设计师的超级力量：AI 时代中的同理心

这件事就发生在一次会议之后，设计那边的 VP 们告诉我，这不可能做成。我说：“嘿，你们看，我不是工程师，但我刚做出来一个 SVG，它可以染成任何你想要的颜色，可以透明，可以用上所有品牌色，商家完全不用动手。”我们还做了一个全新的在线结账页面，直接展示给商家，像是在问：“你喜欢这样吗？如果不喜欢，这里还有其他颜色可以调。”所以这就是那个世界。我也不知道自己现在为什么这么兴奋。我觉得设计师真的拥有这种超能力。

And this was literally right after a meeting where I was told by VPs and above on the design side that this couldn't be done. I was like, 'Hey, check it out. I'm a non-engineer, but I just built the ability to create an SVG that can be colored to any color you want, can be transparent, can have all these brand colors, and a merchant will never have to lift a finger.' And we have a new online checkout that we just present to the merchant, like, 'Do you like this? If not, here are some other colors you can tweak.' So that's that world again. I don't know why I'm so excited right now. I think designers literally have this superpower.

欢迎来到 Dive Club。我是 Rid，这里是设计师永不停歇学习的地方。本期的嘉宾是 Ron Goldin，他曾领导 Google、Shopify、Uber Eats 的设计团队。在我看来，他是这个新 AI 世界中作为设计领导者欣欣向荣的完美典范之一。我们将深入探讨他作为构建者的旅程，包括他如何从零开始打造一款名为 Matchmaker 的新产品。但在那之前，我先请 Ron 分享一下他早期在 Uber Eats 做设计师的经历。

Welcome to Dive Club. My name is Rid, and this is where designers never stop learning. This week's episode is with Ron Goldin, who's led design at Google, Shopify, and Uber Eats. In my opinion, he's one of the perfect examples of what it looks like to thrive as a design leader in this new AI world. So we're going to do a deep dive into his journey as a builder, including how he built a new product called Matchmaker from scratch. But before we get into all of that, I asked Ron to share a story of what it was like as an early designer working on Uber Eats.

我记得入职第一周，我住在纽约市。我这辈子从没做过配送。我想试用自家的产品，于是下载了 Uber 司机端 App，自己走了一遍流程。纽约到处都是 Citi Bike，你基本上可以借一辆，能骑 45 分钟之类的。所以我自己也骑了一趟。首先，我们都是设计师，这种共情体验太强了。新手上线流程糟透了。我好不容易才把司机端 App 搞明白。我按下一个写着“上线”的按钮，然后就坐在那儿，什么都没有发生。我怎样都接不到单，因为 Uber Eats 当时规模还非常小。所以我基本就放弃了。而这正是我 Uber Eats 生涯头几个月的开端——与其说是设计角色，不如说是研究角色。因为 Uber 当时虽然是一家相当大的公司——Uber Rides 覆盖了几百个城市——但它的行事方式仍然很像创业公司。司机端 App 上其实只是加了一两个按钮，让你能接配送单，除此之外没有任何为骑手量身定制的东西。我发现确实如此，于是开始更多地去跑配送——先是在纽约。后来我开始带一些同事一起去跑。我们真的接到了实际订单，把餐送到人手里，但报酬不高，这又是我和 PM 们聊的另一条线。不过后来这件事变成了一项国际化研究。我们的一大挑战是如何把业务扩张到美国以外的城市。一种方式显然是直接开开关，但另一种是亲自看看在墨西哥城、东京、曼谷配送到底是什么感觉。我们真的去了这些地方，然后发现：“天哪，这完全是另一种运动。”在东京那种摩天大楼片区配送时，从楼上大门到对方公寓门口可能要走 15 分钟，中间要穿过好几道安保门。而且没有任何地标。所以我们真的带着翻译在这些城市里骑来骑去，想搞清楚方向，躲开车辆，差点送命——这些都有。我特别想做的一件事情，就是想办法把那种恐惧、焦虑、在自行车上几乎丢掉性命的痛苦，带回到总部，可以这么说。我不知道你还记不记得当时的 Google Cardboard——不过，呃……

I remember my first week on the job, living in New York City. I had never done a delivery before in my life. I wanted to dogfood the product, so I downloaded the Uber driver app and just went through the process myself. In New York City, you've got these Citi Bikes everywhere, so you can basically check out a bike and you get it for 45 minutes or whatever. So I remember taking a ride for myself. And first of all, you know, we're both designers. The empathy building was extreme. The onboarding was awful. I finally ended up getting the driver app to work. I pushed a button that says 'go online,' and I'm just sitting there, nothing is happening. I could not get a delivery job for the life of me, because Uber Eats was still pretty small. And so I basically just quit. And that's kind of the beginning of — I think the first few months of my Uber Eats tenure were really more like a research role than a design role, because Uber at the time really had this strategy: even though it was a pretty large company — Uber Rides was in hundreds of cities — it still acted very much like a startup. So on the driver app, they had literally just added a couple of buttons so you could pick up a delivery, but there was nothing bespoke about it that really catered to the couriers. And so I found that to be true, and I started doing more of these deliveries — first in New York City. Then I started bringing some of my colleagues along with me to do the deliveries. We ended up getting actual gigs and dropping them off to people, and didn't get paid a whole lot, which is part of a whole other thread of conversation I had with my PMs. But it actually turned into an international research endeavor as well. A big part of our challenge was trying to scale to non-U.S. cities. And one way to do it obviously is just flip a switch, but the other is to actually see what it's like to deliver in Mexico City, in Tokyo, in Bangkok. And these are places we actually traveled to, and of course found, 'Oh my god, this is a whole different sport.' When you're doing a delivery in the skyscraper neighborhood of Tokyo, the journey from your front door to the person's actual apartment can be a 15-minute trek through multiple gates of security. There are no landmarks. And so we would literally ride around these cities with translators, trying to figure out which way to go, dodging traffic, almost dying — all of that stuff. One of the things I really wanted to do was figure out a way to bring that feeling of fear, anxiety, the pain of being on these bikes and literally almost dying, back to home base, so to speak. And I don't know if you remember Google Cardboard at the time — but, uh...

哦，哇。是啊。

Oh, wow. Yeah.

是的，是的。我用非常早期的 Insta 360 相机录下了所有这些行程。我把相机装在自行车上，真的去记录整个过程——从躲闪车流，到地图指错路不知道该往哪走，再到被保安拦住说“没有证件不能进来送餐”，全都拍下来了。我就是想把这一切记录下来。然后我把 Google Cardboard 和我上传到 YouTube 的 Insta 360 视频，寄给了我们“Uber 一切”的负责人——这个部门当年就叫这个名字，覆盖 Uber Eats、Uber Rush 以及一堆其他产品。这位 Jason Droy 真的看了。他真的看了我坐在洛杉矶一辆车里的视频——在洛杉矶最大的三明治餐厅外，和司机一起待了 3 个小时，一个订单都没接到。而且这实际上——我是说，这是一个完美的例子。我发给他的正是这段，因为在那之前，我们的假设是服务那些大餐厅，对吧？它们有最多的客流，对吧？所以我试图去模拟那种情况，而这个人真的在车里受罪。他还没吃饭，想上厕所，却什么也干不了。但是等等。这就是一次相当深入的产品和设计调查的开端，我们想搞清楚为什么会这样。我们后来意识到——现在可能算是常识——如果所有骑手都被派去热门餐厅，那骑手太多，而餐不够。对吧？而且事实上，当时骑手端 App 里并没有一个很好的可视化界面，告诉人们哪里有需求。所以人们根本不用 App 去了解需求。他们只是去 Google 地图或 Yelp 之类的平台找热门餐厅，然后把车停在那儿。你在自己住的城市可能也见过——我家附近就有人把车停在 Shake Shack 外面，就这么等着、盼着，对吧？所以这是一个长期项目的开始，最终我们做出了一种相当可靠的可视化方式，来展示我们所说的“热点”——不是那些你想象中肯定是热门、需求量大的地方，而是像一家小小的夫妻老婆中餐馆，突然涌出一小波需求的地方，对吧？这样一来，骑手们就可以开始把一个城市的配送力量分散开来了。

Yeah, yeah. So I had recorded all of these trips using a very early generation of the Insta 360 camera. I would mount it to the bike and record the trips, to literally capture what it's like — everything from dodging traffic, to figuring out which way to go with maps telling me the wrong directions, to even having security guards tell me, 'No, you can't come in without certain credentials, to deliver food.' So I was just capturing all of this stuff. And I sent this Google Cardboard, along with the Insta 360 videos that I'd uploaded to YouTube, to our head of "Uber Everything" — it was called that, covering Uber Eats, Uber Rush, and a bunch of other products. This Jason Droy — he watched them. He literally watched videos of me sitting in a car in Los Angeles, outside the biggest sandwich restaurant in all of LA, for three hours, in a car with a driver, and not getting a single gig. And this is actually — I mean, this is a perfect example of this. That was the one I sent to him, because our assumption up until that point was to serve the big restaurants, right? They're getting all the traffic, right? So I was trying to replicate that, and this guy literally was suffering in his car. He hadn't eaten, he had to go to the restroom, and he couldn't do anything. But wait. So that was the beginning of a pretty heavy product and design investigation into why this was happening. And what we realized — which now is kind of common sense — is that if all the couriers are being told to go to the hot restaurants, then there are too many couriers and not enough meals, right? And in fact, we didn't really have a good visualization at the time on the courier app to even tell people where the opportunities were. So people weren't even using the app to understand demand. They were just going to Google Maps or Yelp or whatever to find popular restaurants, and they would just park there. And you might see this in your own town — down the street from me, people park outside Shake Shack, just waiting, hoping, right? So it was the beginning of a long project that ended up with a really robust way to visualize what we called "hotspots" — places that aren't the tried-and-true places you might imagine have a lot of demand, but places that might even be a small mom-and-pop Chinese restaurant that suddenly happens to see a little pop of demand, right? And so it was a way for people to start to spread out their efforts across a city.

另一个特别有意思的例子和生活方式有关。以往人们会怎么做呢——这真的很疯狂——洛杉矶有个家伙要开车一个半小时去吃一家三明治店，那家店几乎没什么生意，因为他觉得在加州文图拉自己家附近没有足够的需求。可你猜怎么着？其实有。只不过我们以前没有工具来表达这一点。所以现在他可以待在自己家附近，单程就省下一个半小时的油费，还能实实在在支持本地商家，回家也算少了一段通勤。

The other one that was really interesting was a lifestyle thing. What people used to do—and this is crazy—this guy in LA used to drive an hour and a half to get to a sandwich shop that had no business, because he figured in his own neighborhood in Ventura, California, there wasn't enough demand. Well, guess what? There was. We just didn't have the tools to say it. So now he could stay in his neighborhood, save an hour and a half of gas each way, actually support local businesses, and have less of a commute to get home.

插播个简短消息，然后我们继续。Jitter 刚发布了动画组件，你可以一次性设计，在文件里复用，并且通过一次编辑更新所有实例。我个人非常激动，因为我到处都用 Jitter，从给 in-flight logo 做动画，到给 dive 新闻稿里的按钮做动画，都是用它。所以这会让所有内容保持同步变得轻松无比。今天就可以用了，只要访问 dive.club/jitter 就能开始。我期待 Desen 的新版本已经整整一年了。它叫 Surfaces，能让你直接在现有的生产界面上设计和做原型。你需要的所有页面和流程都预装在 Desen 里作为起点。所以你可以轻松做出修改、探索想法，然后准备好了之后，一键分享你的原型。Desen 是唯一一款可以让你直接在从生产环境抓取的真实页面和流程上做设计的设计工具——这对团队来说意义重大——今天就可以开始。访问 dive.com/desen，也就是 d-s-n。

Real quick message and then we can jump back into it. Jitter just released animated components, so you can design once, reuse across your file, and update every instance with a single edit. I'm personally very excited about this because I use Jitter for everything—from animating the in-flight logo to animating the buttons in the dive newsletter. So this is going to make it a breeze to keep everything in sync. And it's available today. Just head to dive.club/jitter to get started. I've anticipated Desen's recent release for a full year. It's called Surfaces, and it enables you to design and prototype directly on top of an existing production interface. All the pages and flows you need are preloaded inside of Desen as starting points. So you can easily make changes, explore ideas, and when you're ready, share your prototype in a single click. Desen is the only design tool that allows you to design on existing pages and flows pulled directly from prod—which is a pretty big deal for teams—and you can get started today. Just head to dive.com/desen. That's d-s-n.

回到正题。从你那么快就说‘我干脆把整个流程从头到尾做一遍’来看，你骨子里就是个研究者。但这个故事的另一个版本肯定是：你把东西写进一份漂亮的备忘录，发给领导层，他们边读边点头，然后它就被丢进某个电子表格的角落里，可能两三个季度之后才会再想起来，对吧？所以我觉得特别有意思的是，你拍下了全部视频——你说，‘我要非常刻意地去思考怎么把这种共情传递出去。’最终看来，这在推动人们真正采取行动方面起了很大作用。

Now on to the episode. You can tell that you're a researcher at heart, just how quickly you said, 'I'm just going to do the whole flow end to end.' But there's definitely a version of this story where you put things into a nice memo, sent it off to leadership, they read it and kind of nod along, and then it gets tossed into a spreadsheet somewhere. Maybe you think about it again two to three quarters later, you know? So the part I think is so interesting is how you captured all of the video—you said, 'I'm going to think very intentionally about how to transfer this empathy.' And ultimately it seems like that played a big role in motivating people to actually take action.

Jason Droy 和我当时的经理 Andy Zabolski 支持我真的把那个视频送到了 Uber Everything 整个组织面前。Jason 说：‘大家需要看看这个。’我们的产品经理、工程师、运营同事都需要亲身体会那种痛苦到底是什么。所以我们做的是把那个视频从两个半小时剪到 3 分钟。但即便是那 3 分钟，我们在全员大会上播放时，只留了 1 分钟。那 1 分钟里，纯粹是坐在车里等着、却什么都等不到的那种感受，对吧？你也许参加过公司的全员大会，也许没有，我不确定，但那种会通常内容塞得满满当当。所以，当几百个人一起坐在会上，尴尬地等着、看着手机 App、却什么都没发生——那个时刻真的很有力量。它也真正催生了后来的 Uber 内部自测（dogfooding）项目。当时，就像我说的，条件很简陋。我是和一群人一起做这件事，后来拉了更多工程师加入。但后来它变成了整个 Uber 内部的正式项目，鼓励大家基本都这么干。我们有研究员——真正的研究员——来规划这些操作，所以它得以规模化，并且把洞察都沉淀下来。所以那种需求——尤其是像 Uber Eats 这样有人类真实参与的服务——我觉得对任何做产品的人来说，都存在一种真实的需求：把自己放在用户的立场上。而且，你知道，这——

Jason Droy and my manager at the time, Andy Zabolski, supported me to actually get that video in front of the entire Uber Everything organization. Jason was like, 'People need to see this.' Our PMs, our engineers, our ops people need to really understand, firsthand, what that pain is. So what we did is we cut a version of that video down from two and a half hours to three minutes. But even that three minutes—we showed it at all hands, and we just left a minute. We left a minute in that presentation for just the feeling of sitting in a car waiting for nothing to happen, right? You've probably been in an all hands of a company—or maybe you haven't, I'm not sure—but they tend to be packed with content. So the power of having hundreds of people in a meeting where we're all just sitting there awkwardly, waiting, looking at the app, nothing going on—it was such a powerful moment. And it actually spearheaded what became, I think, the Uber dogfooding program. At the time, like I said, it was pretty scrappy. I was sort of doing this with a bunch of people. I got more engineers to do it with me, and so on. But it actually became a program across all of Ubers, where people were encouraged to basically do it. We had researchers—real researchers—actually planning the operations of all of this, so it scaled, and we captured the insights and all that stuff. So the appetite—especially for a human service like Uber Eats, where there are real human beings going through the motions of doing something—there's a real appetite, I think, for anyone building products to actually put themselves in the shoes of people. And, you know, it's—

这真的很难做到。

It's so hard to do.

我觉得，如果你至少没有尝过研究是什么滋味——或者至少掌握一点相关技能，再加上一点勇气——就是自己亲自去做，并且知道自己会犯错，而那也没关系。

I feel like if you don't at least have that taste of what research feels like—or at least a little bit of the skill set, and maybe the courage—to just go out and do it yourself, and know that you're going to fail, and that's okay.

听了这些故事，你网站上有一句话我存了下来，现在更说得通了：你说你喜欢在公司的拐点加入。你知道，这时候需要做不一样的事，或者下一个大胆的赌注，或者改变运营方式。所以我很好奇——鉴于你的背景，你横跨过那么多不同的公司，那么多高影响力的职位——有没有另一个例子可以聊聊，让大家了解这种时刻在实际中是什么样的，以及你作为设计师是如何表现的？

Listening to these stories, there's a line on your website that I saved that makes a little more sense now, where you talk about how you like to join companies at inflection points. And you know, there's this need to do things differently, or to take a bold bet, or to change the way you operate. So I'm curious—given your background, you've spanned so many different companies, so many high-impact roles—is there another example we could talk about, just to give people an idea of what this moment looks like in practice, and then how you show up as a designer?

当然可以。是的，是的，是的。我是说，也许我可以讲讲 Shopify。我当时负责领导在线结账业务。如果你去——我忘了，大概是 shopify.com/checkout 之类——他们有个营销页面，上面用大号粗体字写着：‘业界转化率最高的结账流程。’所以想象一下，你接了一份工作，本质上是要改进一个本就该是行业标准、而且已经很优秀的东西。好，这是第一个挑战。再想象一下，Shopify 的 CEO Tobi Lütke 对你整个团队说：‘我们需要重新设计结账流程，把它做得更好。’

Sure. Yeah. Yeah. Yeah. I mean, maybe I'll talk about Shopify. I was responsible for leading online checkout. If you go to—I forget, it's like shopify.com/checkout or something—they have this marketing page, and in big, bold, large text, it says, 'The highest-converting checkout in the industry.' So imagine taking a job where your job is to basically improve what is supposed to be the industry standard—it's already great. Okay, so that's challenge number one. And then imagine that the CEO of Shopify, Tobi Lütke, is telling your entire team, 'We need to redesign checkout, make it better.'

压力山大啊。

No pressure.

压力山大，而且还没有方向。它就是必须变得更好，因为这就是他的——顺便说一句，我很欣赏他这一点——这就是他的 DNA。他总想着颠覆重建，质疑现状，让事情变得更好。我得说，刚加入公司时，这让我很意外。我加入的时机，可能阴差阳错地正好。在我在的那段时间里，我们推出了和 OpenAI 的合作，所以我们几乎是世界上第一批在聊天界面里提供智能体式电商体验的公司。我们还和 Coinbase 一起推出了加密货币——那是 Shopify 第一次在全球范围内支持加密货币。还有结账流程的重新设计。而这只是我入职的头六个月。我觉得自己从来没有在这么短的时间里，在适应期内，交付过这么多东西。说实话，我爱死这种感觉了。

No pressure. And also no direction. It just has to be better, because that is—and I love this about him, by the way—that is his DNA. He's always looking to, you know, tear things down and build them up again, question things, make things better. I would say that was a surprise to me when I joined the company. I joined at, maybe by accident, the perfect time. In the time I was there, we launched the partnership with OpenAI, so we were one of the first companies in the world to basically have that agentic e-commerce experience within the chat interface. We launched cryptocurrency with Coinbase—the first time Shopify had crypto globally. And the checkout redesign. And this was my first six months at the company. I don't think I've ever shipped that much in such a short time during a ramp-up period. And honestly, I loved it.

那种感觉又回来了，就像我在一家公司早期感受到的那种能量——那家公司技术上已经上市、在股票市场交易，但实际运作起来就像一家精干的创业公司。我在大家谈论 OpenAI 之前就做过一些与 AI 沾边的工作，比如基于机器学习的产品工作。但我觉得 Shopify 才是我第一次真正进入 Cursor 和那种 vibe coding 的世界的地方。再说一下时间背景：这是一年半之前的事，简直是远古历史了。当时 Shopify 几乎没有人使用 AI，一个都没有。我属于一小批设计师，是最早试用 Figma MCP 连接 Cursor 的人。对我们来说，这既像魔法，又极其痛苦。我记得当时我和一群人，我们在 Slack 上聊，大家要花几个小时才能把 MCP 连上，因为一年半前的技术就那样。它很笨拙，而且……嗯，反正你需要知道很多门道，你不能直接问 Claude，因为它当时对这些问题还没有答案。所以我们都是自己摸索。加入这个小群体真的很有意思——我们是在这股热潮到来之前的探路人，就在想：这是什么东西？它有用吗，还是只是一阵潮流？我们能用它吗？它会变得有趣吗？然后我想分享一个我的顿悟时刻。Shopify 每年都会办一次大型聚会。我记不清具体叫什么了，但很大一部分——大概 80% 的时间——是一场全公司范围的黑客马拉松。所有人都会参加，不只是设计或工程，所有人，无论你是什么角色。我当时在线上业务，但我喜欢实体设备。他们有一个非常酷的小车库——真的是车库，里面有 POS 终端硬件、NFC 贴纸，如果你想动手摆弄，尽管去。于是我拉了一个向我汇报的设计师，说：‘嘿，你想不想做这个项目？我有个想法，做一个能与 NFC 贴纸联动的 POS 方案。’然后——我说最多八小时——我们从一个想法做到了一个完全可用的东西。给你描绘一下：想象一个带 NFC 芯片的贴纸。你是商家，你不需要静态价签或小牌子，基本上只需要把这些 NFC 贴纸放在毛衣、裤子或任何你要卖的东西上。顾客拿起 Shop App，不需要解锁手机，只需把手机靠近贴纸，就会收到一个‘发现商品’的通知。点一下，它就会说：‘哦，这是一件某某毛衣，90% 羊毛，这是价格、评论等等。’然后点一下就能购买。购买信息会发送到店里的 POS 系统，顾客基本上就自动付款结账了。不用排队，不用跟任何人说话。收银员也知道你不是偷东西，因为系统里有记录，你直接走出店门。所以我们真的基本上做出了这个原型，对吧？就是我刚才展示的那种体验。至于 NFC 连接部分——我们借了一位工程师大约 30 分钟，说：‘嘿，你能帮我们解决卡住的那个问题吗？’因为现在可能 Claude Code 一下子就能搞定，但当时我们需要帮助。

Again, it kind of brought that same energy I had in those early days of a company that—yeah, technically it's public and on the stock market, but it really operated like a scrappy startup. You know, I had worked in AI-adjacent stuff before we all talked about OpenAI—ML-based product work and all that. But I think Shopify was really my first foray into Cursor and that sort of vibe-coding world at the time. And again, to frame this, it's a year and a half ago—this is ancient history. Almost nobody at Shopify was using AI, period. Nobody. I was part of a small cohort of designers that were early testers of Figma MCP connecting to Cursor, and it was magic for all of us, but it was also insanely painful. I remember it took me and a bunch of people—we had the Slack going—it would take people hours to get MCP to connect, because that's where the tech was a year and a half ago. It was just clunky, and just not very—yeah, there were all these things you had to know, and you couldn't just ask Claude because it didn't have the answers to that stuff yet. So we were just figuring it out. And it was really fun to be part of this little crew of folks who were just before the Kool-Aid, like: What is this thing? Is it useful or is this just a fad? Can we use this? Is it going to be interesting? And then maybe I'll share my light-bulb moment with this. So Shopify has this big meetup once a year. I forget exactly what it's called, but a big part of it—maybe 80% of the time—is a massive company-wide hackathon. Everybody, not just design, not just engineering—everybody, whatever your role is. I was in the online space. I love physical stuff. They had this really cool little garage—literally a garage with point-of-sale hardware, little NFC stickers, and if you wanted to tinker, go for it. So I grabbed one of the designers that reported to me and said, "Hey, you want to work on this project? I have this idea for a POS thing that can talk to an NFC sticker." And in—I'm going to say eight hours tops—we went from an idea to a fully working thing. To give you a picture: imagine a sticker with an NFC chip on it. You're a merchant, and instead of having static price tags or little signs, you basically just have these NFC stickers that you can put all over your sweaters and pants or whatever you sell. Then a customer can take the Shop app and just hover—they don't have to unlock their phone—just hover near that sticker, and you get a little notification like "product detected" or whatever. Tap on that. It says, "Oh, this is a blah-blah sweater, made of 90% wool, here's the price and reviews, all that stuff." And you can just tap to buy. It sends the purchase over to the POS in store, so the person basically automatically paid for it—checked out. They don't have to get in line, don't have to talk to anybody. The cash register person knows you didn't steal because it's in the POS, and you can just walk out of the store. So literally, we basically built this prototype, right? The experience I just showed. And the NFC connection—again, we borrowed an engineer for about 30 minutes to say, "Hey, can you help us with that thing we were stuck on?" Because probably today Claude Code could do this in a heartbeat. But we needed help.

那时候你得靠自己努力去争取才行。

You kind of had to earn it back then.

是的。

Yeah.

我想说，所谓‘当时’其实只是一年半之前，这太不可思议了。你的职业生涯非常丰富，但在最近这么短的时间里，却发生了这么多变化。也许我们可以借此机会回顾一下。当你思考自己早年关于设计领导力的理念，以及你在 Uber Eats 甚至 Google 的经历是如何塑造这些理念的——再看看你今天怎么想、怎么做——我很好奇，最大的差异在哪里？

I say back then—it's only a year and a half ago, which is insane. You have this really storied career arc, and yet so much has changed in such a tiny window here toward the tail end. So maybe we could use that as an opportunity to reflect a little bit. When you think about your own early philosophies around design leadership, and how those Uber Eats and even Google experiences were shaping it—and then what you think and how you operate today—I'm kind of curious where the biggest deltas exist.

我觉得你肯定看到了很多不同的说法，但本质上就是组织结构的扁平化，对吧？管理层级变少了。这本身就意味着——如果你是一家公司的管理者——会有几点变化。第一，你管理的下属可能会比以前多不少。我们把这叫 wingspan，也就是管理跨度，对吧？每个经理、高级经理、总监的跨度都变大了。另一个随之而来的变化是：我看到对‘队员兼教练’这种角色的需求更多了。有人来找我，也有我私下教练的人，他们有时候在找下一份工作——我自己也兼职做一些这类教练工作。我看到越来越多的雇主，尤其是上市前的公司，在找这种‘队员兼教练’型的人。这种角色可能有很多定义方式，但对我来说，它通常是管理一个小团队——在我心里大概就是 Shopify 那种规模，两三个、五六个，差不多这个范围。你管理团队，但同时你也和团队一起做东西。你不是待在象牙塔里光做绩效评估和协调会议，而让一个个 IC（个人贡献者）做他们自己的事。你们都是同一个团队的一部分，你有双重责任：确保团队开心、健康、发挥出最佳水平；同时，对于那些特别适合你强项的具体事情，你也要深入其中。以我为例，我最近对一些事情充满热情，而我会发现，要么别人请我做，要么我自己主动投入时间——其中一件就是你提到的‘从零到一’的事情。如果我身处一个产品团队，而每个人都在想非常渐进式的东西——尤其是现在——我不会只写一条 Slack 或邮件说：‘嘿，各位，我有个想法。我要去用 Claude Code 花 20 分钟，写一个大家可以回来看的 vibe coding 原型。’因为在我工作过的每一家公司，原型都能赢得争论，对吧？原型能让人兴奋起来。

I think you're definitely seeing a lot of different ways people have framed it, but essentially it's a flattening of a lot of orgs, right? Fewer layers of management. That in itself means—if you're a manager at a company—a few things. One, you probably have a lot more reports than you're used to. We call that wingspan, right? An increased wingspan for each manager, senior manager, director. The other thing that has changed is along with that: I'm seeing more of a player-coach need. Folks have reached out to me, folks I coach personally who are looking for their next jobs at times—I've done a little bit of that on the side as well. I'm seeing more employers, especially pre-IPO companies, that are looking for this player-coach type. There are probably a lot of ways you define it, but for me it often looks like managing a smaller team—in my mind, a Shopify-size thing of like two, three, five, six people in that range. You're managing the team, but you're also building things together with the team. You're not in this ivory tower just doing performance reviews and alignment meetings while the ICs do what ICs do. You're all part of the same team, and you have this dual responsibility of making sure the team is happy and healthy and doing their best work, and for particular things that are most suited to your strengths, you're also diving in. For me as an example, there are a few things I'm passionate about these days, and I find I'm either asked to or I spend my own time on—one of which you talked about—is zero-to-one stuff. If I'm on a product team and everyone's thinking really incrementally, especially these days, I'm not going to write a Slack or email saying, "Hey guys, I have an idea. I'm going to go to Claude Code, spend 20 minutes, and vibe-code something that people can come back and see." Because in every company I've ever worked at, prototypes win arguments, right? Prototypes get people psyched.

我们正在远离幻灯片，远离 PRD，转向人们真正能感受到、能产生反应的东西——你的情绪会直接变成“我想要这个”。我想要把这个带给我们的客户。这是我花大量时间投入的领域。现在我在全职工作中的每一分钟空闲，都在想办法提升自己的 AI 技能，并把它们应用到手上的工作。比如在支付领域，我会 vibe code 一个类似新数字钱包的想法，只是为了向团队展示不一样的思考方式。我认为，这些工具的可及性、它们的能力，以及人们上手的速度，意味着管理者想顺手做点东西，不再需要专门腾出几天时间在 Figma 里搭建组件、用 auto layout 创建界面、再做可点击原型。你可以用百分之一的速度完成。所以对那些好奇又有动手能力的管理者来说，他们可以在每周 40 小时内做更多事。这也会帮到像我这样不想坐在 9 小时会议里、只做人员管理的管理者。当然，那种管理职责在世界上绝对有一席之地，也非常重要。但我热爱解决问题，热爱贴近问题本身，去看看用户、和他们聊天。所以看到这条路重新成为一群站在两个世界边界上的人更可行的职业路径，我真的非常兴奋。

Moving away from slide decks, moving away from PRDs, moving into something people can actually feel and react to—where your emotion is literally “I want that.” I want to ship that to our customers. That’s the area I spend a lot of time on. Any minute I have right now at my full-time job, I’m trying to find ways to push my own AI skills and apply them to the work at hand. For payments, for example, I’ll vibe code an idea for a new digital wallet, just to show teams different ways of thinking. I think the access to these tools, the power of these tools, and the speed at which people can pick them up means that for a manager to build something on the side no longer requires setting aside a couple of days to build out components in Figma, create something in auto layout, and string up a prototype. You can do that at one hundredth of the speed. So for managers who are curious and crafty, they can do a lot more with their 40 hours. That’s also helped managers like myself who don’t want to sit in nine hours of meetings and just manage people. There’s absolutely a place for that in the world, and it’s super important. But I love problem solving. I love being close to the problems, seeing the users, and talking to them. So it’s been really exciting to know that this is becoming a more viable career path again for people on the fence between those two worlds.

我之前没从这个角度想过。事后看这很明显，但你确实会听到“原型就是新的 PRD”这种说法，或类似的变体。你说得对——不只是我们有了更有效沟通的能力，而是以更高保真度表达想法的机会成本被大幅拉低了，低到作为领导者，你不再觉得自己被从真正需要推动的大目标上拽走。

I hadn’t thought about it that way. It’s very obvious in retrospect, but you hear “the prototype is the new PRD” or some version of that statement. You’re right—it’s not just that we have the ability to communicate more effectively. The opportunity cost of expressing your idea at higher fidelity is driven down so significantly that maybe as a leader you don’t feel like you’re being pulled away from the bigger things you’re responsible for moving.

我曾在会议间隙倒杯咖啡的 10 分钟里，用 Figma Make 或 Claude Code 做过原型。有个例子——我的网站上也有——是我在 Shopify 时的一个经历。当时我们在探索在线结账的重新设计。实际设计还处于进行中，所以仍然保密，但背后用到的技术有一些剧透。想法是做一个更有表现力的在线结账页面。当时有一些假设：商家必须自己动手，才能让结账流程更美观、更有品牌感。他们可能需要上传高分辨率的 SVG 版 Logo，或者自定义颜色。但我们在 Shopify 的调研中知道一件事：商家，尤其是中小商家，要么没时间，要么不太懂技术，所以很可能根本不会自己去做这些。所以到了 2025、2026 年，这类时刻就必须亮起 AI 的灯泡：等等，有些事情人们不愿意做，是因为没时间或不会做——AI 能不能做？我当时提出的就是这个：AI 能不能看看 X 品牌的在线商店，抓取哪怕一张低分辨率的公司 Logo JPEG，把它转成矢量图，这样我们就能替换背景色？能不能再看一下他们网站的 CSS，挑出可能的主色、辅色和强调色，用在他们的结账页面里？

I’ve done prototypes in Figma Make or Claude Code in 10 minutes between meetings while grabbing a cup of coffee. There’s an example—it’s on my website too—from my time at Shopify. We were exploring the online checkout redesign. The actual design is still confidential because it’s still a work in progress, but there’s a tease around the technology behind it. The idea was to make a more expressive version of online checkout. There were certain assumptions about what that would look like: merchants would actually have to do something to make their checkout more beautiful and more branded. They might have to upload high-fidelity SVGs of their logo, or customize some colors. But one thing we knew from research at Shopify is that merchants—especially SMBs—either don’t have the time or aren’t super tech-savvy, so they just might not go through the motions of doing that stuff on their own. So in 2025, 2026, the AI light bulb has to go on in those moments: wait, there’s something people aren’t willing to do because they don’t have time or know-how—could AI do this? That was my pitch: could AI look at an online store for brand X, pull even a low-res JPEG of the company’s logo, turn it into a vector, and then swap out background colors? Could we look at their website CSS and pick out possible primary and secondary accent colors for their checkout?

当时我的设计负责人说：“商家看起来不会那么做，我也不知道这在技术上能不能实现。”而且，在 2025 年之前，设计师会怎么做？他们会去问工程师：“这能做到吗？”没错——事实上我也真去问了。工程师的反应是：“不确定 / 没时间回答你的问题。”其实不用问我也知道他们会这么回答。所以我说：“好吧，这听起来很疯狂，大概率也不会成功，但管他呢。”于是我当时的做法是打开 Cursor，vibe code 了一个内部工具，来做我刚才描述的事情。那真的是史上最丑的东西——就是一个表单，你输入类似 outdoorvoices.com 的品牌名，它会把网站上所有可能是品牌色的颜色变体都抓出来。它还会智能地判断：“这个颜色用在了所有按钮上，很可能是主色候选；这个用在了其他元素上，可能是点缀色或次要色。”

My design leader at the time said, “It doesn’t seem like merchants are going to do that. I don’t know if we can do that technically.” And again, pre-2025, what would a designer do? They’d ask an engineer, “Can this be done?” Totally—which, by the way, I did. The engineers are like, “Not sure / don’t have time to answer your question.” Which I could have told you was their answer even. So I said, “Okay, this is crazy. This is probably going to be fruitless, but what the heck.” So I went into Cursor and vibe coded an internal tool to do what I just described. It was literally the ugliest thing ever—just a form where you type in a brand like outdoorvoices.com. It would pull all the possible color variants from the site that could be brand colors. It was intelligent enough to say, “This was used on all the buttons, so it’s probably a primary color candidate. This was used in some other stuff, so maybe it’s a spot or secondary color.”

然后是 Logo 的部分——这就是我们喜欢 AI 的地方，对吧？我不需要去查 API 文档。我直接说：我想做的是，找到 Logo，看看页头里有什么，或者任何可能的地方，试着找到一个可能带品牌名的 Logo，下载它，再转成 SVG。然后我就开始祈祷，心想：“也许就成了。”后来我找到了一个开源的 API 能实现这个，于是把整个流程做了出来。这件事刚好发生在那个会议上、设计侧的 VP 们告诉我说做不到之后。我说：“嘿，你们看看。我不是工程师，但我刚刚做出了一个 SVG 生成能力，它可以换成任意颜色、可以透明、带上所有品牌色，而商家完全不需要动手。我们可以直接把新的在线结账方案拿给商家看，问他们：‘你喜欢这样吗？不喜欢的话，这里还有别的颜色可以调。’”

And then the logo part—and we love this about AI, right? I don’t have to look for an API. I just say, “Here’s what I want to do. I want to find the logo, look in the header or wherever we can, try to find a logo that maybe has the brand name, download it, and turn it into SVG.” And then I prayed—I’m like, “Maybe this is the thing.” So I found an open-source API that does that, and I built the whole thing. This was right after the meeting where I was told by VPs on the design side that it couldn’t be done. I was like, “Hey, check it out. I’m a non-engineer, but I just built the ability to create an SVG that can be recolored, made transparent, and carry all the brand colors. The merchant never has to lift a finger. We could present a new online checkout to a merchant and say, ‘Do you like this? If not, here are some other colors you can tweak.’”

所以这就是我说的那个世界——我也不知道自己为什么现在这么激动。我觉得设计师真的拥有这种超能力，一种还不知道自己拥有的超能力。我甚至不知道 Claude 或 Cursor 能不能在那件事上帮我，结果它们真的做到了。我记得把这个例子分享给另一位设计负责人，他问：“你怎么知道它会成功？”我说：“其实我不知道，我只是不停地往 AI 黑箱里丢东西。”而且说实话，我已经 vibe coding 两年了，现在做的这些事，在一两年之前要难得多——当时感觉根本不可能。这也是另一件事。

So that’s that world again—I don’t know why I’m so excited right now. I think designers literally have this superpower, an unknown superpower. I didn’t even know Claude would be able to help me there, or Cursor would, and it did. I remember sharing this example with another design leader, and he asked, “How did you know it was going to work?” And I said, “I actually didn’t. I’m literally just throwing stuff into the AI black box.” And I’ll tell you, having been vibe coding for two years now, the stuff I’m doing now was so much harder a year or two ago—it felt impossible. And that’s the other thing too.

这一刻有趣又诡异的地方在于，你甚至不能声称自己是任何这方面的专家，因为下个月它就会——

Like, that's such a fun and weird thing about this moment: you can't even claim to be an expert on any of this stuff, because literally next month it's going to be –

完全不同。

Totally different.

更聪明 20%。它能做到你今天做不到的事。所以我一直想起 Cleo 的一句话，他谈到早期 Facebook 那种“不可避免的光环”：当事情已经真实发生、技术也已经接近成熟时，你就很难再说“不，我们不应该这样做”。而突然间，设计师们有能力把这种“能力的光环”带到台面上。所以当你把它做出来时，就像：“好吧，糟糕，我们再也不能无视它了。”你知道，它确实管用。就像，他把它做出来了。

20% smarter. It's going to be able to do things you couldn't do today. So, there's a Cleo quote that I keep coming back to, where he talked about the aura of inevitability at early Facebook, and how when things are real and already kind of close to having the technology there, it's just so much harder to say no, we shouldn't do that. And all of a sudden, designers have the ability to bring the aura of ability to the table. And so when you gave that, it's like, all right, shoot, we can't ignore this anymore. You know, it works. Like, he made this.

对，对。我要补充一点。我记得那段经历：当时我在 AI 领域算是早期又好奇的人。我的设计团队里，大部分人从来没有碰过 vibe coding。我们只有一个设计师在用 Midjourney——那已经是我团队里 AI 实践的全部了。我记得把这个例子、这个故事拿给和我一起做在线结账的设计师看，她的反应就是：“天哪，我真不敢相信你——一个设计师、一个管理者——居然能做出这个，更别说是在 15 分钟内。而且，你居然有胆量让 VP 说‘不’，然后又说‘其实是可以的’，对吧？”所以我想，回到管理者的角色：就是要勇于推动这些工具，并且亲自试用它们，而不是直接把工具扔给团队，然后期望它们能管用。

Yeah, yeah. I'll add one thing to that. I'm remembering this part of my experience. At that time, I was one of the early and curious people in the AI space. Most of my design team that reported to me had not dabbled in vibe coding at all. We had one designer who was using Midjourney – that was the extent of AI practice on my team. I remember showing that example, that story, to the designer who was working on online checkout with me, and she was just like, 'Holy expletive, I cannot believe that you – a designer, a manager – were able to make that, let alone in 15 minutes. And, by the way, having the audacity to have a VP say no and then being like, actually yes, it's possible, right?' So I think, again, going back to the role of a manager: it's about having the courage to push on these tools and dogfood them, instead of just sending them over to your team and hoping they work.

我想从一个稍微不同的角度来看看当今世界的状况，因为你最近以领导者的身份经历了求职市场。我想知道这次有没有什么让你明显感觉不一样的地方，或者你看到的其他信号，能反映出这个格局正在发生的某些变化。

I want to look at the state of the world today from a slightly different angle, because you recently navigated the job market as a leader. I'm wondering if there was anything that felt clearly different this time around, or other signals you're seeing that point to some of the ways the landscape is shifting.

完全同意。我是说，我们刚才提到的“球员兼教练”这种类型，现在可能更流行。我想不出任何一个我应聘过的领导职位——我做过好几个总监职位，都是非常高级的岗位——有哪一个不是期望你亲力亲为的。句号。完完全全。招聘描述里并不清楚这到底意味着什么，得视具体情况而定。但我给你举个例子。当我在和一家早期创业公司谈的时候，他们居然让我和一个东欧工程师通电话，然后看着我使用 Figma，以确保我知道怎么用 auto layout。

Totally. I mean, we talked a little bit about this sort of 'player-coach' being maybe more in vogue right now. I can't think of a single leadership role that I've applied to – and I've been a director for a number of roles, very senior positions – where they weren't expecting you to be hands-on. Period. Full stop. What that actually means isn't clear in job descriptions; it varies case by case. But I'll give you an example. When I was talking to an earlier-stage startup, they literally put me on a call with an engineer in Eastern Europe and watched me use Figma to make sure I knew how to use auto layout.

不会吧。

No way.

对，对，对。说真的，他们给了我一个设计练习——说真的就是为了测试我是否会使用 Figma，而我当然会。耶，我通过了。不过，那真是我职业生涯头一回。同样地，我觉得 AI 工具未必会写进职位描述，但人们肯定注意到了我对构建的热情。他们注意到，在离职期间，我不仅仅是找工作，还和妻子从零开始创建了一家完整的公司，名叫 Matchmaker。我觉得，看到一个人不仅是个 AI 爱好者，而且真的能戴上产品经理、工程师、设计师，甚至企业家的帽子，去做端到端的产品开发——还要处理在特拉华州注册公司、付各种费用、和合作伙伴制定合同之类的事情。我一直这么说，但我确实认为，最优秀的设计领导者往往是那些花时间当过创业者，或在早期公司工作过的人，还是因为那种主人翁意识。嗯，不过我也确实看到，越来越多的人开始把以前可能被称为“个人项目”的东西看作有价值，对吧？我管这个叫个人项目，因为没人雇我去做 Matchmaker，但它是我和两位联合创始人从零开始创建的。它展现了主动性和反复迭代的韧性，展现了产品思维，展现了技术敏锐度，所有这些。所以，我确实看到更多公司在看你工作之外做的东西时，不把它当作额外加分，而是可能把它当作获得工作的原因，甚至超过传统工作经验。

Yeah, yeah, yeah. Literally, they gave me a design exercise – it was literally a test to see if I know how to use Figma, which, by the way, I did. Yay. I got it; I passed. But yeah, that was a career first. Likewise, I don't think AI tools are necessarily something people are putting on job descriptions, but people have definitely noticed my passion for building. They've noticed that, between jobs, I wasn't only job searching, but also built an entire company from scratch, called Matchmaker, with my wife. I think seeing that somebody is not only kind of a tinkerer of AI, but literally can do end-to-end product development – wearing the PM hat, the engineering hat, the design hat, and also the entrepreneur hat – all the other stuff like incorporating the business in Delaware, paying the fees, figuring out contracts with partners, all of that. I've been saying this for a minute, but I do think some of the best design leaders are people who have spent time being entrepreneurs or working at early-stage companies, again because of that ownership thing. Um, but I do think I'm seeing more people find what previously might have been called 'personal projects' as valuable, right? I call this a personal project because nobody hired me to build Matchmaker, but it was something that me and two co-founders created from scratch. It shows initiative and the resiliency of iterating through it. It shows product thinking, technical savviness – all of this stuff. And so I'm definitely seeing more companies looking at things you built outside of work, not as extra credit, but as possible reasons to get jobs, in addition to or even over traditional work experience.

你知道吗，有意思的是，我收到很多设计领导者的邮件和 Twitter 私信，请求上节目。这是个不同的视角，但我对你感兴趣也是因为 Matchmaker，原因完全相同。“哦，老兄，你不仅有资历，还真的是亲自动手。”没有什么比从 0 到 1 打造一个全新产品和公司更能体现深度了，对吧？这非常能说明你是谁。所以我想知道，我们能不能稍微了解一下幕后情况。我很好奇那个过程是什么样的。也许有没有具体的经验教训，或者你必须克服的障碍，你觉得听众中那些受到启发、想要做类似旅程，或者只是开始更多构建的人，会从中受益？

You know, it's funny – I have a lot of design leaders in my emails and Twitter DMs asking to come on the show. It's a different lens, but what I thought was so interesting about you was Matchmaker too, for the exact same reason. It's like, oh man, not only do you have the pedigree, but talk about hands in the clay. There's no level of depth you can get beyond building a totally new product and company, going 0 to 1 on the whole thing, right? That says so much about who you are. So I'm wondering if we can maybe get a little bit behind the scenes. I'm curious what that process looked like. Maybe were there specific lessons learned or hurdles you had to overcome that you think people listening – who are inspired to do a similar journey or to just start building more – would benefit from seeing?

我觉得我们很多人脑子里都有很多个人项目的想法——我们生活中想要解决的问题。但这次不一样。这是我的妻子 Melanie——她做市场营销，整个职业生涯都在做营销。后来她产生了创业的冲动，成立了一家专门做所谓“品牌合作”的机构。你知道，就像：“我是一个户外用品品牌，我是一个奢侈品牌，对吧？我想让更多人注意到我，建立一定的品牌资产，让人们购买我的产品。”品牌合作就是，好吧，我们去找其他志同道合的组织、品牌或创作者，想办法合作，实现双赢，互相借力对方的受众，等等。于是 Melanie 和她的搭档注意到，他们加入了全世界各种关于营销的 Slack 社区。然后人们会说：“嘿，我们几个月后要参加威尼斯双年展，你知道吗。”

I think many of us probably have a lot of personal project ideas in our head – problems we have in our own lives that we want to solve. This was different. This was my wife Melanie – she's in marketing, she's been marketing her whole career. Then she got the entrepreneurship bug and started an agency specifically focused on what's called brand partnerships. You know, 'I am an outdoor goods brand. I'm a luxury brand, right? I want to get more people to notice me, build up a certain brand equity, and get people to buy my products.' Partnerships is like, okay, let's find other like-minded organizations, brands, or creators, and find ways to work together so we can have a win-win situation, leverage each other's audiences, and so on. So what Melanie and her partner noticed was that they were in all these Slack communities throughout the world that are all about marketing. And people were like, 'Hey, we're doing the Venice Biennale, you know, in a couple of months.'

“我们在找那种对户外感兴趣的网红，对吧？”所以 Slack 里会不断冒出这种小小的基于 Slack 的需求提案，最后都石沉大海。Matchmaker 其实就是想用软件解决这个问题。说回流程，我有一个内部的产品经理兼研究员联合创始人，也就是我的妻子。她是我的智慧来源，告诉我用户是谁、他们可能怎么行事、在找什么。但说实话，我觉得我们现在已经在第三次构建这个产品了。

"We're looking for influencers that are interested in the outdoors, right?" So there are constantly these little mini Slack-based RFPs that are just falling into the ether. Matchmaker was really an attempt to solve that problem in software. Now in terms of the process, I had my in-house PM/researcher-founder, which is my wife. She was my source of wisdom on who the users are, how they might behave, things they're looking for. But to be honest, I think we are now in our third build of the product.

简单描述一下：我们大概一年半前重新开始思考这件事。当时 AI 工具还算可以——还行。我不觉得我一定得用 Cursor 或 Claude Code 做一个完整的定制系统。当时有很多现成产品。其中一个叫 Softr，当时拼写里没有 e。它有点像 Framer，但内置功能更多。我当时想：“哦，这个可以做 MVP，试试吧。”

Just to paint the picture, we started thinking about this again probably a year and a half ago. At that time, AI tools were okay—they were fine. I didn't think I necessarily had to build a whole custom thing with Cursor or Claude Code. There were a bunch of off-the-shelf products. The one was called Softr at the time, with no 'e'. It's almost like Framer with a little bit more built-in functionality. I thought, "Oh, this could work for an MVP, let's try it."

到最后它丑得要命。作为设计师，我当时想：“我受不了，我受不了。”几个月后，我们看到 Cursor 和 Lovable 的能力开始崭露头角。Lovable 真的让我眼前一亮。我想说两件事。第一，作为一个学过计算机科学、人生中有段时间学过六种编程语言的人，我很想彻底投入极客模式，从零开始写代码。但我知道我很有可能会回去做全职工作。我有一个梦想：我可以构建一个产品，交给我的妻子和她的合伙人——他们不一定懂技术——他们可以在 Lovable 里运营这个业务，而我不用实际参与。这就是我的梦想。所以我决定选择 Lovable，尽管它不能满足我内心极客的那部分，但我想体验一下第一次从零到一构建东西的感觉。

And then by the end, it was ugly as hell. As a designer, I was like, "I can't. I can't." After a few months of that, we already saw that the capabilities of Cursor and Lovable started becoming a thing. Lovable actually really caught my eye. I'll say two things. One: as somebody who studied computer science and has learned six different programming languages at some point in my life, I was tempted to go all geek and just code this from scratch. But I knew there was a good chance I was going to go back to a full-time job. I had this dream that I could build a product and hand it to my wife and her partner, who are not necessarily technical, and they could run the business in Lovable without me actually being involved. That was the dream. So I'm going to choose Lovable, even though it doesn't scratch that geek part of me, to see for the first time I built something zero to one.

我非常充分地复刻了自己作为设计师的流程。我的意思是，就是和联合创始人进行讨论，关于用户是谁、我们要解决什么问题。我们会写战略文档、写 PRD，这些都是我们的产物。然后利用这些产物来生成一系列提示词，再喂给 Lovable 来创建第一版。你可以想象，第一版糟糕透顶，而且特别紫。但它只是为了开始打基础。我几乎把 Matchmaker 的第一次迭代看作我所说的“可构建线框”。它能让你实现核心功能。工程师们早就这么做了——这不是新东西。你基本上是在搭建流程的骨架：有一个首页，上面要有最新列表；可能还有一个导航，里面有通知区域，用来显示另一个品牌的聊天。我们有了所有元素。它很丑、不直观、一半功能不能用，但它就在那里。我想这就是 “vibe” 这个词以及 “vibe coding（氛围编码）” 的来源。

I very much replicated my own process as a designer. What I mean by that is literally conversations with my co-founders about who our user is, what problems we're trying to solve. We're writing a strategy document, writing a PRD. We have artifacts for all this stuff, and using that to basically generate a series of prompts, which we then fed into Lovable to create the first version. The first version, as you can imagine, was absolutely terrible and very purple. But it was just to start building the foundations. I almost thought of that first iteration of Matchmaker as what I'd call a build wireframe. It allows you to do the core functionality. Engineers have been doing this for a long time—this is not new. You're basically building the skeleton of the flow. You have a homepage that's going to have recent listings on it. It's going to have maybe a navigation with a notification area if you have a chat from another brand, right? We had all the elements. It was ugly, not intuitive, half of it didn't work, but it was there. And that's, I guess, where the word 'vibe' comes from—vibe coding.

我当时慢慢开始在 Figma 里构建一个设计系统。我知道有些人已经不用 Figma 了，但我发现它对于构建——我们称之为五到八个高保真原型图，展示我想要体验效果的样子——非常有用。我创建了一个非常非常简陋的设计系统，说“设计系统”可能是夸张了。那是一个组件贴纸页：按钮、按钮悬停状态、我的字号层级、调色板，还有几样其他东西。然后我真的就让 Lovable 摄取那些屏幕，然后说：“把全站所有东西都更新成我做的这些新高保真视觉设计。”顺便说一下，我之所以这样做，回到我老派的设计流程，是因为我们不一定知道流程应该是什么样的。

I slowly started to build a design system in Figma at the time. I know some folks aren't using Figma anymore, but I found it very useful for building—let's call it five to eight high-fidelity mockups of what I wanted the experience to look like. I created a very, very humble design system—'design system' is probably an exaggeration. It was a sticker sheet of components: button, button hover state, my type scale, color palette, a few other things. And I literally just had Lovable ingest those screens and said, "Update everything across the site to these new high-fidelity visual designs that I made." And by the way, the reason I did it this way—going back to my old-school design process—is that we didn't necessarily know what the flow should be.

对。

Yeah.

我们几乎是在做一个原型，去感受低保真线框阶段是什么样子。我们拆掉了一些东西。在那个难看的阶段，我们做了几次迭代，确保它在信息架构上感觉没问题。然后我们说：“好了，现在让它看起来超棒。”现在它看起来很棒——我的意思是，我们从此就生活在高保真的世界里了。我们有设计系统，有字号层级。我妻子总是没我也能发布新功能。她真的能创建全新的页面，而且看起来非常符合品牌。它用上了所有该用的东西。而且如今 Lovable 比一年前强太多了，它自己也能做出非常好的设计决策。

We were almost prototyping it to sort of feel what it looks like in that low-fidelity wireframe stage. We tore some stuff down. We had a few iterations in that ugly stage to make sure it felt right in terms of the information architecture. And then we went, "Okay, now let's make this look awesome." Now that it looks awesome—I mean, we are living in the world of high fidelity from now on. We have a design system, we have a type scale. My wife is launching features without me all the time. She can literally create entire new pages and it looks on brand. It uses all the right things. And even today, Lovable is so much better than a year ago that it also makes really good design decisions too.

很好。

That's good.

看到它过去这段时间的进化，我真的印象深刻。但我要说的是，我们到了一个节点。现在它已经完全可用了——更不用说，它还是一家已注册的公司。它能收款，我们有真实用户，有真实列表，但我们仍然处于非常早期。我这样说的意思是——现在大家在社交媒体上经常讨论——因为任何人都能从零开始构建东西，从零开始 vibe coding（氛围编码）。我们全都发现，构建软件不再像以前那么难，但获得受众的难度一如既往。

I've been really impressed even seeing how it's evolved over the past. But I'll just say we kind of got to a certain point. Right now it's fully functional—not to mention, it's an incorporated business. It takes payments, we have real users, we have real listings, but we're still very much in the early stages. And what I mean by that is—people are talking about this a lot on social now—because anyone can build something from scratch, vibe-code something from scratch. We are all finding that building software isn't as hard as it used to be, but growing an audience is as hard as it always has been.

有一个问题我忍不住一直问自己：如果公司主动申请来和你对话，而不是反过来呢？这个问题就是全新的 Dive Talent Network 的基础。而且它正在奏效。现在，我正在帮助我认识的最令人兴奋的初创公司，来雇佣收听这个节目的设计师和建设者。所以，如果你好奇外面有什么机会，或者想加入我的名单，又或者你正在寻找你的下一位设计人才，请访问 dive.club/talent 加入吧。

There's one question that I can't stop asking myself: What if companies applied to talk to you rather than the other way around? That question is the foundation for the all-new Dive Talent Network. And it's working. Right now, I'm helping many of the most exciting startups that I know hire the designers and builders who listen to this show. So if you're curious what might be out there, maybe you want to get on my list, or maybe you're even looking for your next design hire, head to dive.club/talent to join today.

哦，我想请教一下你的经验。你在实际用 AI 进行 vibe coding（氛围编码）方面已经积累了不少经验，不只是内部那些一次性低保真原型，而是真的把一个产品从零做到一，并且经历了不同阶段和反复重建。我想你一定在这个过程中积累了一些经验教训，这些经验塑造了你在与 AI 交互以及最大化利用这项技术时的思考方式。

Oh, I think I want to tap into your experience now, having quite a few reps under your belt in terms of actually vibe-coding—building software with AI. Not only the little one-off low-fidelity prototypes internally, but literally going zero to one on a product and building that through different phases and rebuilding it. And I would imagine that you've picked up some lessons and experiences along the way that have shaped the way you think about interfacing with AI and getting the most out of this technology.

我很好奇，你作为构建者，是否能看清自己的流程发生了哪些明显变化，或者也许是你使用这些工具的方式有所不同，或者任何那些可能一直站在旁观立场的人可以借鉴的东西。

I'm curious if you're able to see clear ways that your own process as a builder has evolved, or maybe just different ways that you approach the tools, or anything that people who maybe have been kind of on the sidelines a little bit could learn from.

嗯，我觉得这里面有很多可以展开的。我是说，正如你提到的，从那之后我确实又做了不少东西。从另一个我还在捣鼓、可能不会上线的 SaaS 工具，到一些实验性质的东西。它有点像在帮设计行业里找工作的人和教练。所以对比一下，Levelable 的第一个版本大概花了我几个月。做到现在这个像样的版本，大概只花了一天。

Yeah, I think a lot to unpack there. I mean, as you mentioned, I've built a bunch of stuff since then. Everything from another SaaS tool that I'm working on that may or may not launch. It's just something I'm experimenting with, almost like helping people who are looking for jobs and coaches in the design industry. So just for comparison, the first version of Levelable probably took me a few months. Getting to a decent version of this took me probably a day.

哇。

Whoa.

对。部分和这些工具的能力有关，但也有一些，正如你说的，是因为我现在更聪明了一点——我知道 AI 是怎么思考和工作的。我也做过一堆实验性的艺术项目。所以，我觉得有几点值得提出来。

Right. Some of it has to do with the power of these tools, but some of it, to your point, is about just being a little smarter now that I know how the AI thinks and works. I've also done a bunch of experimental art stuff. Again, I think there are a few things I'd call out.

我想一是思考如何规划一个项目。我觉得这套技能设计师和产品经理本来就应该一定程度上具备，尤其是资深一些的。那些常规练习仍然有用：非常清楚地定义这个产品的目标是什么，给谁用，必须处理的关键用户旅程是什么。哪怕只是把它过一遍，而且这些都是纯文字，对吧？就像把事情写进文档里。我甚至见过有人用麦克风口述，如果那对你有效的话。但关键是把东西用粗糙、难看的方式先写下来。而我实际上做的是，把我写的那个，姑且称之为策略/产品设计混合体的初稿，丢进 Claude，然后基本会说：“嘿，Claude，我正在做某个东西，就是巴拉巴拉，对吧？”我会把那份满是错别字的丑陋 PRD 直接倒进去，然后说：“给我一系列提示词，以及一个如何最高效地把它做出来的计划。”然后我会说出平台，是 Lovable、Claude Code 还是别的。因为我觉得这些工具现在对自身、对这些平台都非常有自觉，它们真能理解 Lovable 是怎么构建东西的，所以 Claude 真的会吐出一整套计划。它会问你关于计划的问题，指出你思考中的漏洞，这太神奇了。你大概也见过，Claude 的界面真的进化到能直接在对话里问：“嘿，我注意到你想要一个品牌的概念。是什么构成了品牌？你希望我把这个品牌设为公开、私有，还是两者都做？”就像……

I think one is thinking about how you plan for a project. And I think this is a skill set that designers and PMs should already have to some extent, especially if you're on the more senior side. The usual exercises still help: really clearly defining what is the goal of this product, who is it for, what are the critical user journeys that this product is supposed to handle. Even walking through it, and this is all in words, right? Like writing this stuff down in a doc. I've even seen some people dictate using a microphone if that works for you. But just getting it down in raw, ugly ways. And what I've actually been doing is taking that initial, let's call it a strategy/PD kind of hybrid thing that I've written. I will drop that into Claude and basically say, 'Hey Claude, I am building something that's blah blah blah, right?' I'll literally dump that ugly, typo-ridden sort of PRD into it and say, 'Make me a series of prompts and a plan for how to most effectively build this, and then I'll say the platform in Lovable, in Claude Code, in whatever.' Because I mean these tools now are so self-aware of themselves and of these platforms that they can actually understand how Lovable builds stuff, and so they will — Claude will literally spit out an entire plan. It will ask you questions about your plan and identify gaps in your thinking, which is amazing. Like you've probably seen this as well, but the Claude interface has really evolved to literally have inline interactions like, 'Hey, I noticed you wanted a notion of a brand. Like, what constitutes a brand? Do you want me to make this brand public or private or both?' It's like...

所以它就是……它聪明了很多，几乎可以在预规划阶段给你巨大帮助。所以我觉得这是帮我先把第一版做出来的一个方面。

So there's just... it's a lot smarter, and it can almost help you a ton with the pre-planning. So I think that's one area that's been helpful for me to get that first build out the door.

到目前为止对我依然成立的第二件事——我很好奇你是否对此有不同看法——是我普遍发现，当你想在任何平台上一次性把所有东西都建出来，某种程度上会是一场灾难。我更倾向于分块构建。所以有时 Claude 会这么建议，有时我会主动问 Claude。比如：“咱们先谢谢你给的方案，Claude。现在给我一个计划，告诉我怎么制定路线图，或者把这次构建分阶段进行。”

The second thing that so far has proven true to me — I'm curious if you have a different opinion on this — is that I've generally found that when you try to build everything at once with any platform, it's kind of a disaster in a way. I prefer to build things in pieces. And so sometimes Claude will recommend that. Sometimes I will ask Claude about that. Like, 'Let's actually, thank you Claude for your plan. Now walk me through a plan on how to basically create a roadmap or stagger this build out.'

对。所以，举个例子，比如 Matchmaker 这个项目，我们可能会先从几个关键页面开始。设置先不碰，聊天功能也先不碰，对吧？所以我们真的就像跑冲刺一样分阶段构建。每个冲刺可能只有 10 分钟，而不是两周，但你仍然是在按一次一次会话来做，对吧？你在构建、检查，确保自己没有为了做这件事烧掉 1000 个点数。而且你确实是在分块推进。在这个过程中，你和智能体正在建立起对产品和彼此的相互理解，这真的很有意思。它会开始理解你。顺便说一句，我刚才忘了提：所有那些初始上下文、PRD 之类的东西，我觉得像 Claude 这样的一些平台在长期上下文上要好得多。Lovable 以前不是，但 Lovable 里有些地方你可以保存上下文。所以我也做了一件事，就是确保有一个类似项目记忆的地方，对吧？就像你存下来的上下文，告诉它这个东西到底是什么，让它能反复参考并明白：“哦等等，我是在做一个品牌配对服务，不是什么邮件平台，”对吧？就像……

Right. And so you know, as an example, with Matchmaker, maybe we'll start with a few of the key screens. We're not going to mess with settings yet. We're not going to mess with chat yet. Right? So we're literally building out things in phases like you would a sprint. Each sprint might be 10 minutes instead of two weeks, but you're still doing things in sessions, right? Where you're kind of building a thing, checking it, making sure you're not burning through a thousand credits doing it. And you're really doing this in pieces. And as you're doing that, you and the agent are building a mutual understanding of the product and of each other, which is really interesting. It starts to understand you. By the way, something I forgot to mention: all that initial context, that PRD and all that stuff — I think some platforms like Claude are much better at long-term context. Lovable hasn't been, but there are places within Lovable where you can save context. And so that is something I also do: make sure there's a place that's sort of like the project memory, right? Like your stored context for what this thing is, that it can keep referencing to understand, 'Oh wait, I'm building a matchmaking service for brands, not whatever some email platform,' right? Like it's kind of like...

这些上下文似乎真的对这类平台帮助很大。

That context seems to really help these platforms quite a bit.

其实我觉得我有一篇 Substack 文章讲过这个。我都不太记得了，大概写了 12 条关于怎么省 token 的建议，比如让它少说废话、更简洁，对吧？让它去检查自己的工作。呃，视觉方面——我觉得这是目前设计领域一个很大的分歧点。我是说，你得从我已经冰凉的手里把 Figma 抢走。也许迟早会那样，但我就是喜欢在视觉上工作。我是一个“展示而非讲述”的人，我带领团队的方式也是这样。我会把我想到的点子画出来，而不是只用文字。所以当 Claude 里有什么不对的时候，我可以和它来来回回用文字弄 10 分钟，也可以打开 Figma 花 5 分钟画出来，而且那可能更接近我脑子里那个像素级完美的版本。所以我经常会在 AI 构建和 Figma 工具之间同时处理事情，你知道的，它构建出一些东西，我就会说：“哦，这个间距有点不对。我可能要往上移六个像素。”或者如果问题更大，我可能真的会把那个小组件或组件的一部分设计出来，然后说：“不，我要的头像、旁边的品牌名和收藏按钮，看起来应该是这样。”对吧？所以我觉得每个人都有自己的一套工作流，但对我来说，设计工具仍然是一个非常有效的临时补丁，因为 AI 确实会自己做很多事情，它的灵感来自，你知道的，它被训练时看的数百万其他设计师。所以我想这也是最后一点：你知道，我们会开玩笑说 AI 的紫调，但我确实认为现在有很多非设计师在用 AI。

I actually think I have a Substack post on this. I don't even remember. I wrote like 12 points on how to save tokens, but like asking it to be less wordy and more concise, right? Like asking it to sort of check its work. Uh, visuals — I mean, this is where I feel like there's a big divide in design right now. I mean, you'll have to pry Figma from my cold hands. It might happen eventually, but I just love working visually. I'm a show-versus-tell person, and that's even the way I lead teams. I'll sketch out an idea I have instead of just using words. And so when something is off in Claude, I could go back and forth with it in words for 10 minutes, or I could go to Figma and sketch it out in 5 minutes, and it's probably going to be closer to the pixel-perfect version I have in my head. So I will very often kind of dual-task between AI building and Figma tools at the same time. It'll build something and I'll be like, 'Oh, that spacing is a little bit off. Let me just move it up six pixels.' Or if it's something bigger than that, I might literally just design that little component or a part of a component and be like, 'No, this is how I want the avatar to look next to the name of the brand and the favorite button.' Right? And so I think that's just something that everyone's kind of got their own workflows, but for me, the design tools have still been such a helpful stopgap, because AI does seem to do a lot of stuff on its own, inspired by, you know, the millions of other designers it's being trained off of. And so I guess that's the last part of this too: you know, I think we joke about the purpleness of AI, but I do think that you have a lot of non-designers right now.

“他们看着 AI 工具的输出，就会说：‘哦，还挺不错的。’但我们都知道，其中很多看起来非常相似。那是因为它们根本就是在同一个世界的数据上训练出来的。所以我最后想鼓励大家的是：不要等到最后才去想品牌和差异化。一旦你知道了产品是什么、用来做什么，你越早开始定义品牌和观点，就越能把那种灵魂、品牌和观点注入到产品里。这也是我自己的经验：如果你在开头就把想要的视觉语言的小种子定下来，Claude 非常擅长把它延伸开去。有时它可能会走得太远，你会说：‘好了好了，这个我不要。’从那个意义上说，我会在 canvas 里做设计，但偶尔它会根据我最初的品牌语言做出一些东西，让我觉得：‘哦，我其实还挺喜欢的。也许我可以在那边也用上。’它几乎帮助我扩展了自己的思维。我之前跟你聊到 canvas 工具时也在笑，因为我也是一样：我一开始只待在 canvas 里，然后突然就觉得自己会写代码了；接着钟摆又摆动到了另一端，现在我确定自己正落在中间地带——我觉得今天作为设计师，很多策略和技巧就在于知道什么时候该用不同类型的工具。我在这个过程中玩得很开心。”

They're looking at the outputs of AI tools and they're like, 'Oh, that's pretty good.' But a lot of them, as we both know, look very similar. That's because they're literally all being trained on the same world of stuff. The last thing I really try to encourage people is: don't wait until the very end to think about branding and differentiation. Once you know what the product is and what it's for, that's a conversation and definition you should have as early as possible, to start to infuse some of that soul, that brand and point of view, into the product. That's been my experience too: if you have the nugget of what you want the visual language nailed down in the beginning, Claude's really good at extrapolating it. Maybe sometimes it goes a little too far, and you're like, 'Okay, no, no, no, I don't want that.' And in that sense, I'll design it in the canvas, but every once in a while it'll do something based on my initial brand language where I'm like, 'Oh, actually, I kind of like that. Maybe I would do that over here.' It almost helps me scale my own thinking. And I was kind of chuckling with you about the canvas tools too, because same thing—I'm only in a canvas and all of a sudden I'm like, 'Oh my gosh, I can code now.' Then the pendulum swung all the way, and now I'm definitely settling into the middle ground where I feel like a lot of the strategy and skill around being a designer today is knowing when to reach for different types of tools. I'm having a lot of fun being creative in that process.

“在你走之前，我还有两个完全不按套路出牌的假设性问题想抛给你，确保我们把所有话题都聊到了。第一个回到团队建设上。我想让你想象一下——也许就是你现在的角色——你可以在更大的团队里，从零开始搭建一个从 0 到 1 的设计组织。全新的编制，什么都是新的。”

Before I let you go, I have two total left-field hypotheticals I want to toss your way, just to make sure we covered everything. The first goes back to team building. I want you to imagine—maybe it's even in your current role—you get to build a new zero-to-one design org within the broader team, from scratch. New headcount, everything.

“AI 如何影响了你对团队建设、角色定义的思考？甚至可能影响你在这个新团队里优先看重哪些技能或特质？”

How has AI impacted the way you think about team building, role definition, or maybe even the types of skills or traits you want to prioritize on that new team?

“我确实相信，现在每个职能都能用到同样的工具，所以每个人都能做出看起来像产品的东西。我觉得目前我们仍然能比 AI 做得更好的事情是：第一，我们作为设计师的看家本领——你怎么思考‘我要解决什么问题’？我要解决什么用户问题？我要解决什么商业问题？以及什么样的体验最适合那个问题？有些人会笼统地称之为产品思维。对我来说，这向来是高级及以上设计师的硬性标准，而且到了 2026 年，这对设计师来说更是不可妥协的。你不能只是执行，不能只是接需求。我不是说设计师过去就该这样，而是这种做法已经不再有价值了，因为产品经理已经不需要你再去做那件事了。第二点，我们之前也聊到一点，所以我觉着那是第一类技能。另外，轻量级调研绝对是我认为仍然很有价值的特质。我待过的每一家设计公司、每一家公司，研究资源从来都不够。所以我仍然认为，那些有胆量、有勇气想尽办法去做接地气的咖啡店测试，或者用 user testing.com 这类工具，并且知道那是什么样子、知道什么时候有用、然后直接动手的设计师，是非常重要的。第二点，我们稍微聊过，就是观点。有人把它叫作品味，我觉得这有点自命不凡，但我愿意称之为品牌声音、品牌定位。在一堆已经涌现出来、而且看起来几乎一模一样的产品中——因为它们基本上都是同一个智能体或同一段代码生成的——你要怎么实现差异化？尤其是在创业和从 0 到 1 的世界里，这个问题尤其重要。能够在产品设计、一点品牌，甚至某种程度上的视觉设计之间来回切换，这种能力非常重要。”

I really do believe that now, because every function has access to the same tools, everyone can build something that looks like a product. I think the things we can still do better than AI for now are: one, our bread and butter as designers—how do you think about what problem am I trying to solve? What user problem am I trying to solve? What business problem am I trying to solve? And what is the experience that best suits that problem? Some people would just call that more generally product thinking. I think that has always been a very strong criteria for me for senior and above designers, and it would be a non-negotiable for designers in 2026. You can't just be execution. You can't just be order takers. Not that I think designers ever should be, but that is no longer valuable because a product manager doesn't need you to do that anymore. Two—which we talked about a little bit—so I think that's one kind of skill set. And scrappy research is definitely something that is still a valuable trait. Every design firm, every company I've ever worked for has always been under-resourced in research. So I still think designers that have the boldness and courage to find ways to do scrappy café testing, or user testing.com, or whatever it is, and know what that looks like, know when it's useful, and just do it—that's huge. The second one, we talked about this a little bit, is that point of view. Some people call it taste. I find that a little pretentious, but I'll call it brand voice, brand positioning. How do you differentiate in a sea of tools that are already coming out and look exactly identical because they're all being generated by the same agents basically, or code? How do you differentiate, especially in the entrepreneurial world of startups or zero-to-ones where it's most relevant? That ability to swing between product design and a bit of branding, and to some extent visual design as well, is very important.

“你知道，这就是那些令人愉悦的瞬间。这是我在 Matchmaker 里真的很想做到的东西。你知道，当你做产品的时候——我知道你能感受到这种同理心——在做产品时，我会对视觉设计宽容很多，因为我会觉得：它只要能跑就行。让它跑起来已经太难了，我怎么还会去在意每一个小像素呢？但真等它跑起来之后，我又会发现：好吧，你猜怎么着？有什么能阻止一个竞争对手明天用 vibe coding 直接做出一个一模一样的东西呢？”

You know, it's those moments of delight. This is something I really tried to do with Matchmaker. You know, when you build products—I know you can feel this empathy—when you're building products, I have found myself much more forgiving of visual design because I'm like, it just has to work. It's been so hard to just get it to work that I'm not going to worry about every little pixel, right? But actually, once I get it to work, I really do find—well, you know what? What's stopping a competitor from literally vibe coding the same exact thing tomorrow?

“对。”

Yeah.

“他们花一两天就能做出来，对吧？所以难点不在这里。因此，第一是找到客户——真正的增长问题；第二是拥有一个真正差异化的产品，它感觉更高级，和用户有情感连接。这些正是设计世界所负责的事情。所以我觉着这是两个不太明显的技能组合——也许可以说三个：产品思维、调研和品牌——它们都极其重要。最后我要说的是：那些有一定技术背景、不怕盯着终端然后说‘这到底是什么鬼’的人。我甚至会说，读一读 Claude Code 的输出，理解它在做什么，真的会让你成为一个更成功的构建者。所以我认为这是我在寻找的另一种人——也许是稍微学过一点计算机科学的人，或者那些曾经是设计师、写过一点代码、或者曾经是设计工程师的人。”

It'll take them a day or two, right? It's not that. So finding the customers is number one—real growth stuff—and two, having a really differentiating product that feels more premium and has an emotional connection with its users. Those are things that fall squarely in the design world. So I think those are two non-obvious skill sets—maybe three: product thinking, research, and branding—that are extremely powerful. The last one I'll say is people who have a somewhat technical background and are not afraid to look at a terminal and be like, 'What in God's name is this?' I'd even say reading the output of Claude Code and understanding what it's doing actually makes you a more successful builder. So I think that's another persona I'm looking for a little bit—or people who maybe studied a little bit of computer science, or people who have been designers and have coded a little bit, or have been design engineers.

“我认为这些技能——虽然从技术上来说，确实，AI 替我们做了一切——仍然非常有用。因为我确实遇到过这种情况：我会说，‘好了，Lovable，我知道你在干什么。你卡在一个 for 循环里了，或者说你陷在一个不断递归某个奇怪东西的世界里，而我知道怎么帮你把它调试好。’我真的觉得这是我以前辅修计算机科学时练出来的肌肉。我无法解释我的大脑为什么莫名其妙地把那套能力应用到了这个世界，但它确实还在起作用。”

I think those skills—even though technically, yeah, AI does it all for us—are really helpful. Because I've certainly found myself where I'm like, 'Okay, Lovable, I know what you're doing. You're stuck in a for loop or something. You're stuck in this world where it's just recursing over some weird thing, and I know how to help it debug itself.' And I really do think that's a muscle I built as a minor in CS back in the day. I can't explain how my brain has somehow applied it to this world, but it still does.

所以，我想说的是，我现在理想中的设计团队——再说一次，我觉得这就是招聘这么难的原因——我们对人的要求很高，但要的是那种 T 型人才，可能在某些领域有几项专长。作为领导者，我一直相信，T 型意味着公司领导有责任去了解每个人的优势和热情所在，知道如何在各个方向上培养他们，让他们变得全面。所以我当然不是那种只会招完全符合我清单上每一项的人。具体看岗位，我可能会有一些硬性要求，比如消费级移动端，如果那是我工作中的大头。如果你整个职业生涯都在做企业级 SaaS，那可能会难一点，视情况而定。但比如有人从没做过用户研究，我会培训你，或者把你介绍给一位研究员，让你们配对，给你一些机会，让你在安全的环境里慢慢适应。如果你不太擅长产品思维，那视觉设计这块会有点难——我觉得这在某种程度上是通过视觉设计体现出来的，可能需要教育或自学。然后，老实说，我觉得产品思维在很大程度上就是经验。

And so, I would say my dream design team right now is – and again, I think this is why hiring is so hard – it's like we're asking a lot of people, but it's like that T-shaped generalist that maybe has a few of those areas. Now, as a leader, I've always believed T-shaped means it is your responsibility as a leader at a company to understand what people's strengths are, what their passions are, and to know how to grow people in the areas to make them well-rounded. So I'm certainly not someone who will only hire people who check off every box on my hiring list. Depending on the role, I might have a certain non-negotiable, like consumer mobile, if that's a big part of my world. If you've done enterprise SaaS your whole career, maybe it's a little trickier, depending. But if someone's never done research before, I will train you, or I'll connect you with a researcher and pair you up, and give you moments to safely start to feel comfortable. If you're not as much of a product thinker, visual design gets a little harder – I think that comes through visual design to some extent, and maybe it's education or self-education. And then, honestly, I think product thinking is just experience to a large extent.

你说的很多内容，我在做这个角色的时候也注意到了——我有点像红娘，连接很多品牌和设计师，这真的让我充满活力。但我注意到的是，市场对视觉品牌设计师的需求大幅上升。很多人在用『品牌工程师』这个词，因为他们希望你更懂技术。他们希望你拥有建造者的心态，但同时也是能真正推动产品视觉形象前进的人。我认为我的理解是，这来自于超越基线的需求，因为现在每个人到达基线的速度真的非常非常快。

A lot of what you're saying, I'm kind of noticing in playing this role – connecting, I guess, my own sort of matchmaker thing between a lot of brands and designers – it's super life-giving for me. But something I'm noticing is there's a real big uptick in demand for more visual brand designers. A lot of people are using the phrase “brand engineers” because they want you to be more technical. They want you to adopt the builder mindset, but be the person that can really push what the visual identity of this product is forward. And I think my understanding is that it's coming from this need to elevate beyond the baseline, because everybody's getting to the baseline really, really, really quickly right now.

而且我觉得具体来说——对，你说的这些都同意——我认为动效设计是其中非常非常庞大的一个方向。我在 AI 出现之前就说过很多次。在一个很多产品设计出于实际原因都变得非常相似和平坦的世界里，你能设计多少次长信息流加轮播卡片呢？有很多种排版方式，但我真的觉得现在很多品牌表达可以借由动效设计来实现。而且我认为这是产品设计师非常低估的一项技能。众所周知，动效做不好的时候会很糟，做好的时候又很棒。我有一个关系很近的同事，他是真正的动效设计师，他总说：『显然，少即是多，要非常克制。没有什么比 UI 上出现一个跳舞的热狗更糟糕了。』所以我认为这是其中一个领域——同样，现在的 AI 工具不会用令人惊叹的动效给你惊喜，这在今天还做不到。所以，能够做创意指导、美术指导，在 After Effects、Rive 这些工具里创建素材来展示和说明，或者制作那些 Bodymovin 动画——这种技能，我认为属于视觉技能的一部分，而且我发现非常难找。所以如果你是这样的人，并且具备我们刚才说的所有其他能力，那你现在就是顶尖 1%。

And I think specifically – all that, yes – I think motion design is one of those things that is so huge. I've said this for a long time, before AI. In a world where a lot of product design is very similar and flat for practical reasons – how many times can you design a long feed with carousel cards? There are a bunch of ways to style that, but I really think a lot of brand expression today can come through motion design. And I think it's a very underutilized skill by product designers. As we all know, it can be so bad when it's bad and so good when it's good. I mean, one of my close colleagues is a legit motion designer, and he's like, “Less is more, obviously. Be very subtle. Nothing worse than a dancing hot dog type of moment on your UI.” So I think that's one of those areas where, again, the AI tools right now are not going to surprise you with an amazing motion thing – that's just not doable today. So again, being able to creative direct, art direct, create artifacts in After Effects, Rive, whatever, to show and tell, or to create those Bodymovin animations – that skill set, I think, is part of the visual set, and it's something I find very hard to find. So if you are that person and have everything else we described, just know you are in the top 1% right now.

我完全也看到了。最能引起共鸣的作品集通常都带一点动效。如果你真的想脱颖而出，如果你能让你的作品动起来，或者只是加入一个非常流畅的转场，甚至只是你思考作品构图的方式本身有动感，那甚至都是很大的加分项。AI 显然在推动很多这方面的发展，但有一部分也只是新工具把门槛降低了很多。我以前一直对 Lottie 望而却步。我以前会在 After Effects 里瞎弄。那真是一条艰难的制作管线。创建一个东西，然后让它和 Lottie 配合？太痛苦了。

I've totally seen that too. The portfolios that resonate the most have a little bit of motion. If you really want to separate, if you can make your stuff move, or just bring in one really seamless transition, or even just the way you think about the compositions of your work – if those move, even that is a big plus one. AI is obviously pushing a lot of this forward, but some of it is just new tools lowering the floor so much. I was always so intimidated by Lottie. I used to mess around with it in After Effects. That's a hard pipeline. To create something in After Effects and then get it to work with Lottie? That sucked.

是啊。

Yeah.

突然之间，我现在在用 Jitter。就像，『哦，这太容易了，我 10 分钟就能搞定，而且效果还挺棒。』或者最近让我很惊讶的另一件事是，给定一个 SVG，Claude 在给 SVG 加动画方面非常厉害，超出了我之前的想象。SVG 动画的上限比我想象的要高。只是我以前没有足够的能力去实现。然后我突然想到——回到你之前关于原型制作的黑盒类比，就是那个结账体验——我完全不知道这是否可能。我就想，这可能吗？你能做到类似的效果吗？我就说，你能不能试一下？然后它就试了。我说，天哪。好，太酷了。如果我们做 XYZ 呢？你懂吧？

All of a sudden now I've been using Jitter. It's like, “Oh, this is really easy. I can do this in 10 minutes, and it's quite awesome.” Or another thing that's been blowing my mind recently is, given an SVG, Claude's pretty freaking good at just animating the SVG, beyond what I thought was possible. The ceiling for SVG animations is higher than I thought it was. And I just didn't have the skills in order to do it. And all of a sudden I'm like, kind of back to your blackbox analogy from prototyping that checkout experience – I have no concept if this is possible. I'm like, is this possible? Can you even do something like this? I'm like, can you just try it? And then it tries it. I'm like, oh my god. Okay, cool. What if we did XYZ, you know?

完全同意。而且顺便说一下，在动效这方面，对立面是：我去过的每一个地方，工程团队在动效上的能力往往都比较弱。首先，能找到一个热爱设计、能做像素级工作的前端工程师，我们就已经很幸运了。但即使是我工作过的顶级 FAANG 公司里的那些工程师，很多人也会说，『我不知道怎么处理动效，同时还要保证性能。』所以，这也许又是一个设计工程师可以填补的空缺，真正把动效和代码结合在一起。

Totally. And the flip side, by the way, on the motion side, is that every single place I've ever worked, the engineering skill set tends to be weaker on the motion stuff. First of all, we're all very lucky to find a front-end engineer that loves design and can do pixel-level work. But even those engineers at top FAANG-level companies I've worked at, many of them are just like, “I don't know how to work with motion and have it be performant.” So again, maybe an opportunity for a design engineer type to fill that gap potentially, and really kind of master motion and code together.

嗯，这是一个当设计师的好时代，我们今天的对话也正好说明了这一点。能听到你的经历，还有你现在做的各种小实验和东西，以及你怎么思考不同的角色和格局，真的非常开心。你真是知识和经验丰富。非常感谢你今天来和我们分享，Ron。

Well, it's a fun time to be a designer, and everything about this conversation just points to that. It's been a real joy to hear about your journey and then all the little experiments and things you're building now, and just how you're thinking about the different roles and landscape. Wealth of knowledge and experience. I really appreciate you coming on and sharing it with us today, Ron.

哦，太开心了。和你聊天真的很愉快，伙计。

Oh, such a joy. Such a great hang to spend time with you, for sure, man.