's gonna be more and more tech companies because of AI. They're gonna be trying to hire designers, but there's gonna be this very uncomfortable 

00:37:27,876 --> 00:38:14,696 [Dan Weiner]
issue that we're gonna have where designers with experience don't necessarily have the experience that the companies are trying to hire. I think that's very sad for a lot of designers, and I feel a lot of empathy for them. I- if you're working at a company that, that doesn't, isn't focused on those foundational layers, can't really use AI to get any benefits, and you're working at that company and, you know, you're busy and you're being stretched already, and people are saying, "Hey, you should, you know, be more curious. You should be showing your value with AI." And it's like, well, I'm already working 10 hours a day. I'm stressed out. I wanna ha- you know, hang out with my family. We're not really getting any value from AI in my job. Like, what am I supposed to do to turn into that designer that people are looking for? 

00:38:14,696 --> 00:38:18,326 [Mark Swaine]
Yeah. That's a great point. Yeah, yeah, yeah. 

00:38:19,416 --> 00:38:29,956 [Dan Weiner]
I think the future's looking bright for designers. I just... The question for me is, like, what kind of designers? Are they the designers that we're thinking of, that we recognize, that... Or is it a different kind of designers? 

00:38:29,956 --> 00:38:42,796 [Mark Swaine]
And can designers go the pace? Can they put the dedicated time into this new way of working experimentally, exploring tools, models weekly? I do know designers need to move fast. 

00:38:42,796 --> 00:39:57,956 [Dan Weiner]
Yep. And I, I think for, for me, I feel like there's a bit of a responsibility. As a design leader, I always think to myself, I need to help designers have a better career, not just perform better in their role now with me. I need to be setting themsel- them up for a better career. So if I hadn't been their manager versus being their manager, their career is, is in a better place, uh, after working with me. That's my goal. I believe that it's one of those sort of like win-win things that you can do because for me, if I can make designers more successful, it probably means they're adding more value at Kit and adding more value to their role, and then it also means that, you know, they hopefully have better careers. So I feel like it's a way I can add value in both places. And so now I feel there's this real responsibility where you've got to try and enable the conditions for that designer to be successful in this new future, whatever that is, which is so uncertain that we're all trying to figure out. Any design leaders are listening, I think we have this responsibility to make space for designers to figure out how they can use AI in their workflows, how they can have a case study where they can demonstrate that they provided genuine value to their customers but accelerated it using AI or increased efficiency or, or something like that. 

00:39:57,956 --> 00:40:13,896 [Mark Swaine]
Yeah. We'll be experimenting for the next couple of years to see where roles land, and I think it would be great to chat again to see maybe in the future how the Kit team is evolving. So look, thank you again, super conversation, and, uh, I look forward to chatting again, Dom. 

00:40:13,896 --> 00:40:15,535 [Dan Weiner]
Thanks, Mark. It's been a pleasure. 

00:40:15,535 --> 00:40:39,976 [Mark Swaine]
[upbeat music] If you've learned anything from this episode, I ask one small favor. Please leave a five-star review on Apple Podcasts or share this episode across your networks. It will really help the growth of the podcast and will allow me to continue to secure amazing guests that will part their insight with you. Thanks for listening to the official UX Institute podcast, and talk to you next time.