AI 时代的设计系统：从令牌到核心

基本上在一夜之间，LLM 让设计系统变得重要得多。那么，在当今世界，我们应该如何看待一个系统到底是什么？

Basically overnight, LLMs made design systems way more important. So, how should we think about what a system even is in today's world?

如今，系统已经成为一切成功工程工作的核心。我们越是依赖工具来生成代码，就越需要依赖系统来告诉工具该编写什么代码。

By now, the system becomes the centerpiece of all successful engineering work. The more we rely on a tool to create our code, the more we rely on a system to tell that tool what to code.

这会如何改变设计组织的形态，以及我们作为团队协作的方式？

How is this changing the shape of design orgs in the way that we collaborate as a team?

这种‘从任何地方开始’的理念就是新的工作方式。你可能从文档开始，可能从画布开始，可能从前端开始。你可以从这些任何地方入手，但最终你会在各种工具之间来回切换。

That 'starting from anywhere' philosophy is the new way of working. You might start in a document, you might start in a canvas, you might start in the front end. You might start in any of those places, but ultimately you're going to be bouncing through the tools.

欢迎收听 Dive Club。我是 Rid，这里是设计师永不停止学习的地方。

Welcome to Dive Club. My name is Rid, and this is where designers never stop learning.

本期节目的嘉宾是 Louie Orriach，他在 Figma 担任设计倡导者，不断帮助团队应对当今不断变化的环境，尤其是在设计系统方面。所以，我们打算深入聊聊他今天观察到的所有趋势，以及这些对设计师意味着什么。但首先，我请 Louie 给我们梳理一下自他上次上节目以来，设计系统领域发生了哪些变化。

Today's episode is with Louie Orriach, whose role as a design advocate at Figma means that he's constantly helping teams navigate today's changing landscape, especially when it comes to design systems. So, we're going to do a little deep dive into all of the trends that he's noticing today and what it all means for designers. But first, I asked Louie to give us a rundown of everything that's changed in the world of design systems since he was last on the show.

自从我们上次谈到 Figma 的 variables 发布以来，过去两年半、三年里，一切都摆上台面，一切都变了。没错，但你还是能看到一些相同的脉络。所以我要说，当时大家过度强调了 token 的重要性，要把它弄对，要重构系统。那花掉了大量时间。有些公司花了一年、两年才走到重构那一步。一旦那件事稳定下来，我觉得行业又给我们扔了一个曲线球，引入了人工智能，也就是基于 LLM 的代码生成，人们开始重新思考系统到底是什么、用途是什么，至少是那些走在前沿的人。那让我有点害怕。我觉得也让很多人有点害怕，不知道这对他们的角色、他们每天的工作产出意味着什么。然后，可以说那是一年多以前。现在我们进入了一个更稳定的阶段，工作流看起来更清晰了，而且我们开始看到——可能谢天谢地——文档对于一致、可预测的输出的重要性。所以，我觉得经历了“所有事都重要”到“什么都不重要”的摇摆之后，现在我们的落脚点是：你一直在做的那些事情，对于我们要建立的新工作流来说，都极其重要。

The last two and a half, three years since we spoke about the launch of Figma's variables, everything's on the table and everything has changed. Yeah, but you still have threads of the same. So I'd say that there was an over-indexing on the importance of tokens, getting that right, refactoring systems. That took an enormous amount of time. Some companies took a year, two years to get to that point of refactoring. Once that felt stable, I think the industry threw us a curveball and introduced artificial intelligence, LLM-based code generation, where people started to rethink what systems actually were and what their purpose was, at least those at the forefront. And that made me a bit scared. I think it made a lot of people a bit scared about what that meant for their roles and what their daily output would look like. Then, let's say that was a year ago, plus. Now we're in a position where there's a little bit more stability in what that workflow looks like, and we're starting to see, probably thankfully, the importance of documentation in consistent and predictable output. So I'd say there's been swings back and forth between everything being important, nothing being important, and now we're landing at a position where all those things you have been doing are exceptionally important for the new workflows that we're trying to establish.

对，我笑了，因为我自己也有一个小小的观察视角，就是一直在采访人。你知道，从我们上次聊天到现在，我采访了大概 200 人，差不多这个数。确实有一种情绪在说，设计系统已死，云云。然后就是，一切都关乎速度，设计系统只会碍事。而现在钟摆完全摆到了另一头，每个人都在加倍投入设计系统，因为它明显是速度的驱动力。而且他们写了那么多文档，其实真的很有价值，只是不再是为人类写的了。

Yeah, I'm smiling because I've also had my own little vantage point, just interviewing people. You know, I've interviewed like 200 people since we last talked, something in the ballpark of that. And there was definitely this sentiment around how design systems are dead, yada yada yada. And it's like, everything's about speed, design systems just get in the way. And the pendulum has swung to the complete opposite end, where everyone's doubling down on design systems because it's so clearly a driver of speed. And all the documentation they wrote is actually really valuable, it's just not for people anymore.

是的，我会说，完全坦白讲，2025 年 1 月的时候，我以为我们全完蛋了。我们陷入困境，因为我们的工作不再有意义。还好那只是暂时的感觉，好吧，持续了几个月。就像你所说，那些文档的重要性，无论它们以前是什么形式，都将成为每个人未来工作流的基础。它们可能只是换一种写法，面向不同的格式、不同的输出，但指南对我们做的每一件事来说都变得至关重要。

Yeah, I would say, to be totally honest, in January 2025, I thought we were all toasted. We were in trouble because our jobs were no longer relevant. And that was, thankfully, a temporary feeling—well, a couple of months. And like you said, the importance of those docs, whatever form they were in before, is going to be the foundation of everyone's workflow going forward. They might just be written slightly differently, for a different format, for a different output, but guidelines are becoming paramount for everything we need to do.

对，也许我们可以再深入一点。你说系统的目的某种程度因为 AI 发生了转变。背后的原因是什么？

Yeah, maybe we can just go a little bit deeper there. You said the purpose of systems has kind of shifted as a result of AI. What's behind that?

很多人把系统看作一个团队，可能待在角落里，做自己的事，提升质量门槛，确保业务能够高效、快速地扩展。但现在，系统成为所有成功工程工作的核心，因为我们越是依赖工具来生成代码，就越需要依赖系统来告诉工具该写什么代码。而且我们对系统的投入越少，就会产生越多——说实话——垃圾进入系统，接下来的几年里它们会像垃圾填埋场一样堆积起来，最后总得有人去清理。所以，除非我们现在把这些系统当作变革的核心，否则几年后我们就会陷入那种局面：得靠顾问进来收拾烂摊子。

A lot of people looked at systems as a team, kind of potentially in the corner, doing their thing, raising the bar of quality, and making sure that the businesses can scale efficiently and at speed. But now the system becomes the centerpiece of all successful engineering work, because the more we rely on a tool to create our code, the more we rely on a system to tell that tool what to code. And the less we invest in a system, the more we're going to generate—honestly—rubbish into our systems that will just pile up like a landfill over the next couple of years, and somebody's going to have to fix it. So unless we now treat those systems as a centerpiece for change, we're going to end up in that situation in a few years where a consultant's going to have to come in and fix all the mess.

你最近发表了一篇文章，叫‘生成式设计系统’。你能谈谈你的想法吗？

You recently published an article called a generative design system. Can you talk a little bit about your thinking there?

这完全不是什么新说法，但人们确实想自动化很多东西，我们现在正在摸索哪些可以自动化、哪些应该自动化。我跟那些确实想自动化组件生成的团队聊过，也跟那些完全不想这么做、想自动化其他环节的团队聊过。可能是 token 管道，可能是文档的某些方面。但我们正在走向这样一个世界：大量此类工作可以通过工具或 LLM 完成，随便你怎么称呼。作为个人、作为公司，甚至作为公司里的一个团队，我们必须决定：我们现在真正的角色是什么，工具正在接管什么，而且要尽快决定，趁这个决定还没被替我们做出来。

This isn't a new phrase at all, but people want to automate a lot of things, and we're in a process now of figuring out what can be automated, what should be automated. And I speak to teams who do want to automate the generation of their components, and speak to teams who absolutely do not want to do that, and want to automate some other aspects of it. It might be token pipelines, it might be some aspects of documentation. But we're moving to a world where a lot of this work can be done via a tool or an LLM, whatever framing you want to put on it. And we have to decide—as an individual, as a company, or even as a team within a company—what our roles actually are now, and what the tools are taking over, and decide that as soon as we possibly can, before that decision is made for us.

快速说个消息，然后我们继续。刚刚发布了更智能版的 Lovable。它在解决复杂任务上提高了 71%，这意味着它可以更自主地做更多工作。还有更智能的规划、提示队列，所以你可以把请求堆在排队里，Lovable 继续干活。我最喜欢的部分是自动化测试，这意味着 Lovable 现在会像真实用户一样测试你的应用。它会打开自己的浏览器、走流程、研究边界情况，并帮你抓 bug。最棒的是，当它发现问题时，会当场修复。我真的相信 Lovable 是当下构建软件最简单的方式，所以前往 dive.club/lovable 试试这个新版本吧。

Real quick message and then we can jump back into it. An even smarter version of Lovable just released. It is 71% better at solving complex tasks, which means it can do more work more autonomously. There's more intelligent planning, there's prompt queuing, so you can stack requests while Lovable works. And my favorite part, there's automated testing, which means Lovable now tests your apps like a real user. It opens its own browser, navigates flows, investigates edge cases, and catches bugs for you. And the best part is when it finds a problem, it fixes it right on the spot. I genuinely believe Lovable is the easiest way to build software today, so head to dive.club/lovable to try the new release.

你知道我兴奋的是什么吗？Granola MCP。它允许你把 AI 工具直接连接到你的 Granola 会议笔记。作为一个最近一直在 vibe coding 很多工具的人，这个发布相当重要，因为我的会议笔记是我拥有的最有价值的上下文之一。现在我可以找到特定主题、提取行动项，并基于我的会议历史回答任何问题。这些现在都已可用，而且我已经在疯狂用它构建了。所以，如果你想加入我，去 dive.club/granola-mcp。那是 granola-mcp。好了，回到正题。

You know what I'm excited about? The Granola MCP. It allows you to connect your AI tools directly to your Granola meeting notes. And as somebody who's been vibe coding a lot of tools recently, this release is a pretty big deal, because my meeting notes are some of the most valuable context that I have. And now I can find specific topics, pull out action items, and get any question answered based on my meeting history. This is all available today, and I'm already building with it like crazy. So if you want to join me, head to dive.club/granola-mcp. That's granola-mcp. Now, onto the episode.

我想从你的视角出发，听听你的见解，毕竟你的工作会接触很多这样的团队。你看到了很多，知道什么在起作用，也知道要避免哪些坑。随着 AI 带来的一切变化，你看到团队在采用哪些最具创新性的技术或流程？

I want to tap into your perspective as somebody who works with a lot of these teams, given your position. You see a lot—what's working, what pitfalls to avoid. What are some of the most innovative techniques or processes you're seeing teams adopt as a result of everything that's changing with AI?

我看到的是很多人感到被赋能。可能是一个有技术好奇心的设计师，也可能是一个有设计好奇心的工程师，或者是这个三角形的另一个角：一个觉得自己无法交付的产品经理。所有这些人都觉得自己可以更接近生产，或者至少是更高保真度。我很不愿意说一切现在都可以直接生产，因为那可能不是重点。也许我们可以做出更高保真的原型，更快地测试和验证，然后仍然把它们扔掉——因为很多时候，原型的目的就是测试一个想法。但在这个过程中，是什么阻止任何人为之做出贡献？我在 Figma 内部或者其他团队那里都看到这种情况，借用一句俗话，人们正在“渗入”另一个人的职位描述。我认为这很好，因为我们是团队。从历史上看，不管我们喜不喜欢，人们都各自在自己的筒仓里工作。你把一个文件交给工程师去构建，或者产品经理把 PRD 交给你去调研。问题是，为什么我们不都以专业知识为所有这些阶段做出贡献？这并不是说专业知识不重要。绝对重要，我们现在有机会成为自己领域里更大的专家，但我们都作为一个整体更紧密地协作，而不是依赖那些筒仓。

What I'm seeing is a lot of people feeling empowered. It could be a technically curious designer, a design-curious engineer, or another angle of this triangle: a product manager who doesn't feel like they can ship. All of these people feel like they can get closer to production, or at least to a higher fidelity. I'm very nervous about saying everything can be production now, because maybe that's not the point. Maybe we can get higher-fidelity prototypes that we can test and validate much faster, and still throw them away—because that is the point of a prototype a lot of the time: to test an idea. But what is stopping anybody in this process from contributing to that? I see this internally at Figma and externally with other teams, where people are, for want of a better phrase, bleeding into another person's job description. And I think that's great, because we are teams. Historically, whether we like it or not, people have worked in their own silos. You pass an engineer a file to build, or a product manager passes you a PRD to investigate. The question is: why aren't we all contributing to all these phases with expertise? That's not to say expertise isn't required. Absolutely, we now have an opportunity to become even bigger experts in our field, but we all collaborate more tightly as an individual unit, not relying on those silos.

我知道这期节目的很多答案都取决于团队类型。所以，也许我们先在大公司的领域里待一会儿。如果有听众还没有采取多少措施来赋能非工程师，让他们达到这种更高保真度的原型制作水平，或者让他们能更多地接触代码里已有的东西，那么团队现在正在利用这个时机做哪些基础性的事情？这些事情应该是当今几乎每个人都应该牢记在心的。

I know a lot of the answers for this episode depend on the type of team. So maybe we'll hang out in bigger-company land for a bit first. If somebody's listening and they haven't taken many steps to empower non-engineers to reach this higher-fidelity level of prototyping, or to have more access to what exists in code, what are some baseline things teams are doing to take advantage of the moment—things that should be top of mind for pretty much everybody today?

这是一个与时髦无关的回答，但更多地作为一个团队待在一起，建立让你能够同步和异步协作的仪式。那是一个 Slack 频道、一份文档、一个 Figma 文件，或者你更喜欢在其中协作的任何文件，但要做为一个整体来做，而不是说：“我是产品经理，所以我拥有 PRD。我是设计师，所以我拥有设计。我是工程师，所以我拥有这个想法的高保真版本。”而是设计师能接触到代码库，工程师能贡献到 PRD，设计师也能在这个过程的各个阶段之间灵活进出。说到大公司，我看到很多大型企业给自己定了一个非常激进的 Q1 目标，基本上就是要彻底重塑他们的流程。这很令人兴奋，也有一点吓人。我们曾以为确定无疑的设计流程现在可以重新定义了，因为事情可以通过跨职能团队更快完成——而这正是我们一直想要的。我们一直想要这个。这不是什么意外或不同寻常的事情。我在和 CTO 们交谈，他们告诉我，这不再是设计负责人的角色，而是技术负责人的角色。CTO 们告诉我，他们现在的工作效率高多了。我的反问是：你用省下来的时间做什么？答案是：做更多的产品。人们看到自己的成长机会现在更宽了，他们可以根据用户反馈做出更多产品，而反馈也能更快获得，从而建立更好的业务。

This is a non-zeitgeisty response, but hanging out together as a team more, building rituals that allow you to collaborate synchronously and asynchronously. That means a Slack channel, a document, a Figma file or whatever file you prefer to collaborate in—but doing it as a unit, not saying, 'I'm the product manager, therefore I own the PRD. I'm a designer, therefore I own the design. I'm an engineer, therefore I own the high-fidelity version of this idea.' It's a designer having access to the code base. It's an engineer contributing to the PRD. And it's a designer flexing in and out of these phases of the process. Speaking of large-company land, I'm seeing so many large enterprises giving themselves a very aggressive Q1 target of basically reinventing their process. That's exciting, and a little scary. The design process we thought was true is now up for grabs, because things can be done faster as a cross-functional team—which we always wanted. We always wanted this. This isn't surprising or out of the ordinary. I'm speaking to CTOs who tell me this is not a design leader's role, it's a technical leader's role. CTOs tell me they're able to work much more efficiently now. My question back is: what are you doing with the spare time? And the response is, building more products. People see that their opportunity for growth is wider now, and they can make more products based on user feedback, which they can also get faster, and build better businesses.

当流程可以重新定义时，我确实觉得很多团队可能会过于热心。我自己也时不时看到这种情况。我相信你作为更多承担这种角色的人，也遇到过。对于那种情况，有没有需要避免的陷阱，或者一些迹象或信号，会表明：也许我们正在追错目标？

When process is up for grabs, I do think a lot of teams maybe get overzealous. I've even seen it from time to time—and I'm sure you have, as somebody in more of that role. In that scenario, are there pitfalls to avoid, or signs or signals that would point to the fact that maybe we're chasing the wrong thing?

是的，当然。这要联系回系统核心。我有一段时间一直担心，如果设计系统是一个孤立的团队，那就意味着人们可以说这是别人的工作。对我来说，设计系统是一个组织里的质量标杆，而那是每个人的工作。所以，当流程被重新定义时，你就有了机会说这是别人的工作，或者不是。我们仍然需要有清晰的责任边界和对每个个体的产出期望。设计师仍然在做设计师角色特有的事情。不是说产品经理现在可以贡献原型了，所以我就不做原型了。我们必须接受一点：我们的角色有一个最低限度的基本产出，再加上“plus”。而这个 plus 由你的行业、你的公司、你的团队、你的领导层、你的经验以及你在组织中的级别来定义。更乐观的人会想要拓宽自己的技能，变得更像一个通才。悲观的人会在角落里抓狂，想着：“我现在不知道能做什么，因为我什么都能做。”组织需要决定预期产出的最低标准可以是什么、应该是什么。然后设计师可以延伸到自己想去的不同方向。对我来说，最接近的机会是成为一个更好的写作者——作为设计师，多写文档，因为在我看来，这是沟通想法的最好方式。你可以使用音频，也可以录下自己的声音，不管怎样，在你投入像素之前把事情写下来，是让一个想法获得支持的最好方式。除非你获得支持，否则你无法把想法交付出去。所以文档对我来说仍然是核心。

Yeah, for sure. This links back to the systems centerpiece. I've had a concern for a while that design systems being a siloed team means people can just say it's someone else's job. To me, a design system is a bar of quality in an organization, and that's everyone's job. When we're reinventing process, there's an opportunity to say it's somebody else's job—or not. We still need clear lines of responsibility and output expectations for each individual. The designer is still doing something specific to a designer's role. It's not, 'Oh, the PM can contribute to prototypes now, therefore I'm not going to make one.' We have to have this level of acceptance of what our role can do at a bare minimum, plus. And that plus is defined by your industry, your company, your team, your leadership, your experience, and your level within the organization. The more optimistic person is looking to widen their skill set to become more of a generalist. The pessimistic person is freaking out in the corner, thinking, 'I don't know what I can do now, because I can do everything.' It's up to an organization to decide the bare minimum of what expected output can or should be. Then designers can stretch into different angles they'd like to. For me, the closest opportunity looks like becoming a better writer—as a designer, write more documents, because that in my opinion is the best way to communicate an idea. You could use audio, you could record your voice—regardless, writing down before you commit to a pixel is the best way to get buy-in for an idea. And you're not going to get the ability to ship an idea unless you get buy-in. So a document is still central to that for me.

在当今世界，这其实是个挺大胆的观点。现在流行语不就是“原型是新的 PRD”吗？

Kind of a spicy take in today's world actually. Like 'the prototype is the new PRD' is the jargon now.

嗯，问题是，一旦你投入像素，就会有人对视觉设计有意见。但你需要让它获得签字批准。在一个有前瞻性的组织里，你必须让一位设计负责人、产品负责人——不管头衔是什么——在投入之前批准你的想法。而我所见过的做到这一点的最好方式，是写一份文档，说明你的发现和计划。我们作为跨职能团队围在一起，添加评论，征求其他想法，也许再加上一个粗略的示意图。但在我们去做那个高保真原型之前，让我们先确保我们在做正确的事情。顺便说一句，这违反我的直觉。我的直觉是尽快投入像素。我也不知道。

Well, the thing is, as soon as you commit to a pixel, someone's going to have an opinion on the visual design. But you need to get it signed off. In a forward-thinking organization, you have to have a design leader, a product leader—whatever the title is—sign off on your idea before we commit to it. And the best way I've seen to do that is to write a document of your findings and your plan. We huddle around it as a cross-functional team, add comments, ask for other ideas, maybe add a loose mock-up of your idea. But before we go and build that high-fidelity prototype, let's make sure we're building the right thing. That goes against my instinct, by the way—my instinct is to commit to pixels as soon as possible. I don't know.

我两者都做。当然，我认为两者都有价值。

I do both. I see value in both for sure.

所以如果说有什么的话，我觉得我很欣赏这种更细致入微的观点，因为上节目很容易就说‘用原型而不是 PRD’。我想花点时间再深入聊一下团队结构这块，不如我直接给你抛一个假设场景。假设你被请去当顾问，服务的是一家正在成长的 B 轮公司，他们到现在基本上没在设计系统上投入过，为此很痛苦。他们没有专职的负责人，也没有专门的团队在维护设计系统。在几乎是一张白纸的情况下，你会抓住哪些机会？你会建立什么样的体系或安排什么样的人来持续管理它？

So if anything, I think I just appreciate the more nuanced take because it's very easy to come on here and say prototypes instead of PRDs. I want to double down on the team structure piece for a second and maybe I'll just toss a hypothetical your way. So let's say that you're brought in as a consultant for, let's say, a Series B org that's growing and feeling the pain of having basically made no investment into the design system up until this point. They don't have any dedicated people. There's no dedicated squad working on the design system. Given almost like a blank canvas, what would be some of the opportunities that you would pursue, and what types of systems or people would you put in place to steward it moving forward?

首先，我要回到关于质量的观点：质量是每个人的职责，也是每个人的事。如果我们不认可或不追求质量，那就该认为自己正在失败。这是我的深切感受。我们必须设定一个高标准，而具体长什么样则因公司而异。但不追求质量，就必然要承担后果，因为我们会吸引更少的客户、做成更少的业务、产生更少的营收。所以，“我们对质量的定义是什么、如何达到”应该贯穿到我们做的每一件事里。这与体系无关，因为我认为对质量的追求会让你在不知不觉中先形成一个体系。但从战术上讲，我认为你需要一个对技术好奇的设计师，和一个对设计好奇的工程师。而且，这个体系不能是事后才想的，不能是边角项目，也不能是额外工作之外还要做的某件事，它应该像一条线一样贯穿整个组织产出的每一件东西。我们是不是在搭建组件？那本身就是体系。我们没必要给它贴上标签，因为我担心给“设计系统”打上标签，会引发我们不想谈的预算问题。

Firstly, I would say back to my point about quality: quality is everyone's job. And quality is everyone's business. And if we don't accept or pursue that, we should be treating ourselves as failing. And that is something I just feel deeply. We have to have a high bar, and that is company-dependent what that looks like. But if we're not pursuing that, there should be consequences, because we're going to attract less customers, we're going to sell less business, and we're going to produce less revenue. So that should just be sewn into everything we do—what is our definition of quality and how do we get there? That is regardless of the system, because I think that the pursuance of quality gets you to a system before you've even know it. But tactically, I guess you need a technically curious designer and a design-curious engineer. And for the system not to be an afterthought or a side project or something you're doing on top of extra work, it should just be a thread through everything that you're producing as an organization. So are we building components? That's a system. We don't need to put the label on it, because I worry that putting a label on something like design system can introduce conversations we don't want to have about budget.

而围绕着招聘、专业化分工的讨论，与一位前端工程师真正做好本职工作相比——我认为，一位设计师真正做到位，体系会作为副产品自然产生。当我们开始扩展规模，也就是在你的场景里，要构建另一个产品时，最简单的例子就是引入深色模式。那时才需要真正开始系统地整理诸如令牌（token）之类的东西。那是当时可以讨论的话题。但到那时候，因为我们之前已经做对了，我们有 CSS 变量，可以直接扩展。如果到那时前端里还到处是原始值，那就说明我们之前没做好，我们在构建过程中就会意识到，根本没法再像 Slack 那样继续下去了。我在自己的工作中也有这种感觉。差别就在于，哪怕只是在一个像 claw.md 这样的文件里加入检查规则，内容大概是“嘿，这些地方不要用原始值，碰到某类组件应该这样获取”，然后在 GitHub 审查时自动执行，你知道吗？其实非常神奇。你会感受到，对“这个体系应该是什么样”——或者用你的话说，“质量基准应该在哪”——投入哪怕一点点，LLM 就能帮你从那时起在每一次 PR 中持续命中这个基准。太神奇了，真的。

And hiring and specialism versus a front-end engineer doing a really good job. And a designer doing a really good job, to me, produces a system as a byproduct. When we start to scale it up, so in your scenario, you're building another product. The easiest one is introducing dark mode. That's when you have to start systematizing properly things like tokens. That is a conversation we can have at that time. But by that time, because we've already done a good job, we have CSS variables and we can extend them. If by that time we have raw values across our front end, we haven't done a good job and we would know that as we're building it. That's like Slack is just not going to fly anymore. I'm feeling that even in my own work. Like the difference of even just putting checks in like a claw.md file where it's like 'hey, we don't use raw values in these places. Here's how we reach for certain types of components' and it just is automated in like a GitHub review, you know? It's amazing actually. Like you can feel how a very small amount of investment into what this system should be, or in your language, where the quality bar should be, and then how much the LLMs help you continuously hit it with every single PR from that point. It's amazing. It really is.

不过，我要提出场景的另一面：因为我们能快速生成任何想要的东西，而早期公司成功的很大一部分取决于它给终端用户带来什么感受——用户有多想再次回到你这里？这一点极其重要。可以说，对体系的偏离才是制胜关键，因为每次与用户对话，我们都得用不同的方式吸引他们。那么，一个无限灵活的品牌长什么样？到这一步，我们归根到底说的是品牌。那也许意味着原始值是没问题的，因为对于这个落地页的这个变体，我们正在做实验。而一旦很快进入系统化，我们就是在做承诺。所以，确实存在这样的争论：令牌（token）还重要吗？还是必须从第一天就有，因为我们无时无刻不在考虑扩展？我们要在“实验”和“对质量的承诺”这两个世界之间跳舞。实验也可以是一种质量，但过度令牌化可能意味着它太固若金汤，以至于我们无法偏离。而产品或组织在每一个阶段都得判断自己处在这条线的哪个位置。而且，既然一切都能生成，扫一眼代码库看有没有原始值只需半秒，那令牌化在那个时点真的有用吗？还是你只是在为了推像素而推像素？所以我们现在进入的世界是：原始值可不可以？如果不行，那你怎么管理？如果原始值可以，那就让它们只限于品牌感觉的范围。这可以写进你说的那份指导文档里。我们可以表达品牌应该让人产生什么感觉，然后依靠 LLM 帮我们达到。

Let me present the other side of the scenario though is that because we can generate anything we want at speed, and so much of the success of an early-stage company is how much it makes the end user feel. How much do you want to return to you? That is incredibly important. Which arguably means deviation from the system is the win or the key, because we need to be able to attract people differently every single time we talk to them. So what does an infinitely flexible brand look like? Because we're boiling it down to the brand at this point. And that might mean the raw values are fine, because for this permutation of the landing page, we're experimenting. And the sooner we get into a systematization, we're committing to something. So there is definitely an argument of do tokens matter anymore or absolutely they have to be there from day zero, because we're thinking about scale at every moment. And we have to dance between these two worlds of experimentation and committing to quality. Experimentation can also mean quality, but an over-tokenization of something can mean it's so robust we can't deviate from it. And every stage of your product or organization has to figure out where you lie on that line. And maybe because everything can be generated and can take you half a second to scan a codebase to see raw values, does the tokenization actually help you at that point? Or are you pushing the pixels around because you can? So the world we're entering now is, is a raw value fine? Or if it's not, then how do you manage that? If a raw value is fine, let's just limit them to what the brand feels like. And that can be written in a document, the guidelines that you're talking about. We can express what the brand should make someone feel like and rely on the LLM to get us there.

而到这时，我说话有点自相矛盾了。因为一方面，内心深处我想的是体系、体系、体系。另一方面，我想到的是我每天沟通的所有团队，他们都在为“采用”这个词而挣扎——大家没有在正确的地方使用正确的东西。而那个世界就要被彻底颠覆了。不，它已经发生了，已经发生了。是的。

And this is where I like dance over my own words because in some ways, deep down I'm thinking system, system, system. In other ways, I'm thinking all the teams I talk to every single day who struggle with this word adoption, that people are not using the right thing in the right place. And that world is just going to be flipped upside down. Now, it's already happening. It's already happening. Yeah.

现在甚至连“体系”这个词都有点奇怪。因为做抽象化的目的是：为了跨代码库进行扩展性修改，把需要你拧的旋钮降到最少。可现在，那种痛苦已经大大消减，区别只是调用一次 LLM 花 10 秒还是 25 秒。比如我就无所谓，你知道吗？它照样能把所有东西都找出来。所以到那个点，体系到底是什么？而且我确实有点喜欢你把它拉回到“品牌”这一点，以及你想唤起的那种感觉。因为具体如何构建，我不知道，也许意义真的没那么大了。不，我不确定。这很奇怪，它已经装不进我原有的框架了。

Even the word system is weird now, because like the point of abstracting something was to minimize the amount of knobs that you have to turn in order to make changes that scale across the codebase, but now it's like the pain of that has diminished so significantly, where it's the difference between a 10-second LLM call and a 25-second LLM call. Like I don't care, you know? It's still going to be able to find all of the things. So like what is a system at that point? And I do kind of like how you're bringing it back to brand actually, and like the feeling that you're trying to evoke, because the specific way that that is constructed, I don't know, maybe it does mean less. No, I'm not sure. It's weird. Like it doesn't fit into my box anymore.

我们面临的挑战——这根本不是新观点——如果我们仅仅依赖代码生成，而没有对品牌感受的严格定义，最终就会让所有东西看起来都一个样。是的。说实话，在大多数情况下这没问题。但如果你真想运用你那个拿到一大笔融资、需要加速的场景，光是这样就不够了。所以我们必须想清楚：它看起来是什么样、感觉如何、语气语调是什么、意象是什么。而这就把我们从一个“组件是什么”的问题，拉回到“公司应对加速的体系是什么”的问题。

The challenge we have, and this isn't a new point at all, but if we rely just on generation of code without a very strict indication of brand feeling, we end up with all the things looking the same. Yeah. And honestly, that is fine for the majority of cases. But if you really want to take your scenario that received a ton of funding, we need to accelerate, that is not good enough anymore. So we have to figure out what that looks like, what that feels like, what the tone of voice is, what the imagery is. And it dials us back out of what is a component to what is the company's system of approaching acceleration.

把「设计系统」这个标签从我们所做的事情上拿掉，反而会有帮助。它会把你从角落里、从孤岛中带出来，带到组织变革的中心。它几乎就像是产品的基线质量标准。但如果你始终只停留在那条基线上，产品可能会很无聊，就像有人周末 vibe code 出来的东西。所以，有了基线之后，关键是你策略性地偏离系统的那些点，去做那些根本不在我们给模型的规则集里的事情。这里是有层级之分的。一个人自己花一个周末把它搭出来，也许能帮他拿到融资继续做下去。但这并不意味着你 12 小时做出来的东西就应该直接上线，它只是一个验证点。然后一路加速到那条线的顶端，你拥有一个完整的 token 化系统，可以扩展、甚至可以并购公司进来，以更横向的方式建立一个更强大的品牌。但你必须先搞清楚自己在这条线上处于什么位置——这条线不会移动——才能知道产品的每一个点需要多高的保真度，而不仅仅是系统本身，因为系统哺育产品，产品也哺育系统。没错，系统过去的很多价值主张在于确保一致性和标准化，但现在我们确实感觉这些东西很多会开箱即得。

So, taking the label of design system off what we're doing is going to help. It's going to bring you out of the corner and out of the silo, to the center of an organization's change. It's almost like the baseline quality line for a product. But if you only stayed at that baseline at all times, then it's probably going to be a boring product that someone were to vibe-code in a weekend. So, given the baseline, now where do you strategically deviate from the system and do things that simply don't exist in the set of rules we're giving to the models? And there are elevations there. The person spinning something up by themselves at a weekend might use it to secure some funding to develop it further. That doesn't mean the thing you built in 12 hours is what you should ship. It's just a validation point for something. Then you accelerate all the way up to the top of the line, where you've got this fully fledged tokenized system, which you can expand and even acquire and merge companies into, and build out a more robust brand in a more horizontal way. But you have to figure out where you are on this line—which won't shift—to understand what level of fidelity you need at every single point of a product, not just the system, because the system feeds a product, and the product feeds the system. Yeah, it's like a lot of the value proposition of a system was in making sure that things are consistent and standardized, but it does kind of feel like we're going to be getting a lot of that out of the box.

所以，这已经不够了。你得想办法让产品产生一种依附感。而且，我觉得 Shadcn 在这里起到了很大的作用。现代 Web 技术栈有很强的网络效应。我们越转动这个飞轮——大家都在用 Tailwind，大家都在用 Shadcn，大家都在 Vercel 上构建——基线就越来越高。

So that's not enough anymore. You have to find ways to create a level of attachment to a product. And I think Shadcn played a huge role in this. There are so many network effects around the modern web stack. The more we spin this flywheel—everybody building with Tailwind, everybody building with Shadcn, everybody building on top of Vercel—the baseline is rising.

对吧？所以，一个好看、干净、一致的设计，在很多方面只是入场券。我不希望人们把时间花在重新发明已经存在的东西上。我觉得那样的日子不多了；但让设计看起来像你的品牌，才是你在一个已经被验证的系统之上投入精力的方式。我们花了太多时间试图让无障碍成为某个人或某家公司的优先事项。如果它能开箱即用，那就拿着吧。你不需要去自己定制一个 popover 组件的版本。你已经被说服了。这非常了不起。质量门槛，就像你说的，更高了。所以，让我们专注于那些能帮我们卖出更多产品的事情。这对追求像素级完美的设计师来说可能不是最有吸引力的主张，但这就是工作本身。而且这很令人兴奋。你可以更接近营收、展示影响力，也更容易得到你迫切想要的那次晋升，因为你能够更好地接触到利益相关者。

Right? So a good, clean, consistent design is table stakes in many ways. I don't want people to spend their time reinventing something that already exists. I think those days are limited, but making it feel like your brand is the way you can put energy on top of a system that is already proven. We've spent so much time trying to figure out how to make accessibility someone's priority, or a company's priority. And if you can get that out of the box, take it. You don't need to go and custom spin up your own version of a popover component. You've already been sold. And that is amazing. The quality bar, like you said, is higher. So let's focus on what will help us sell more products. That might not be the most attractive proposition for the pixel-pushing designer, but that's the job. And that's exciting. You can get closer to revenue and showing impact, and get closer to that promotion that you're desperate for, because you're able to get better access to stakeholders.

我想深入聊聊工作流，以及设计师的实践。还是想借助你的视角，因为你跟很多团队聊过。你如何看待企业级和初创公司之间目前的差距有多大？不仅是在这些系统中设计师的运作方式，更宽泛地说，就是他们如何面对作为设计师的日常工作——不管在今天这个时代，这个身份到底还意味着什么。

I want to dig into workflows and the practice of a designer. Again, tapping into your vantage point as somebody who talks to a lot of teams. The first thing I'd love your take on is: How wide is the gap right now between enterprise and startups, in terms of the way designers are operating in these systems, but also more broadly, approaching their day-to-day role as a designer—whatever the heck that still means today.

我感觉差距在越来越大。之所以觉得越来越大，是因为工具的可及性因公司的安全标准而差异巨大。如果你在一家传统企业里，想批准一个 AI 工具可能非常难。如果你是一家高速成长的初创公司，你大概开箱就能拿到这些东西。签份合同，工具就到手了。所以，有与没有的鸿沟，或者说我们过去所说的数字鸿沟，在这个微观层面似乎也在扩大。这让我有点担心，因为我不想让那些所在组织不给他们工具访问权的人，觉得自己被困在某个岗位上，无法跳槽到更进步的组织。这是行业需要尽快解决的问题，因为它可能变成潜在的问题。但谈到工作方式，初创公司冲得很猛，他们一向如此，总是想突破边界。但我注意到的是，人们想要更快地进入生产。他们几乎想跳过设计——这个说法可能有点戏剧化，但给我的感觉就是这样：我有一个想法，我现在就想让它在某个域名上线。最快的方式是什么？这就是我上周在聊的对话。而反过来，对于拥有数百万用户的企业软件，你在发布方式上必须更慎重。于是问题就变成：我们如何缩短上线时间？能用什么工具加速反馈？如何让更多用户进入 beta？如何用功能开关确保一小部分受众先帮我们测试？我们对软件的社区化方式是什么？我们如何建立一个可以依赖的团队，让他们因为在意我们而持续给我们反馈？我觉得随着公司从初创走向企业，最终每个人都会用上这套打法。但对我、对我们来说，幸运的是，社区、人、用户、热情，对一个想法的成功仍然无比重要。只是我们现在做得更快了，但无论多想提速，质量这一关不能跳。

It feels like it's getting wider. And the reason it feels like it's getting wider is because the access to tools is so vastly different based on your security standards as a company. If you're in a legacy enterprise organization, it might be really hard to get an AI tool signed off. If you're a high-flying startup, you've probably just got these things out of the box. You sign the contract and you just have these tools. So the haves and have-nots, or the digital divide, as we used to call it, in this microcosm, seems to be getting wider. It does give me a little bit of concern, because I don't want people to feel like they could be stuck in a role and not be able to get out of it into a more progressive organization if they're in one that doesn't give them access to tools. So that's something the industry needs to figure out pronto, because that could be a potential problem. But how people are working—the startups are high-flying, they've always been doing that, they're always trying to push the boundaries. But what I have noticed is that people want to get to production a lot faster. They want to almost skip design, and that is a potentially dramatic phrase, but it's how it comes across: 'I have an idea, and I want to get it live on a domain now. What's the easiest way to do that?' And that was the conversation I was having last week. And on the flip side, for enterprise software where you've got millions of users, you have to be a bit more considerate about how to release something. So that then looks like: How can we shorten the time to production? What tools can we use to accelerate feedback on this? Or how can we increase the number of users that get access to a beta? Or how do we use feature flags to ensure that a subset of the audience can test this thing out for us? What's the community approach to our software? How can we build a team of people that we can rely on for consistent feedback, because they care about us? And that is a playbook that I think everybody will use eventually as you go up the stack from startup to enterprise. But thankfully, for me and for us, community, people, users, passion is still incredibly important for the success of an idea. It's just that we're doing it faster, but we can't skip the quality no matter how fast we want to move.

过去一年，Dive Club 让我非常清楚一件事：设计的实践正在改变。旧的反馈流程在今天的世界已经不太管用了。所以我很兴奋地宣布，Inflight 正式进入公开测试。这是我一直想要的反馈工具，它针对一个以 AI 速度运转的世界而建造。我可以分享我的原型，用视频演示提供背景，而 Inflight 能让我轻松获得继续推进所需的精确反馈——无论是给方向投票，还是给新想法放行上线。所有这些都通过一个链接完成，我可以丢进 Slack，或者分享给核心用户来测试新原型。我每天都在用 Inflight，它彻底改变了我分享工作的方式。所以我很期待你来试用这个产品，如果你想一起聊聊，就发邮件到 rid@inflight.co。

One thing that Dive Club has made abundantly clear to me over the last year is that the practice of design is changing. And the old process of getting feedback just doesn't quite cut it in today's world. That's why I'm excited to announce that Inflight is officially in open beta. It's the feedback tool that I've always wanted, and it's built for a world that moves at the speed of AI. So I can share my prototypes, give context in video walkthroughs, and Inflight makes it easy to get the exact feedback that I need to move forward, whether it's voting on directions or maybe even getting the green light to ship a new idea. And all of this is available in a single link that I can drop into Slack or maybe even share with power users to test out a new prototype. I use Inflight every day, and it's totally transformed the way that I share work. So I'm excited for you to try the product, and if you ever want to jam about it, just email me at rid@inflight.co.

聊到工作流，我不得不问你关于你刚发布的 Claude MCP。我记得就是昨天的事。

I can't talk workflows without asking you about the Claude MCP release that you just dropped. I think it was literally yesterday.

那么，你脑海里想到的哪些东西让你感到兴奋？

So, what are some of the things that unlock in your mind that you're excited about?

这真的让我对那个世界感到兴奋——我很早以前作为一个专业人士就以为自己会进入的世界：脑子里有个想法，想好好测试它，想把它放进浏览器里，想感受它，想知道交互、转场、时机。而且能够从画布到浏览器，测试，然后觉得‘那其实不太对，我们回去，再即兴发挥一下。’对于一个非技术背景的设计师来说，这个世界会打开很多扇门，因为你可以更快地传达你的意图，也可以把它推得更远。我们在内部常说的一句话是‘不是更快，而是更远’。这能让非技术人员跳进浏览器，试一些东西，打开开发者工具，说：‘哦，那个间距应该是 12，不应该是 10 或 8。’然后跳回 Figma，让这个东西旋转着冒出 10 个版本。把那个推到浏览器，把这个推到浏览器。啊，还是不太对。我们通过 MCP 或随便什么工具回来，形成我们自己的反馈循环，不需要牵涉工程师——他们可能要几天或几周才能回复，这不是他们的错，而是流程问题。所以这部分流程感觉可以大幅提速。更进一步的关键是系统。如果你能把画布、浏览器以及仓库里的系统全部串起来，那么我们的迭代速度会快得多。对我来说，这就是我们一直想挖的金矿。

It really does get me pumped for the world that I thought I was entering as a professional a long time ago, where I have an idea, I want to test it properly. I want to get it into a browser. I want to feel it. I want to know interactions, transitions, timing. And being able to go from the canvas to the browser, test it, and be like, 'That's actually not quite right. Let's go back. Let's riff on it.' That world for a non-technical designer is going to open so many doors because you can communicate your intent so much faster because you can push it further. A little phrase we throw around internally is 'not faster, but further.' That enables people who aren't technical to jump in the browser, test something out, open the inspect, say, 'Oh, that spacing should actually be 12, not 10 or eight.' Let's jump back into Figma and have 10 versions of this thing spinning around. Push that one to the browser. Push that one to the browser. Ah, still not quite right. Let's come back via MCP or whatever we're using and have this feedback loop for ourselves without involving an engineer that might take days or weeks to come back, through no fault of their own, just through process. That part of the process feels like we could speed it up massively. The key to unlock even further is the system. If you can push from your canvas to the browser and back, all connected up to a system that lives in a repo, then we're going to iterate so much faster. That is to me the gold mine of what we've been trying to do.

我也是。我闭上眼睛，想象着永远把这两个窗口并排放在一起，那才是我想要的工作方式。没错，关键是系统，它们指向同一个东西，但有时候我也就想要一块白板，你知道吗？我就想快速涂几笔，然后说：‘不不不，像这样。’然后立刻把这些交给 Claude 或 Codex，或我正在用的任何工具，自如地进进出出。感觉我们会回看过去这——我现在不知道是什么时间了，时间线在我脑子里已经糊了——但大概就是过去三到六个月，对吧？这是雷达上的一个小点：作为设计师，我们当时面对一个虚假的选择，你必须选一条路，而且很难跨到另一条。结果，我每天醒来都会问：‘我还算设计师吗？’因为在很多初创环境里，就像你描述的，我直接打开 localhost，把一个想法变成现实，然后点击发送。我所做的全是在写代码。那我还算设计师吗？我不知道。有趣的地方在于，角色的边界在变宽，你可以随时进入或退出任何对你有意义的事情。当我刚开始以专业身份做前端时，我停下来了，因为搭建环境实在太复杂。我记得 Docker，哦天哪。

Same. I close my eyes and I picture just having those two windows side by side at all times. That's how I want to work. There's an element of yes, it is the system and they're pointing at the same thing, but sometimes I also just want it to be a whiteboard, you know? I just want to be able to quickly scratch and like, 'No, no, no, kind of like this instead.' And just immediately have that go to Claude or Codex or wherever I'm working, and fluidly move in and out. It feels like we'll look back on the last, I don't know what it is at this time. The timelines are blurring in my head, but it's the last three to six months, right? As this blip on the radar where we had this false choice as designers where you had to kind of pick a path and it was very difficult to traverse to the other one. As a result, I wake up every day and I'm like, 'Am I even a designer anymore?' Because a lot of times in the startup environment like you're describing, I just open up localhost, bring an idea to life and hit send. All I'm doing is writing code. Am I still a designer then? I don't know. This is the beauty that the role is getting wider and you tap in and out of whatever makes sense to you. When I first started professionally contributing to front end, I stopped because things got so complex to set up an environment. I remember Docker and oh god.

NPM install、React、打包——我当时就想，这不是我现在该在的地方。首先，你说这些术语的时候我完全不知道你在说什么；如果请工程师帮我搭环境都要花几个小时，那我们基本上就走错路了。所以我们退出来，让设计师不用理解那些东西也能做到这些。想法胜出。它不是一个需要你跨越的技术障碍，反而把人推得更远——这正是发生在我身上的事。它把我推离生产，推回那种在画布上打磨像素的角色。现在我们可以再退回来，也许在中间某个地方相遇：画布对构思、协作和探索大方向非常重要，但浏览器是我们最终落地的地方。我们在这两种媒介之间来回切换，得到正确的想法，我觉得这正是我们想要的。

NPM install, React, bundling — and I just thought that is not where I am right now. Firstly, I've got no idea what you're talking about when you say these terms, and if it takes a couple of hours for an engineer to help me set something up, we've kind of gone down the wrong path. So we reverse out of that into a designer being able to do this without thinking about what any of those things mean. The idea wins. It's not a technical barrier that you have to jump over and pushes people further away, which is what happened to me. It pushed me further away from production and more into the polishing pixels on the canvas kind of persona. Now we can reverse out of that one and maybe meet somewhere in the middle where the canvas is very important for ideation, collaboration, and general direction exploration, but the browser is where we commit to it. We bounce back and forth between these different mediums to get the right idea, which I think is exactly what we want.

我喜欢‘想法胜出’这个说法。确实感觉这就是那个世界的发展方向。我非常看好设计师。任何有想法的人都能受益，这正是它的美妙之处。我喜欢‘造物者’这个词，因为它不会限定你在组织里的具体角色。如果工程师想玩交互细节，完全没问题。如果设计师发现某个卡片在特定屏幕上的边框颜色与背景的对比度更高，想调整一下，请直接去改那个 token 值。不要被以前的流程束缚。而且你多次把话题拉回协作，我也很有感触——是的，我确实被赋能了，我能做很多事，但在远程公司里，有时候感觉就只有我和我的智能体。

I love the idea wins. It definitely feels like that's where we're headed in that world. I'm pretty bullish on designers. Anybody with an idea is the beauty of this. I like the word maker because it kind of doesn't define specifically your role in an organization. If an engineer wants to play around with interaction details, absolutely go for it. If a designer wants to tweak the border color on a card because they've spotted that there's a better contrast against the background on this particular screen, please go and change that token value. Don't feel restrained by the process that was before. And like how multiple times you brought it back to collaboration too, because something that I'm feeling a lot is like, yeah, I'm empowered, for sure. I can do a lot, but sometimes it kind of feels like especially in a remote company, it's just me and my agents.

不，我整天做的事情就是指挥我的智能体，最后我和 Claude 合作的时间比和其他人合作的时间多得多。我不知道，这就像一个小小的信号，让我觉得：嗯，这感觉有点怪。我还没想好该怎么处理。变化太多了。

No, like all I am doing is just wielding my agents all day, and I end up collaborating with Claude so much more than other people. And I don't know. It's one of those tiny little signals where I'm like, huh, this feels weird. I don't know what to do with it yet. So much is changing.

但你觉得自己在学习吗？

But do you think you're learning?

哦，我学得疯狂。我认为在一个工具之间来回往返，在你向所有人承诺之前先让自己对想法更有把握——你怎么看？我觉得这没什么问题。把所有事情都外包给某个工具，才是我更担心的。因为，没错，我们是掌控局面、把产品推向市场的专家。如果我们把想法、产出、对任何事物的定义都外包给工具，那我们就要问问自己，我们到底想做什么。如果我一整天只是在一个工具上按‘开始、开始、开始’，我作为一个造物者、创作者，会感到沮丧。有时候我也确实会这样。我在做 Figma 插件的时候，我完全不看那些代码，但只要能用，我就很开心。但我仍然拥有这个想法。我在笑，因为昨天我基本上一整天都在做前端。我现在用 Conductor，这有点像巴甫洛夫的狗——你就从一个喇叭声跳到下一个。感觉就像在打地鼠：好，这个智能体需要我。好，现在那个智能体需要我。现在这个智能体需要我。

Oh, I'm learning like crazy. I think that bouncing back and forth between you and a tool to get more sure of your idea before you commit to everybody — what do you think about this? I don't see a problem with that. Outsourcing everything you do to something is where I get a little bit more concerned. Because yeah, we are the experts in the seat to bring something to market, and if we're outsourcing our ideas, our output, our definitions of anything to a tool, then we just need to ask ourselves what we really want to be doing. And I would feel deflated as a maker, as a creator, if I was just pressing go, go, go all day on a tool. Sometimes I do. If I'm making a Figma plugin and I'm coding it, I absolutely don't look at that code, but if it works, I'm happy. But I still have the idea. I'm laughing because yesterday basically all I did was work on front end. I use Conductor now, and it's basically Pavlov's dog a little bit because you just live from one toot of the conductor horn to the next. It feels like I'm playing this whack-a-mole game where it's like, all right, that agent needs me. Okay, now that agent needs me. Now this agent needs me.

我有时候会掉进这个陷阱。回到你刚才说的，获得认同、尽早且频繁分享非常重要，因为当你能整天和智能体就你的想法来回切磋，并且作为一个设计师走得比从前远得多——我会抠这个前端架构的每一个细节、所有悬停状态和交互——就会忍不住一直打磨、一直打磨，而不是跳出来说：‘嘿，我在做这个，这是我想的这么几个方向。’ 当我的流程更固定、总是从我在画布上按 F 键画一个矩形开始时，这种平衡来得非常自然；但现在我第一步就能跳得更远。有利有弊吧，我就这么说。别误会，我也从代码起步。我觉得‘从任何地方开始’这种哲学就是新的工作方式。你也许从文档开始，从画布开始，从前端开始——可以从其中任何一个地方开始，但最终你都会在工具之间来回切换。说实话，甚至可能从 Slack 线程开始。对。如果你想继续即兴发挥、得到最好输出，这个过程最终还是会回到线性穿过这些阶段。不过，我不反对从代码开始，因为有时候你得靠手感才知道哪儿不对，而且很多时候在画布上画出来，比直接说‘嘿，给我生成一个网页’要慢得多。我会说‘不对，不是这样。’ 但我要说，这些工具不是用来做‘嘿，给我做个 App’或‘嘿，给我做个网站’的。它们的用途不是这样，我也不认为我们应该这么用。但把它落到非常具体的事情上，比如组件级创建，对比宽泛的‘请帮我解决世界难题’——你不觉得用这些工具就该是这样吗？我今天收到一条 YouTube 评论，一个怀疑者说：‘我还没见过 AI 设计解决过真正的问题。’ 我就想，等等，你说的 AI 设计是什么意思？因为对我来说，那就像是在假设我们做的事就是喊一句‘给我做个网站’。我是用 AI 获得一种以前从未有过的控制力，而且是在工具之间不断跳进跳出。但别误会，普通人开一家餐厅就需要一个网站，如果 15 分钟能搞定，他们就会用。我完全没意见，因为如果我能去读网站上的菜单，看起来还挺不错，能拿到联系方式、知道他们在哪儿、知道价格多少——作为消费者我也不介意。当我们开始谈软件，就完全是另一场游戏了，‘给我做个软件’是行不通的。我现在在帮我一个哥们儿，他刚开了一家本地承包公司，他特别想要个网站。我问他‘你想在网站上放什么？’他答不上来。

I feel this trap sometimes. Going back to what you were talking about earlier, it's so important to get buy-in and to be sharing early and often, because when you can play ping pong with an agent all day on your idea and go way further than previously possible as a designer—where I'm sweating the details on every single part of this front-end architecture and all of the hover states and interactions and everything—it's tempting to just keep refining and keep refining versus popping out and being like, 'Hey, I'm working on this. These are the couple of directions I'm thinking.' That came very naturally when my process was slightly more defined and always started with me hitting F on a canvas and drawing the rectangle. Whereas now, I can jump much further as step one. A lot of pros and cons, I guess. I'll put it that way. Don't get me wrong. I was starting with code too. And I think that 'starting from anywhere' philosophy is the new way of working. You might start in a document. You might start in a canvas. You might start in a frontend. You might start in any of those places, but ultimately you're going to be bouncing through the tools at some point. You might start in a Slack thread, to be honest. Yeah. And that will always come back to a process that goes linearly through these phases if you want to keep riffing and get the best possible output. But yeah, I don't have a problem with starting in code, because sometimes you need to feel something to know what's wrong, and it can take a lot longer often to draw out on the canvas than to say, 'Hey, generate me a web page for this thing.' I'd be like, 'No, that's not right.' I would say though, these tools are not for 'hey, make me an app' or 'hey, make me a website.' That's just not how they're designed, and not necessarily how I think we should be using them. But boiling it down to something real specific, even like component-level creation, versus broad 'please solve the world for me'—don't you think using these tools is about that? I got a YouTube comment today from some skeptic saying, 'I'm yet to see an AI design that solved a real problem.' And I'm like, 'Hold on a second. What does an AI design mean to you?' Because to me, that's like assuming all we're doing is just saying 'make me a website.' I'm using AI to get a level of control that I've never had before, and bouncing in and out of tools constantly. But don't get me wrong. The average person who starts a restaurant needs a website, and if they can get that done in 15 minutes, they're going to take it. That is totally fine by me, because if I can go and read the menu of the website and it looks kind of nice, get the contact details, know where they're based, know what the prices are—I as a consumer also wouldn't mind that. When we start to talk about software, we're entering a much different game. And that's where 'make me software' doesn't fly. I'm helping my buddy right now who's starting a local contracting business, and he really wanted a website. I asked him questions like, 'What do you want on your website?' He has no answers.

他根本不知道，你懂吗？

He has no idea, you know?

所以‘给我做个网站’的价值不仅在于可能用户没有技能，还因为绝大多数不做设计、不碰软件的人，连表达自己想要什么的能力都没有。正因如此，能直接在页面上吐出个东西来，大概是弄清你真正想要什么的最容易的方式，因为你看了会说：‘啊，我不要那个，我要这个。’ 所以我确实觉得这价值巨大。另外，这也不是什么新鲜事，因为过去人们花 5 美元找人做个网站，也已经有段时间了，而那让更多人成为创作者——或者在这个例子里，更多在意网站的人。而我如今看待这件事的方式也一样：更多人将能创造自己的想法。我们没有资格告诉别人什么能做什么不能做。但一旦那个餐厅网站需要再开一家连锁或分店，他们可能就需要一个更好的网站了。那时候他们大概会找专业人士来建一个更好的系统。

So 'make me a website' is not only valuable because maybe the skill isn't there, but also because the ability to articulate what you want isn't there for the vast majority of people who don't do design or don't operate in software. And so being able to just spit something out on the page is probably the easiest way to even arrive at what you actually want, because you can see it and be like, 'Ah, I don't want that. I want this instead.' So I do see a ton of value in that. Additionally, this isn't a new thing either, because we've had the ability for people to go and spend $5 on someone making them a website for a while, and that made more makers—or more people who cared about websites in this case. And in the same way, I look at this now: more people are going to be able to create their ideas. We're not in a position to tell people what they can or can't make. But as soon as that restaurant website needs to open another chain or another store, then they might need a better website. That's where they probably engage a professional to build a better system.

好，理论上谈世界如何改变是一回事，但我想聊聊你的故事。因为你在 2025 年 1 月给我们画了那幅图景——你知道，就是那种‘这一切要去哪儿？我的角色意味着什么？’的焦虑。显然从那时起一切都变了。那你能不能聊聊你和 AI 的个人经历，以及这些拐点出现在哪里？

All right, so it's one thing to talk theoretically about how the world is changing. I want to talk a little bit about your story, though, because you drew us this picture of January 2025, and there's that nervousness about where is all this going? What does this mean for my role? Obviously everything has changed from that point. So can you talk a little bit about your personal journey with AI and where these inflection points are?

嗯，我花了一段时间自省，酝酿着接下来会发生什么的念头，然后走出来时更强了。我都没法跟你说我过去五年买了多少门 React 课程。

Yeah, I took a while for self-reflection and brewing the thought of what was about to happen, and came out stronger. I can't even tell you the amount of React courses I bought over the past five years.

我记得你至少发过三四条相关推文。

I can remember at least three or four tweets from you about it.

你一直在拿这个打趣我。我一门都没学完，因为我就是没兴趣也没时间。但去年二季度，2025 年，我想：等一下。插件可以作为进入我一直想学的东西的一个非常简单的切入点。打开一个 AI 工具——在那个早期阶段，早期阶段，这么说有点好笑。在这一切的早期，是一张企业版 ChatGPT 的许可证，我让它帮我做插件，像原始人一样把代码从浏览器复制粘贴到我的 IDE 里。像个原始人，但居然成功了。我当时想：好吧，我们手里真有非常强大的东西，我可以跳过那些浪费钱的课程，很快上线一个插件。一个星期、两个星期，差不多那样。后来更快了，几天。再后来几个小时。我想我在过去 12 个月里大概发布了五六个 Figma 插件或 widget，这挺疯狂的，因为它们有用，能解决问题。有几千个用户，人们给我反馈。它是开源的，大家可以去 GitHub 仓库看，想贡献也可以。我有一个仓库收获了一个 GitHub star。

You've been calling me out on that. I haven't finished any of them, because I just don't have the interest or time. But Q2 last year, 2025, I thought, 'Hang on a second. I could make plugins as a very easy entry point into what I've been trying to learn.' Pull up an AI tool—in the early days, early days, that's funny to say that. In the early days of this, it was an enterprise ChatGPT license, and I was asking it to help me make a plugin, copying code from the browser into my IDE like a caveman. Like a caveman, and it worked. And I thought, 'All right, we've got something really powerful here, where I can jump over those courses that I've wasted money on and ship a plugin very quickly.' A week, two weeks, something like that. Then it got faster. Couple days. Then it's a couple of hours. And I think I've published maybe five or six Figma plugins or widgets over the past 12 months, which is kind of nuts, because they work. They solve problems. There are thousands of users. People are giving me feedback. It's open source. People can look at the GitHub repos. They can contribute to it if they want. I received a GitHub star on one of the repositories.

你看你！这到底怎么回事？太厉害了。

Look at you go. What the hell's happening here? That's amazing.

而且我觉得这就是我们眼前的东西。对。就是利用这些工具，快速为我们自己的用例打造实用工具。所以我已经从浏览器里的 ChatGPT 来回复制代码，进化到更集成的工作流——用一款能让你在 IDE 里跑 LLM 的工具。我就是在聊天、来回碰撞、一遍遍重写，回到 Git 工作流、分支那些东西里，把代码托管在 GitHub Pages 上做落地页。这打开了以前被我关上的门。我能更快地推进这些想法，更快上线，更快获得验证，然后解决问题。

And I think that is what we have on the table. Yes. The ability to make utility tools for our use cases at speed using these tools. So I've gone from the ChatGPT in the browser, copying code back and forth, to a more integrated workflow where I'm using a tool that allows you to have LLMs based in the IDE. I'm just chatting, bouncing back and forth, rewriting, rewriting, rewriting, getting back into Git workflows and branches and all that sort of stuff, and hosting my code on GitHub Pages for landing pages. This just opened doors that I had previously closed. And I'm able to work faster on these ideas, get them out faster, get validation faster, and solve problems.

所以，所有这些插件或小工具，都来自与社区的交流：他们遇到什么困难，我就发现一个帮忙的机会。这样实际上我也能把工作做得更好。说到励志，我也是那种买了多得不好意思承认的编程课程的人。我会上一整个训练营。其实我都不确定自己算不算正式毕业了。但我花了所有钱。他们把我钱全拿走了，就是最后没给我那颗金星星。我觉得自己一度有些放弃希望。有那么几年，我心想：“我永远做不成这个，永远做不到。” 但回到你用的那个词——赋能，大概就是这样吧。我也说不清。现在只要有一点空闲时间，我就想做点什么。我就觉得自己能从指尖射出闪电，你知道吗？太神奇了。我收藏了一个 Google 表格，里面存着创业网站的清单，存了两年吧。大概有 800 到 900 个网址。我一直想找个地方把它放出来。前两天晚上，我坐在沙发上打开笔记本电脑，对着 AI 聊起来：“我想做这个网站，这是表格，我们能做点什么？” 它就做了规划，我们一起搭了个东西。然后我合上电脑，之后就没再管过。我能做一堆新东西，但很多还是没做完。不过，我仍然觉得晚上那一小时用来探索非常值。而且我能以更高的保真度去探索。十多年前，我跟一个工程师一起工作。我几乎每天都去找他：“我又有一个创业点子，咱们做个某某领域的 Twitter 吧。” 或者 “做个新的物业管理系统吧。” 诸如此类。他会说：“嗯，通讯录里有些开发人员，大概要花 5 万到 10 万美元来做。” 我就想：“好吧，点子到此为止。” 现在我会想，花几天时间写提示词，大概就会发现这个点子其实不好。但至少我能伸展一下、把它试出来。也许某天某个点子会对我有意义，我就能投入进去。但现在投入的不是金钱，而是时间，还有一点钱，比如按月订阅的服务。作为设计师，我想有这种想法的不止我一个。有时候你就是想把东西往墙上扔，看看什么能粘住。大概率大多数都粘不住。那也没关系。但现在我们能更快地做这件事。这也改变了成功的定义。因为当软件很贵的时候，要成功就必须有某种规模，对吧？在必须投入那么多时间和金钱的情况下，专门为身边一小群朋友做东西就没有意义。但现在不一样了，比如我做过一个超级碗 App，只是给家人玩，一起看球时玩玩。有 15 个人用了，我觉得非常成功。明年我还会做。而且这大概只花了我两小时。不过我觉得关键的是，我们做得出某样东西，不代表它就非得传播开来或有大范围的吸引力。因为现实是，如果你想把它当生意做，那从一开始就找错了方向。嗯。

So, all of these plugins or widgets have come from just conversations with the community about what they’re struggling with, and spotting an opportunity to help. So, I’m in effect able to do my job better as well. Inspiring because I’m also the kind of person who has bought more coding courses than I care to admit. I would do a whole boot camp. I don’t actually think I ever technically graduated from it. But, I spent all the money. They just took all my money. I just didn’t get the gold star at the end. I think I actually had some giving up of hope. There was like a couple years. I was like, “I’m never going to be able to do this. Like, I’m never going to be able to do this.” And going back to your word, empowering, I guess. Like, I don’t know. I’m at the point right now where, if I have any free time at all, I just want to build things. I just feel like I can just shoot lightning out of my fingertips, you know? It’s amazing. So, I’ve been keeping a Google sheet of startup websites for a couple of years. There’s probably 8 or 900 websites in this Google sheet. I’ve had a goal of just putting it somewhere. And two nights ago, I opened my laptop on the couch and just chatting away and saying, “I want to make this website. Here’s my Google sheet. What can we do?” And they planned it out. We built something. And then I closed the laptop and haven’t done anything with it since. I can make a bunch of new stuff, and I still don’t finish them. But, I’m still able to feel like that hour or so in the evening was really useful to just explore. And I can explore a higher fidelity. I used to work with a guy over a decade ago, an engineer. And I would just come to him on a daily basis like, “I’ve got a new startup idea. Let’s build the Twitter for this.” Or, “Let’s build a new property management website.” Whatever it was. And he’d say, “Yeah, I’ve got some developers I’ve got in my contact book. Probably going to cost us 50 to 100,000 to make it.” And I think, “Okay, end of the idea.” And now I’m thinking, spend a couple days prompting away, and probably realize the idea is not good. But, at least I’ve been able to flex that and test it out. And maybe one idea one day will mean something to me. I can commit to it. But, instead of committing financially now, it’s committing with time. A little bit of money, like a monthly subscription. As a designer, I think I’m not alone in this. Sometimes you just want to throw things at a wall and feel what sticks. And probably most of the stuff doesn’t. And that’s okay. But, we can do it faster now. It also changes what the definition of success is, too, because when software is expensive, in order to be successful, it has to have some semblance of scale, you know? It wouldn’t make sense for you to make something for your immediate group of friends in a world where you have to spend the amount of time and money required to do it. But now, like, I made a Super Bowl app just for my family, just to play games together while watching the game. And 15 people used it, and it was a massive success. And I will totally do it next year. And again, yeah, it took me about 2 hours. The key thing I think though is, just because we can make something doesn’t mean it has to be widespread or it has to have mass appeal, because the reality is, if you wanted to do that as a business, that’s the wrong way to place a start. Yeah.

你现在需要去推销这个点子，而不是把它做出来。

You need to go and market the idea now rather than build it.

我也做过一些当副业的创业项目，但那不是重点。有时投入了好几年时间去做一个东西，却没人知道它。市场已经变了，而且不会回头。我们能用到这些工具，能更快地做出东西、上线运行，不代表就会有人看到它。也不会有人来注册。所以，把范围收得更紧、做小一点。也许做给家人的东西，才是最能满足这种手痒感觉的地方。有个词，我在这期播客里想尽量多提，就是“利基经济”。因为我觉得这就是未来的方向。你知道，你可以做出一个解决一千人问题的东西。就像你给家人做的那几个插件。别人不会觉得“这是你的创业项目”或“这是你的生意”，对吧？但换个角度想，作为设计师，如果你能这么做 20 次呢？你只要能找到问题，就一个个解决，你知道吗？

And I’ve also worked on sort of startups as a side project where that hasn’t been the focus, where a couple of years have been sunk into building something with no one who knows about it. And that market has shifted. And that won’t change. Just because we have access to the tools, we can make things faster, we can get it into production, doesn’t mean there’s any eyeballs at all on it. And no one’s going to sign up to this thing. So, keep it tighter, keep the scope smaller. And maybe something for your family is the best place to get that itch scratched. There’s a phrase that I’m on a mission to use as often as possible on this podcast, which is the niche economy. Because I totally think that’s where this is headed. You know, you can create something that solves a problem for a thousand people. Kind of what you’re doing with your family plugins, you know? People wouldn’t look at that and be like, “That’s your startup.” Or, “That’s your business,” you know? But, all of a sudden, like, well, what if you could do that 20 times as a designer? You just solve any problem that you can find, you know?

那么，考虑到你所有这些摸索，考虑到过去 18 个月这段弯弯曲曲的旅程——对你来说可能像过了大半辈子，对我来说也是。就你现在这个位置，你到底是怎么看待接下来职业生涯的？

So, given all of your explorations, given this windy journey of the last 18 months for you that probably feels like a lifetime, I know it does for me. How the heck are you thinking about your career moving forward given where you’re at?

说实话，我不知道，因为从来就没知道过。我从来没规划过职业，就是抓住出现的每一次机会。换个说法，我投了很多次篮，有的进了，有的没进。但我不会停止投篮。我也不清楚终点在哪里。行业的变化总会尘埃落定，事情会起起落落。我基本就是坚持下去，专注做我喜欢的事：帮助整个社区更好地构建软件。这会通向哪里？希望只是更好的软件。我的角色是什么？用我的平台去让别人也能做到这一点。我有一年的职业规划吗？没有。两年、五年、十年？绝对没有。六个月呢？也没有。

Honestly, don’t know, because I’ve never known. I’ve never planned my career. I’ve just taken the shots that have come up. Or, put it a different way, I’ve taken a lot of shots. And some of them make the basket, and some of them don’t. But, I would not stop shooting. And I don’t know where that ends. The industry shift will settle inevitably. And things will boom up and down. And I’m kind of just holding on and focusing on what I like to do, which is help the community at scale build better software. Where does that go? Hopefully, just better software. What does my role in that look like? Using my platform to enable people to do that. Do I have a one-year career plan? No. Two, five, 10? Absolutely not. Six months? No.

三个月呢？

Three months?

没有。光是每周都在变化的东西的数量，就足以让我觉得，像我这样的人去规划职业是徒劳的。因为我连公司的路线图都没法依靠，去定位自己该处在什么位置。这对所有人都是如此。所以我会继续做我喜欢的事，尽量往前推，尽量施压。如果我不享受这个过程，那还有什么意义？

No. The amount of things that are changing on a weekly basis means that I think that it’s a fruitless task for someone like me to try and plan their career, because I can’t even rely on a company’s road map at the moment to know where I fit in that. And that’s for everyone. So, I’d say I’m going to continue doing what I enjoy, pushing that as far as possible, putting as much pressure on that as possible. If I’m not enjoying it, then what’s the point?

不过，作为更偏向教育角色的人，你会不会觉得要跟上所有东西有压力，尤其是现在变化这么快？

Do you feel pressure though as somebody who is in more of like an education role to keep up with everything, especially given how quickly it’s moving?

说实话，这挺难的。我已经接受一个事实：我永远不会什么都懂。以前不懂，以后也不会懂。市场、这个游戏变得广阔得多，简直疯狂。工具，以及在某些场合对设计师要变得更有技术性的期待，都是这样。我永远没法聊到一个全职 React 工程师的水平。这是绝对不可能的。但希望我能促成设计师和工程师之间的对话，帮他们在公司里建立更好的工作流程。所以，我确实感到一种个人压力，不是职业压力，就是需要知道在什么层级需要什么知识，才能把事情做完。我会想办法的。

That is difficult, I’ll be honest. I’m resigned to the fact that I will never know everything. I didn’t, and I never will. The market, the game is much wider, which is kind of crazy. The tools, the expectation for designers to, in some places, become a lot more technical. I’m never going to be able to talk at the level of a full-time React engineer. That’s just not going to happen. But, hopefully I can help facilitate a conversation between the designer and that engineer to build a better workflow for them in that company. So, yeah, I do feel a personal pressure, not professional pressure, to just know what is required at what elevation to get the work done. And I’ll figure that out.

它让我有机会去尝试这个角色的不同方面、这个行业的不同部分，并找到新的乐趣。过去一年，对我来说，就是重新开始做 GitHub 贡献。至于未来 12 个月，谁知道呢？我不想失去与写作重要性的联系，也不想失去对质量标准的重视。像素依然极其重要。就我个人而言，我一直在大量关注视觉设计。所以这永远都在。这是开放的，我会抓住机会。 我觉得自己一直是个彻头彻尾的超级早鸟。我痴迷于每天刷 Product Hunt，坚持了好几年，试用每一个工具。这对我来说太自然了。但我一直在纠结于一个现实：我已经不可能再是那样的人了。不可能。工具太多了，每周变化都太大。所以试图跟上几乎是一种失败状态。但你必须在内心建立新的筛选标准，来决定把时间投在哪里、什么值得追求。为什么我的 Twitter 信息流里 80% 都是 OpenAI 技巧？我甚至都不知道。当你只是想努力不掉队的时候，这有点令人紧张。而且我相信每个人都有这种感觉，所以我当时才会问那个问题，因为我知道你确实有压力去跟上这些东西。

It gives me the opportunity now to lean into different parts of the role, different parts of the industry, and to find new enjoyments. Over the last year, that has meant getting back into GitHub contributions. As for the next 12 months, who knows? I'd hate to lose touch with the importance of writing. I'd hate to lose touch with the importance of quality bars. Pixels are still incredibly important. And personally, I've been focusing on visual design quite a lot. So that's always going to be there. This is open. I'll take my shots. I think I'm someone who has always been an ultra early adopter everywhere. I was obsessed with scrolling Product Hunt every single day for years, trying every single tool. It just came so naturally to me. And I've wrestled with the reality that I actually can't be that person anymore. It's impossible. There are too many tools. Too much is changing every single week. And so it's almost a failure state to try. But then you have to develop new internal razors to figure out where I invest my time, what's worth pursuing. Why is 80% of my Twitter feed OpenAI hacks? I don't even know. It's a little intense when you're just trying to hang on for the ride. And I'm sure everybody feels that, which is why I was asking the question, because I know you do feel some pressure to stay up to date with the stuff.

我每周都会做活动。人们总是举手，非常坦诚地说：“我不懂，我不知道该做什么。”这让人听着很难受。但我们有平台，可以让人们了解什么是可能的。而且我认为，你很容易被无限的可能性压垮。还有你刚才说的，打开 Twitter，看到某一类内容，你可能会被那些看似在做非常相似事情的人淹没。我想，这可能是 Twitter 社区让人觉得有点失败的原因之一——除了显而易见的原因之外。因为人们不知道该在哪里立足，也不知道该聊什么，因为一切感觉就像每个人都在说同样的话。社区仍然非常重要，但如果人们不知道去哪里找志同道合的人，他们就会感到孤独。所以社区对人们的职业满足感和知道该做什么仍然极其重要。我只是不总是知道它在哪里，因为它感觉有点碎片化。

I do events every week. And people always put their hand up and honestly say, 'I don't get it. I don't know what to do.' That is tough to hear. But we have the platforms to enable people to understand what is possible. And I think you can easily be paralyzed by limitless possibilities. Even what you talked about then, opening Twitter and seeing a specific type of content—you can just be flooded by people seemingly doing very similar things. I think that's probably one of the reasons, other than the obvious, why the Twitter community felt like it failed a little bit. People didn't know where to fit in or didn't know what to talk about, because everything just felt like everyone was saying the same thing. Communities are still very important, but people don't know where to go to find like-minded people, so they just feel alone. Community is still incredibly important for people's career satisfaction and to know what to do. I just don't know where it is all the time, because it feels a bit fragmented.

在我让你离开之前，我想问一个有点出格的问题，因为我觉得现在有很多叙事。其中一些可能完全可以忽略，只是被夸大了。有一些是有一些道理的，设计师应该认真对待。我想知道，在你看来，有没有什么叙事是“其实没那么多道理、你可以放心忽略的”，而另一方面，有没有“世界正朝这个方向发展，我们作为设计师不能把头埋进沙子里”的？

I want to try a little bit of an out-there question before I let you go, because I think there are a lot of narratives out there right now. Some of them are probably quite safe to ignore; they're just overblown. Some of them have some truth, and designers should take them seriously. I'm wondering if anything comes to mind for you in terms of a narrative that doesn't have that much truth to it right now. Like, 'You can safely ignore this,' whereas, actually, 'This is where the world is heading, and we can't just bury our heads in the sand as designers.'

周末我去了父母家，晚饭时的谈话格调发生了很大的变化。我父母没上过大学，但 AI 成了饭桌上——其实是饭后喝了点酒之后——的话题。至少在我的圈子里，提到这些东西的第一反应通常是负面的，但大家都在给我讲他们的使用场景。我哥哥有金融背景，他妻子是心理学家，我姐姐在行政岗位，在一家卖灭火器的公司工作。他们并不是我们这类人，但他们都在说这些工具如何让他们的工作和生活变得更轻松。所以我认为我们告诉自己的主流叙事，并不是大众市场的叙事。而且随着尘埃落定，人们会找到每天都能帮到他们的真实用例。作为一个深度参与其中的人，这让我很难受，因为我整天在 Twitter 上被负面观点轰炸。但我能看穿这些，看到普通消费者确实有使用需求。我有个做市场营销的朋友，自己订阅了 OpenAI。使用 Projects 功能，他一直在运行大量文档，对日常生活帮助极大。他有注意力缺陷多动障碍（ADHD），这对他帮助很大。所以我把自己从“先看负面”的思维里拉出来，这对一个天生的悲观主义者来说相当难。

I went to my parents' house at the weekend, and I felt a very strong tonal shift in dinner conversation. My parents didn't go to university, but AI came up as a topic at dinner—or after dinner, after a few glasses of wine. And the prevailing thought, at least in my circles, is probably negative-first when it comes to this stuff, but people were just telling me their use cases. My older brother has a finance background. His wife is a psychologist. My sister works in admin for a company that sells fire extinguishers. These aren't people like us, but they were talking about how their jobs and lives have been made easier with these tools. So the prevailing narrative I think we tell ourselves is not the mass-marketed narrative. And as the dust settles, people will find real use cases that help them every day. That's hard for me, as someone so into it, to see, because I get hit with negative takes on Twitter all day. But I can see through it to the everyday consumer who has a use case for this. My friend who works in marketing has a personal OpenAI subscription. With Projects, he's got tons and tons of documents running all the time, helping him immensely in his everyday life. He has ADHD. It helps him a lot. So I pull myself out of the negative-first mindset, which is difficult as a natural pessimist.

我个人的叙事——我有，但不一定会讲出来的是——不管我们愿不愿意，作为设计师、制造者、普通人，大多数软件都很烂。而且人们并不真的在乎像素。让人们在意的这个挑战，一直落在设计师身上，包括我以前担任的职位。但现在的角色——因为市场上的软件太多、网站太多、企业太多——我们必须专注于销售、营收、用户、注册和收购。所以像素对业务成功来说次要。我们可以在有意义的时候再处理像素，或者根本不处理。随着我们制造更多东西，将会有更多软件被创造出来——更多网站、更多没有设计背景的人做的应用。我们只能接受，不能试图与之对抗。有些职位里，你的经理或经理的经理并不真正在意像素。而那是一场不值得打的仗，至少现在不是。

My personal narrative—one that I have but don't necessarily project—is that, whether we like it or not, as designers, makers, people, most software sucks. And people don't really care about the pixels. That has been a challenge for designers, including me in previous roles, to convince people to care about it. But the role now—because there's so much software on the market, so many websites, so many businesses—we have to focus on sales, revenue, users, sign-ups, and acquisitions. So pixels are secondary to the success of the business. And we can get to the pixels when it makes sense, or not at all. As we generate more things, more software will be created—more websites, more apps by people who don't have a design background. We just have to accept that and not try to fight it. There will be roles where your manager or your manager's manager doesn't really care about the pixels. And that is a fight not worth fighting, I don't think, at the moment.

我在想我自己的答案。我觉得现在有一种叙事：那些可能对 AI 稍微更怀疑一点的人说，角色边界的模糊只是因为企业想少付钱，他们当然想让你做更多，这确实正在发生，但一切都会稳定下来。我并不真正相信这是真的。我认为我们画了一条人为的线，把前端界面和实际前端代码分开，仅仅是基于我们当时拥有的技术能力。而我认为真正的前端实现会落入设计的范畴。它实际上是一整套完全属于 UX 定义的东西，但作为行业，我们一直拒绝承认这一点。所以我们把它推给了前端工程师，他们并不像我们那样在意细节。但总有一天我们会意识到，那其实就是 UX 设计，而这正是新的角色。

I was thinking about my own answers. I think there's a narrative right now where people who are maybe slightly more skeptical of AI say the blurring of roles is just because businesses want to pay less, and of course they want you to do more, and that's what's really happening, but everything will settle. I don't really believe that's true. I think we have drawn an artificial line between a picture of the front end and the actual front end, based only on the technical capabilities that were available to us. And I think the actual front end will fall under design. It's actually a whole collection of things that totally fall under the definition of UX, but as an industry we've refused to admit that. And so we've punted it off to front-end engineers who don't really care about the details as much as we do. But at some point we're just going to realize that's just UX design, and that's the new role.

现在有越来越多的企业被创建出来。小团队这个充满希望的观点意味着，更多的地方有更小的团队，在解决更多的问题。作为一个喜欢创业公司那种快节奏氛围的人，这令人兴奋。但对于喜欢在超大型组织工作的人来说，这可能是相反的感受。

There are more businesses being created. The hopeful argument of smaller teams means smaller teams in more places, solving more problems. As someone who enjoys that startup, accelerated-pace atmosphere, that's exciting. But for someone who enjoys working in very large organizations, that might be the opposite feeling.

但我确实看到更多创业公司、更多机会去融资并打造自己的想法。现在呢，是的，我完全同意。就像有一条界线——我这是用经济学思维在想——你知道，图表上真真切切有一条线：如果你在这个问题上没法产生这么多利润，那它就不值得去做。做这件事没有经济意义。而这条线刚刚倒下了。所以有整整一个子集——有整整一类问题，以前没有人有经济动力去解决，现在却完全敞开了。看看你周围，比如本地小生意。天哪，有太多问题，任何收听本期节目的人都完全有能力自己解决。大胆一试吧。

But I do see more startups, more opportunities to raise money and build your own ideas. Now, yeah, I totally agree. It's like there was a line—this is my economic brain thinking—but you know, there was a literal line on a graph where if you couldn't generate this much profit based on this problem, it's not worth pursuing. It doesn't make financial sense to do this. And that line has just fallen. So there's a whole subset—there's this whole category of problems that previously nobody had the financial incentive to solve—that are just wide open. Look around you, like local businesses. Oh my goodness. There are so many problems that anybody listening to this is totally capable of solving on their own, fully. Take a shot.

大胆一试。是的，没错，我喜欢这句话。嗯，我想不出还有比“大胆一试”更好的结尾了。

Take a shot. Yes, yes, I like it. Well, I can't think of a better line to end on than take your shot.

呃，Louie，感谢你来做客，给我们分享了一些内幕消息，也坦诚地讲了实际情况，因为我觉得你和所有人一样有同感，你知道，这是一个疯狂的年代。有时候我们需要把现实照亮，我很感谢你今天来和我们一起做到这一点。

And, Louie, I appreciate you coming on and giving us a little bit of the insider scoop, but also just telling it how it is, because I think you're feeling this like everybody else, you know, it's a crazy time. And sometimes it's good to just shine a light on the reality, and I appreciate you coming on today and doing that with us.

总是很开心。非常感谢。

It's always fun. Thank you very much.

一如既往地荣幸。在你走之前，我想用一分钟时间给你介绍我最喜欢的产品，因为我经常被问到我的技术栈里有什么。Framer 是我搭建网站的工具。Genway 是我做研究的方式。Granola 是我在评审时做笔记的工具。Jitter 是我给设计做动画的工具。Lovable 是我用代码实现想法的方式。Mabin 是我寻找设计灵感的地方。Paper 是我像创意人一样设计的方式。而 Raycast 是我每一步的快捷键。这些公司都是我精心挑选的，以便我能全职做这些节目。所以，目前支持本节目最重要的方式就是去了解一下它们。你可以在 dive.club/partners 找到完整列表。

Pleasure as always. Before I let you go, I want to take just one minute to run you through my favorite products because I'm constantly asked what's in my stack. Framer is how I build websites. Genway is how I do research. Granola is how I take notes during crit. Jitter is how I animate my designs. Lovable is how I build my ideas in code. Mabin is how I find design inspiration. Paper is how I design like a creative. And Raycast is my shortcut every step of the way. Now, I've hand-selected these companies so that I can do these episodes full-time. So, by far, the number one way to support the show is to check them out. You can find the full list at dive.club/partners.