AI 时代重新思考设计流程

好，想象一下这个场景。季度初，你的团队需要决定做什么，你说：“太好了，我完全知道该怎么做，我知道每一步该怎么做。”于是你去做用户研究，把研究结果做成一个用户画像。这个画像很完美，有人口统计数据，有假名字，还有你生成的照片。基于这个用户画像，你又画出一条完美的用户旅程地图，展示这个人使用你产品的每一步、他们遇到的所有问题、经历的所有情绪变化。然后作为设计师，你说：“好，我们去组织一场头脑风暴，把跨职能的伙伴都叫来。”在那里，你们头脑风暴所有能帮助达成 OKR 的问题。然后你写出完美的问题陈述，可能以“我们如何能”开头，或者用这种格式：“作为一个（角色），我想要（需求），以便（目标）”，这样你就能真正理解用户的目标和问题。然后你说，好，又该头脑风暴了，更多便利贴，但这次是头脑风暴解决方案而不是问题。接着你选一个方案，说，好，我去画线框图，先画低保真的，然后再稍微高保真一点，像这样。然后又该做研究了，你拿着方案去了解用户有哪些问题，然后他们再把方案做得更高保真，看起来很棒，像这样。万岁，你刚刚完成了这个完美的设计流程，解决了公司所有的问题，赚了几百万美元，创造了一个人们真正喜爱的产品。

Okay, imagine this. It is the start of a quarter. Your team has to figure out what to do and you're like, 'Cool. I know exactly what to do. I know the steps to follow.' So, you go do some user research. You go and take that user research and you make this persona. It's perfect. It's got demographic data and it's got a fake name and a photo that you've generated. And from that user persona, you make this perfect user journey map and it shows every step of this person's journey throughout your product, all the problems they have, all the emotional sequences that they're going through. And then you as a designer, you're like, 'Okay, cool. We're going to go lead a brainstorm and bring all your cross-functional partners in.' And there you're going to brainstorm all the problems that you can solve to help meet your OKRs. And then you write the perfect problem statement. Maybe it starts with 'how might we' or maybe it's actually in this format: 'As a blank, I want to blank so that I can blank' so that way you can really understand the user's goals and problems. And then you're like, okay, it's time for a brainstorm again. A lot more sticky notes, except this time you're brainstorming solutions instead of problems. And then you pick a solution. And you're like, cool, I'm going to go do wireframes. I'm going to do them really low-fi first. And then I'm going to go a little higher fidelity, maybe like this. And then it's time for research again. You're going to take the solutions. You're going to understand what problems there are from users. And then they're gonna make them higher fidelity and they're going to look great like this. And hooray, you just did this perfect design process. You solved all the company's problems. You made millions of dollars. You created a product that people really love.

对，对，是的。我是说……还有，嗨，我叫 Jenny Wen。我通常住在纽约布鲁克林，但我其实来自加拿大多伦多地区。工作上，我是 Anthropic 的设计负责人，负责核心的 claude.ai AI 产品，这是一个基于云端基础模型的聊天机器人。我是今年一月加入的。在那之前，我在 Figma 工作了五年半，担任设计总监。我领导团队，负责我们增长团队中的几个新产品，比如 FigJam、Slides 社区、Figma Buzz、网站 CMS 以及我们的增长团队。我也是 FigJam 的第一位设计师，我把它带到了发布，并扩展了那个团队。在那之前，我在 Dropbox 工作了几年，是 Dropbox Paper 的设计师。那是一个文档编辑器，不知道你还记不记得。我这周刚收到邮件说他们要关闭移动应用了，安息吧。它是最早挑战 Google Docs 规范的文档编辑器之一，让文档编辑变得非常愉悦。

Right. Right. Yeah. I am calling... Also, hi, my name is Jenny Wen. I'm normally based out of Brooklyn, New York, but I'm actually from the Toronto area in Canada. Work-wise, I'm a design lead at Anthropic where I work on the core claude.ai AI product which is a chatbot built on top of the cloud foundational models. I joined earlier this year in January. But before that I was at Figma for five and a half years where I was director of design. I led teams at a high level that led a handful of new products in our growth team. So FigJam, slides community, Figma Buzz, site CMS and our growth teams. And I was also the first designer on FigJam and I took it to launch and grew that team as well. And before that I was at Dropbox for a few years where I was a designer on Dropbox Paper. It's a document editor. I don't know if you remember it. I just got an email this week that they're shutting down the mobile apps. Rest in peace. It was just one of the first document editors that challenged the norms of Google Docs and made it really sort of pleasant to use.

好，我们说到哪儿了？你知道，你遵循了这个设计流程，解决了世界和平，做出了漂亮的产品，赚了很多钱，对吧？好，这就是你放 DJ 搓碟音效的地方。那听起来一点都不真实，对吧？但与此同时，这就是过去几年每个人都在宣扬的流程，学校也是这么教的，也许有些人甚至为此拿了 MBA。但我们现在真的需要对这个环境做一个现实检验。首先要考虑的是 AI。我不知道，也许你听说过，但它代表人工智能，以防你真的不知道。基本上，工具在变化。AI 让我们更有能力。有一堆新工具冒出来。我觉得每天都有新工具出现，我都 overwhelmed 了，不知道要试哪些。因此带来的变化是，基本上一个 PM 能比你写出完美问题陈述或组织头脑风暴更快地做出一个可用的原型。他们不用做任何研究、不用做用户画像等等。我在几家公司都看到这种情况，实际上现在期望产品经理去 vibe coding 或做原型。这过去只是设计师的责任，但现在变成了越来越多职能的责任。但与此同时，这也让设计师更强大。你知道，不再是设计师应该会编程，而是设计师可以编程。我们不再只是做静态图片，我们可以轻松做原型，还可以自己实现到完美。所以现实检验是，这个流程过时了。设计流程对今天的工具和技术来说真的过时了，越来越不适合当下。

Okay, so where were we? You know, you'd followed this design process. You solved world peace and you made this beautiful product and you made so much money, right? Okay, so this is where you have that DJ scratch sound effect. That did not sound very realistic at all, right? But at the same time, that's the process that everyone has been preaching for the past few years and has been taught to us in schools and maybe some people even have like an MBA in this process. But we actually need to do a reality check of this environment that we're in right now. So, first thing to consider is AI. I don't know, maybe you've heard of it, but it stands for artificial intelligence in case you actually weren't aware. Basically what's happening is tools are changing. AI is making us more capable. There are a bunch of new tools popping up. I feel like there's new tools that come up every day and I'm overwhelmed. I don't know all the tools to try. And the thing that's changing because of this is basically a PM can get to a working prototype faster than you can write this perfect problem statement or write a brainstorm. They do this without doing any research, making any personas, etc. And I've seen this at a few companies now where actually it's the expectation that product managers for example are vibe coding or prototyping. And this is something that just used to be the designer's responsibility but now becomes the responsibility of more and more functions. But at the same time it's also making designers more powerful too. You know it's no longer should designers code but designers can code. We don't have to just make static pictures anymore. We can prototype really easily and we can implement to fit and finish all by ourselves. So the reality check here is that it's outdated. The design process is really outdated for today's tools and tech. It's feeling less and less right for this moment.

我经常听到的另一件事是“用更少做更多”。我认为过去几年这已经很普遍了，我从独立贡献者设计师和行业领导者那里都听到了。有很多裁员和减员。也许你也受到了影响，无论是作为领导者还是设计师。我认为人们只是意识到拥有更小的团队是可能的，有时甚至更有效。协调开销更少。我认为小团队其实非常强大。但与此同时，我们的角色在演变。你知道，我们是设计师，我们有这些现有的职责。但由于 AI 之类的东西，我们被期望承担越来越多的角色，而且这实际上更可能，比如更像 PM 和更有战略性，或者也实现和原型化我们的工作。我认为同样的事情也发生在我们周围的所有其他职能上。所以在这个世界里，我们还有时间遵循这个僵化的流程吗？我们能从中削减什么来为这些现在被期望做的事情腾出空间？我认为好事是，我不认为我们需要所有这些步骤。所以现实是我们的角色在改变，我们被期望用更少做更多，我们再也没有时间一丝不苟地遵循流程了。

Another thing I'm hearing a lot is do less with more. I think this one's been pretty universal for the past few years and I'm hearing about this both from IC designers as well as leaders across the industry. There have been a lot of layoffs and headcount reductions. Maybe you've been impacted by it too, either as a leader or a designer. And I think people are just realizing that it's possible and sometimes even more effective to have smaller teams as well. It's less coordination overhead. And I think small teams are really powerful actually. But at the same time, our roles are evolving. You know, we were designers. We had these existing responsibilities that fell within our bucket. But because of things like AI, we are expected to span more and more parts of our roles and that's actually more possible like being more PM-shaped and strategic or also implementing and prototyping our work. And I think the exact same thing is also happening to all these other functions around us. So in this world, do we even have time to follow this rigid process? Like what could we be cutting from it to make room for all these other things that we're now expected to do? I think the good thing is like I don't think we need all these steps. And so the reality is our roles are changing. We're expected to do more with less and we just don't have time to follow a process to a tea anymore.

我越来越多地听到这些词，你知道，品味、工艺和质量。我们越来越频繁地抛出这些词，并在设计流程的语境中谈论它们。我认为由此得出的一个假设是，在一个你可以用 AI 开始制作任何东西的世界里，真正重要的是你选择和策划你所制作的东西的能力。我认为这是一个好假设。比如，如果你能一次提示就生成这样的东西，这对设计师意味着什么？这其实不是很好，但它提高了门槛，也提高了设计师被期望做到的下限和基线。你的工作必须比这更好，你才有价值，才可能被雇佣。

I'm hearing more and more of these words, you know, taste and craft and quality. We're throwing these around more and more and talking about them in terms of the design process. I think the running hypothesis that comes from this is that in a world where you can start to make anything with AI, what really matters is your ability to choose and curate what you make. I think that's a good hypothesis. Like if you can one-shot prompt something that looks like this, what does that mean for designers? Like this is not really that good, but it does raise the bar and it raises the floor and the baseline of what a designer's expected to do. Like your work has to be better than this for you to be valuable and somebody that could be hired.

我还注意到，我们开始倾向于使用这类应用，比如那些“不无聊的软件”应用。这个是他们的一款相机应用，它就是个相机应用，但很漂亮，有各种触感旋钮和音效，用起来很愉悦。这感觉不像你随便用几个提示词就能做出来的。或者像 Linear 这样的应用，它是一个任务追踪应用。我们都用过不同的任务追踪应用，它们多得很，但 Linear 做得很精致，速度快，用起来很灵敏。人们喜欢用它，并且愿意付费。或者像 Notion Calendar 这样的应用，功能上其实差别不大，但感觉工艺很高，值得付费。我认为这些应用是“反 AI 垃圾”的。它们让人觉得付费有价值，用起来很享受。当我跟这些应用背后的设计师以及我认识的最具工艺水平的设计师交流时，他们并不是用某种固定流程来做工作的，更不用说设计流程了。

I've also noticed that we're starting to gravitate towards apps like this, like the not boring software apps. This one is their camera app, which is just a camera app but it's beautiful and it has all these tactile knobs and sound effects and it just feels delightful to use. This doesn't feel like you could just make it with a few prompts, or apps like Linear, which is a task tracking app. We've all used different task tracking apps. They're a dime a dozen, but it's well-crafted. It's fast. It's snappy to use. People like using it and prefer to pay for it. Or apps like Notion Calendar, which again, it's not that different functionality-wise, but it feels high craft and worth paying for. I think these apps are the anti-AI slop. They feel valuable to pay for. They're enjoyable to use. And when I talked to the designers behind these apps and the highest craft designers I know, they didn't make their work with any set process, let alone the design process.

所以我认为现实是，设计流程中并没有鼓励卓越工艺的步骤。它就不是适合这个的流程，而且它也从来没有以一种能帮助你做出卓越工艺的方式来补偿你。最后，每个项目都如此不同，而且我们正在做的项目范围也在变化。设计流程就像是在向你推销某种东西，比如一种格式，一种可以让你崇拜的逐步流程，它说每天你这样做，它甚至向你推销整个学位来学习设计流程的做法。但我想，最终发生在我们身上的是，它让人们崇拜这个流程，把大量时间花在这些流程产物上，而不是最终结果。我在作品集里总是看到这种情况，80% 都是流程产物，你能看出他们花了大量时间在这些东西上，然后最后只有一屏，就是最终体验。设计流程教我们按照这些步骤，以非常僵化的顺序，带着这些箭头。但我认为现实是，用户根本不在乎你做的流程产物，或者你是否做出了完美的用户旅程。他们在乎的是他们感受到和看到的最终体验。而且每个项目都如此不同。你知道，每个项目都有不同的利益相关者、复杂性、问题空间、未知因素、技术约束、时间线、业务需求、人员配置等等。不可能同一套步骤适用于所有这些不同的公司。你不可能用这种方式做出伟大的作品。它就是不能产生更好的作品。而且你不可能对每个项目都重复这个流程。但奇怪的是，行业内的情绪一直是：相信流程，遵循步骤，有一个严谨的设计流程，你就做得很好。好吧，我要在这里说，不要相信流程。它真的没什么道理，尤其是在这个新时代。如果这是你热爱、信任、依赖或被工作要求必须遵循的流程，我很抱歉，但老实说，这是我的政策。这会让人不舒服，但我认为，为了帮助我们引领这个新时代，这种不适是必要的。

So I think the reality is there aren't steps in the design process that encourage great craft. It's just not the right process for this, and it never really compensated in a way that helped you do great craft. And finally, every project is just so different, and the range of projects that we're working on is changing too. The design process was like selling you something like this, like a format, a step-by-step thing where you can worship, and it said every day you do this and it'll even sold you entire degrees in ways to do the design process. But then I think what ended up happening to us is that it made people worship this process and spend a lot of time on these process artifacts, not the end result. I see this in portfolios all the time where there's like 80% process artifacts and you can tell they spent so much time on these things and then there's one screen at the very end that is just like here's the end experience. The design process taught us to follow these steps in this very rigid order with these arrows. But I think the reality is the user doesn't give a damn about the process artifacts you made or whether you made the perfect user journey. They care about the end experience that they're feeling and seeing. And every project is just so different. You know, every project has different stakeholders, complexity, problem space, unknowns, technical constraints, timelines, business needs, staffing, etc. There's just no way that the same set of steps works across all of these different companies. You just can't produce great work this way. It just doesn't produce better work. And you can't repeat this process for every single project. But oddly enough, the sentiment in the industry has been, trust the process, follow the steps, have a rigorous design process, and you're doing your job really well. Well, I'm here to say, don't trust the process. It just doesn't really make sense, especially in this new era. And I'm sorry if this is a process you love, trust, rely on, or is mandated by your job, but honestly, it's my policy here. This is going to be uncomfortable, but I think discomfort is what is warranted to help us shepherd in this new era.

不过，我明白我们是怎么走到这一步的。你知道，在 2000 年代初，我们在做这类工作，非常华丽，但我们正从平面设计和网页设计转向设计完整体验的应用，以及人们用手机时需要考虑周围环境的东西。我们被视为那些进来把东西弄漂亮的人。但有一个整个运动，你知道，设计需要有一席之地。我们需要合法性。我们还需要为更复杂的体验进行设计。我们不再只是设计简单的、一步步的网页，而是在为复杂的移动应用进行设计。所以，设计思维和以用户为中心的设计理念普遍爆发了。我认为这些都是好事，因为我们不仅考虑事物的外观，还考虑它们的感觉、它们如何运作、它们如何影响人们的日常生活以及对业务的影响。但正因为如此，我认为我们大大倾斜了平衡。你知道，我们过分强调策略而忽视工艺，以至于我们的整体工艺受损。所以现在我认为我们需要重新平衡一下，确保我们的工作兼顾策略和工艺。但我们也必须解开这些现在出现在我们世界里的新因素的一团乱麻。比如设计门槛降低、AI 带来的指数级变化、角色转变、人手减少的工艺、以及新工具。哦，幻灯片掉了。但这些变化可能比过去 10 到 15 年更快。所以为了我们自己，我们不能再相信旧流程了。

I get how we got here though. You know, in the early 2000s, we were doing this sort of work, which was gorgeous, but we were pivoting from graphic design and web design into designing fully experienced apps and things where people were on their mobile phones and had to consider all the environment around them. And we were just seen as these people who just come in and make things pretty. But there's this whole movement, you know, design needed to see at the table. We needed to have legitimacy. We needed also to design for more complex experiences. And we weren't just designing a simple, step-by-step web page anymore, but we were designing for complex mobile apps. So then there's this explosion in the idea of design thinking and user-centric design generally. I think all good things because we considered not just how things looked but how they felt, how they worked, how they could impact the day-to-day of someone's lives and their impact on the business as well. But because of that, I think we tilted the balances a lot. You know, where we prioritize strategy way over craft to the extent that our overall craft suffered. And so now I think we need to retip the balances a little bit more, make sure our job balances both strategy and craft. But we also have to untangle this hairball of mess of these new factors that are in our world right now. Things like lower barriers to design, all this exponential change from AI, role shifting, less headcount craft, and new tools. Oh, there goes the slide. But these things are changing maybe faster than they have been in the last 10 to 15 years. So for our sake, we cannot trust the old process anymore.

但即使在这之前，我想我已经不信任这个流程了。部分原因是，在过去几年里，我在行业中看到的设计师的作品，让我越来越怀疑这个流程能否创造出伟大的作品。大多数时候，当我想到我最自豪或最敬畏的作品时，它们通常不是用这个流程或类似的东西做出来的。我见过的最好的作品是从解决方案出发，而不是从问题出发。实际上，是一个每个人都倾向于并让每个人都惊叹的解决方案。当一个团队非常在意，花大量时间迭代细节，努力把它们做对时。或者当一个设计师有强烈的直觉或信念时。也许这不是每个人一开始就同意的，或者不会直接出现在用户研究研究中。当一个设计师跳过步骤，或者为了得到他们需要的东西而编造步骤，或者当一个团队做某事只是为了让人微笑时。不是因为它在人物角色或用户旅程中出现，而是因为他们想让人微笑。所以，让我们从这个开始，稍微分解一下。从解决方案出发。出于某种原因，在设计流程中，这感觉非常不合法，你就是不能这样做，否则你就是个糟糕的设计师。但一个例子是，在 Anthropic 的团队，这在我之前，当他们创建 Claude Artifacts 时，Artifacts 基本上是这个右侧面板，向你展示交互式代码。基本上，这个团队是第一个真正创建这种模式的 AI 团队，但现在它被用于所有这些不同的 AI 应用。它基本上即时生成代码，让你就在那里与之交互。

But even before this, I think I was already not trusting this process. And that's partly because in the last few years, the work of designers that I've seen in the industry has just grown increasingly skeptical of the ability of the process to create great work. Most of the time when I think about the work that I'm proudest of or most in awe of, they're usually not using this process or anything close to it. The best work that I've seen is starting from solution first, not a problem. Actually, a solution that everyone gravitates towards and makes everybody say, 'Wow.' It's when a team cares so much they spend a lot of time iterating on the details and just trying to get them right. Or when a designer has strong intuition or conviction. Maybe it's not something that everyone agrees with off the bat or doesn't show up directly in a user research study. It's when a designer skips steps or when they make them up to get what they need, or when a team does something to just make people smile. Not because it came up in a persona or a user journey, but because they wanted to make people smile. So, let's start with this one and break it down a bit. Starting from solution first. For some reason with the design process, this felt super illegal and you just couldn't do it or you're a bad designer. But one example is when the team at Anthropic, this predated me, when they created Claude Artifacts, where artifacts are basically this right panel that is showing you the interactive code. Basically, this was the team was the first kind of AI team to actually create this pattern, but it's used across all these different AI apps now. And it basically generates code instantly and lets you interact with it right there.

这是团队里一位研究员做的原型。看起来不怎么样，但基本上，当 Claude 生成代码时，它会在旁边创建这个交互式工件。看起来真的很有意思。但团队里一位设计师 Michael 看到后说：“我喜欢这个。”然后他迭代了一下，做了自己的原型。整个团队的反响非常积极，大家都觉得这里有点东西。它不是来自产品问题陈述。但当我们发布它时，它非常成功。正是这个东西终于让人们意识到，AI 不只是写代码，而是创建我可以使用的交互式应用。我认为它从根本上改变了人们对 AI 的看法。如果我们一开始从问题陈述出发，可能永远也到不了这一步。我们直到真正看到解决方案，才知道这是一个值得解决的问题。

This was a prototype that one of the researchers on the team built. It didn't look awesome, but basically, when Claude was generating code, it would create this interactive artifact on the side. It was really interesting to see. But one of the designers on the team, Michael, saw this and said, 'I love this,' and he iterated on it and built his own prototype. The reactions across the team were so positive, and they just felt like there was something there. It didn't come from a product problem statement. But when we launched it, it was really successful. It was this thing that finally made people realize that AI is not just writing code, but it's creating these interactive applications that I can use. I think it fundamentally changed the way that people perceived AI. If we had started with a problem statement first, we might never have gotten here. We didn't know it was a problem worth solving until we actually saw the solution.

这过去主要发生在拥有优秀原型师的团队里，但有了这些新工具，我觉得它开始随时发生了。而且不只是设计师。它开始成为一种常态。但在旧的设计流程里，这感觉有点不合规矩，对吧？

This used to happen mostly on teams with really good prototypers, but with all these new tools, I think it's starting to happen all the time. And not just with designers either. It's starting to be the expectation. But this feels sort of illegal in the old design process, right?

但我觉得我们现在越来越多地看到的是，尤其是如果你在开发 AI 或基于 AI 的产品——我认为随着时间推移，我们中越来越多的人会这样做——我们有了这种新技术。它和我们见过的任何东西都截然不同。它是新的。我们大致知道它擅长什么。但正因为它是新的，我们有很多以前不用 AI 就解决的问题。我们基本上必须审视所有这些问题，然后说：“嘿，我们如何把这项新技术反向应用到所有这些问题上？我们如何用不同的方式解决这些问题？”我认为这与上一个时代有根本性的不同，因为那时技术相对静态。我们没有任何东西能如此改变游戏规则。正因为如此，我们不得不寻找要解决的问题，而不是先有解决方案。在 Anthropic，我们基本上每次模型变得更好时都会这样做。我们看看它能做什么、擅长什么，然后必须弄清楚它能解决哪些问题来帮助人们。

But I think what we're seeing more and more now, especially if you're working with AI or products built on top of AI, which I think more and more of us probably are over time, is that we have this new technology. It's drastically different from anything we've ever seen. It's new. We sort of know what it's good at. But since it's new, we have all these existing problems that we've solved before without AI. And we basically have to look at all of them and say, 'Hey, how do we take this new technology and apply it backwards to all these problems? How can we solve these problems in a different way?' I think this is fundamentally different from the last era because the technology was relatively static. We didn't have anything that changed the game this much. And because of that, we had to look for problems to solve instead of solutions to have. At Anthropic, we basically do this every time the model gets better. We see what it's capable of and what it's good at, and we have to figure out what problems it could solve to help people.

那么，举个例子，在过去一年左右，Claude 模型在这些方面变得更好了。它有了更大的上下文窗口，更好的工具使用能力，能够编排智能体，能够规划并进行多步推理。所有这些能力使得你们今年早些时候能够推出一项研究功能，Claude 基本上可以一次花 10 到 20 分钟，写出一份完整的研究报告。这有点神奇，对吧？

So, for example, in the last year or so, the Claude model has gotten better at a bunch of these things. It's gotten larger context windows, better ability to use tools, it's able to orchestrate agents, it's able to plan and do multi-step reasoning. And all those capabilities made it possible for you to launch a research feature earlier this year where Claude can basically go off for 10, 20 minutes at a time and write a full-on research report. That's kind of a marvel, right?

它基本上是在部署这些子智能体来做不同部分的工作，放慢速度思考要写什么，而不是像以前那样，AI 只是瞬间吐出一个答案，而不怎么思考。但这都是因为技术先让我们达到了这个地步。所以，不按顺序做事是可以的，尤其是在这个技术变化比以往任何时候都快的 AI 时代。我认为它改变了我们需要使用的流程。

It's basically deploying these sub-agents to do different parts of this, slowing down to think about what to write, as opposed to before when AI would just spit out an answer instantaneously and not think that hard about it. But this is all because the technology got us there first. So it's okay to do things out of order, especially in this AI era where the technology is changing faster than ever. I think it changes the process that we need to use.

我看到的另一种做出优秀作品的方式是，团队对细节极其在意。你能举个例子吗？

Another way that I've seen really great work made is when the team just cares ruthlessly about the details. Can you give an example?

比如，我们在 2021 年推出了 FigJam，它运行得相当不错。但发布之后，我们不只是给它添加新功能。我们继续迭代，提高质量。所以在接下来的几年里，我们花了很大一部分时间只是迭代细节，确保产品的核心机制真正感觉良好。我们迭代了吸附和对齐。我们改变了画布上选择边框的显示方式，让它不那么杂乱。我们迭代了很多颜色，确保我们得到了正确的范围，颜色不太刺眼。我们添加了很多方法来改变字体大小并调整它们。我们给图片加了边框，让它们在画布上更突出。我们迭代了很多关于输入时形状如何溢出的问题。我们调整和优化了很多工具栏交互，还有更多。我可能光讲我们随时间做的迭代就能填满整个演讲时段。

So, for example, we launched FigJam in 2021, and it worked pretty well. But after the launch, we weren't just adding new features to it. We continued to iterate on it and make the quality better. So we spent a large percentage of the time in the years that followed just iterating on details, making sure the core mechanics of the product actually felt good. We iterated on snapping and alignments. We changed the way we showed selection borders on canvas to feel less noisy. We iterated on the colors a bunch, just to make sure we got the right range and the colors weren't too loud. We added a bunch of ways to change the font size and adjust them. We gave images a border so they'd stand out on canvas. We iterated a bunch on how shapes would overflow as you were typing. We tweaked and refined a bunch of the toolbar interactions, and so many more. I could probably fill a full talk slot just talking about the iterations we made over time.

但实际的过程并不是标准的设计流程，对吧？它更像是一个漫长的质量尾巴和永无止境的迭代。

But the actual process for this was not the standard design process, right? It was more like a long tail of quality and iterations endlessly.

没错。这不是双钻模型真正能容纳的东西。我们总是在尝试学习和迭代，让事情变得更好。这需要时间，你知道，需要集中的时间。设计流程并没有为此留出空间。这是你必须关心并为之分配时间的事情。

Exactly. It's not something that the double diamond thing really makes room for. We're always just trying to learn and iterate and make things better. They take time, you know, they take concerted time. They're not accommodated for in the design process. It's just something that you have to care about and you have to allot time for.

我看到的另一种做出优秀作品的方式就是凭直觉行事。我觉得这是我最喜欢的方式之一，因为我喜欢直觉。但我认为直觉对设计师来说就像一个脏词。好像你不够以用户为中心，你不在乎，你不够以数据为中心。如果你只是用直觉，那就不合法。但直觉不是猜测，对吧？

Another way that I've seen great work made is just operating on intuition. I think this is one of my favorites because I love intuition. But I think intuition is like a dirty word for designers. It's like you're not being user-centric enough, you don't care, you're not data-centric enough. That's not legit if you're just using your intuition. But intuition is not guessing, right?

对。我觉得人们认为直觉就像感觉或猜测，或者谁最有说服力，但我实际上认为直觉非常有价值，而且非常罕见。这是你必须培养的东西。我认为直觉是快速做出合理判断的能力。它是长时间深入思考或收集证据的捷径，因为你非常深入、专业地了解这个领域。我认为拥有出色的直觉是值得尊重和真正向往的。我不认为我们应该轻视它。我认为我们应该努力培养它。

Right. I think people think it's like vibes or guessing or whoever has the best argument, but I actually think intuition is incredibly valuable and it's actually incredibly rare. It's something you have to build. I think intuition is the ability to make reasoned judgments quickly. It's a shortcut to thinking really long and hard or having to gather evidence because you know the subject area so deeply and in such an expert way. I think having great intuition is something to be respected and to really aspire to. I don't think we should be looking down on it. I think we should be trying to build it.

那你如何培养直觉呢？

So how do you build intuition?

我认为这些是我一直在尝试培养直觉的一些方法。我就是通过不断阅读关于产品的反馈来培养它，无论是在 Twitter、Reddit 还是其他任何能找到产品反馈的地方。有时是内部反馈板。我觉得销售团队可能也会收集很多反馈，所以和他们谈谈。我确保定期参加用户研究会议。我会回看旧的会议。我会阅读旧报告。我也会参加那些与我负责的产品部分无关的用户研究会议，因为我想更多地了解我们的用户群。我定期查看仪表板。我只是看看整体使用和活动的趋势。

I think these are some of the ways that I have been trying to build mine. I just build it by constantly reading feedback about the product, whether that's on Twitter or Reddit or wherever you find feedback about your product. It's sometimes internal feedback boards. I think your sales team might be collecting it a lot too, so talk to them. I make sure I go to user research sessions regularly. I watch back old ones. I read old reports. I will go to user research sessions that are not about my part of the product just because I'm learning more about our user base. I'm looking at dashboards regularly. I just see what the trends are on overall usage and activity.

我还在借鉴一些认知偏见和设计原则，这些是从学校教科书和心理学里学到的，因为我觉得这些东西能帮你预测人们的行为和整体决策。但我觉得我在这里真正做的是构建我对用户和最终产品的内部世界模型。这样我就不用对每个决定都做研究或 A/B 测试。我试图做出有根据的猜测来快速行动。但我也认为这能带来更好的设计工作，因为你现在可以把这种直觉不仅应用于大的产品决策，也应用于小的决策，比如怎么把这些按钮分组，或者在登录页上显示什么。因为直觉不是猜测。建立强大的直觉是让你成为优秀设计师的关键，因为你会用这种直觉来应对各种规模和尺度的决策，比如选择正确的设计模式来捍卫设计，或者做出快速选择。这是你无法真正用钱买到或让 AI 生成的东西。

I'm also drawing on things like cognitive biases and design principles that you learn in a textbook from school and psychology, because I think these things help you predict how people behave and make decisions overall. But what I'm doing here, I think, is really building my internal model of the world of our users and our end product. So that way I don't have to research every single decision or A/B test every single decision. I'm trying to make educated guesses to move quickly. But also, I think it creates better design work because you can now apply this intuition not just to big product decisions but also to smaller ones, like how do I group these buttons together or what do I show on this login page. Because intuition is not guessing. Building a great intuition is what is going to make you a great designer, because you're going to wield this intuition for decisions of all sizes and scales, like picking the right design pattern for defending a design or making a snap choice. That's something you can't really pay for or get AI to generate.

我见过的另一种做出优秀作品的方式是跳过步骤，甚至自创步骤。这也是我的最爱之一。你可能知道 Google Ventures 的设计冲刺。它相当死板，他们非常坚持每一步都很重要。你不能跳过，因为每一步都有存在的理由。我跑过很多次这个冲刺，但说实话，大多数时候我都觉得这些步骤对我来说并不重要，感觉没什么用。而且我觉得尤其是在初创公司，一整周的时间太长了——在初创公司里就像一年。所以多年来，我跑过很多不同的变体，混合步骤、删除步骤等等。这是我最喜欢的方式之一，就是把大部分时间花在原型制作和头脑风暴上，因为我发现原型是你在冲刺中产出的东西，人们真正看到它，真正理解概念。所以我也把它压缩到三天而不是五天。我觉得这对我和我所在的团队是有效的，但我每次都尝试调整，也许其他方式对你和你的团队更有效。

The next way I've seen great work made is skipping steps and making them up. Also a favorite of mine. So you probably know the Google Ventures design sprint. It's pretty rigid, and they're really adamant about how every step really matters. You can't skip them; there are reasons they're there. And I've run this sprint a bunch of times, but honestly, most of the time I'm like, these steps don't really matter to me right now, and they don't feel useful. And I think especially if you're at a startup, a whole week is such a long time—that's like a year in startup time. So over the years, I've run a bunch of different variants of it where I mix the steps, remove them, etc. And this is one of my favorite ways to do it, where I just spend most of the time prototyping and ideating, because I find that prototypes are the thing that, when you produce them out of the sprint, people really see them and really understand the concept. So I also do it in three days as opposed to five. And I think this is the thing that's worked for me and the teams I've been on, but I try to play with it every time, and maybe something else works better for you and your team.

另一个流程是亚马逊的逆向工作法，你写一份新闻稿，想象记者会怎么写你的功能。我喜欢这个目标，但在 Figma 使用时感觉太静态，和我们不相关。所以我想，怎么让它对我们真正有用？我觉得 Figma 的等价物是推文，因为我们经常和用户交流，通过那里沟通。所以我为头脑风暴做了一个组件，不是写新闻稿，而是让他们想象，嘿，人们会在 Twitter 上对这个功能有什么反应？这个组件实际上用于我们构建的一堆大型 jam 功能，比如光标聊天、握手、表情和印章。你可以在这里看到这些功能的最初草图。当我们真正发布产品时，我们看到了类似这些的推文，这真的很酷，因为这意味着我们构建的功能确实唤起了我们想要的情感。

Another process thing is the working backwards Amazon thing, where you write a press release to imagine how a reporter would write about your feature. And I like the goal of this, but it felt static and not relevant to us when we were using this at Figma. So I thought, how do I actually make this work for us? And so I think the Figma equivalent is the tweet, you know, because we always talked to our users a bunch and communicated with them through there. And so I made this component one time for a brainstorm where instead of a press release, I got them to imagine, hey, how would someone react on Twitter to this feature? And this is the component in use actually for a bunch of big jam features that we built, like cursor chat, handshakes, emotes, and stamps. And you can actually see the first sketches of those features here. And when we actually did launch the product, we saw tweets that resembled these, which was really cool to see because it meant that these features that we built actually evoked the emotions that we wanted them to.

我还尝试过类似的事情，写假落地页标题，想象我们如何打包功能并头脑风暴这些功能。这是对新闻稿的另一种改编，感觉更快，对我们在 Figma 更相关。我喜欢每次都试着弄清楚什么步骤对我的团队最有意义，以获得最佳结果。这对我来说其实很奇怪地有趣，因为我有点把现有的设计流程看作宜家家具的说明书，有点难懂。而且就是同样的步骤。你每次都跟着做，应该会得到结果，对吧？但问题是，当你设计东西时，情况有点不同。你实际上不知道最终结果是什么。你不知道它会是这个比利书架。它也可能是一把椅子、一盏灯，甚至一个热狗。你只是开始这个过程，你实际上不知道你最终要得到什么，不像宜家家具。所以如果你不知道你在构建什么，你怎么知道步骤是什么，对吧？你怎么能用同样的步骤每次都得到那个结果？所以每次你都必须既绘制地图，又定义你在构建什么。这看起来很难，对吧？没有手册。所以这里又是那个脏词“直觉”发挥作用的地方。你必须在这个新世界中使用直觉的另一部分来知道如何完成一个项目。没有人会给你一本能解决所有问题的手册。所以作为设计师，你的价值在于磨练那个流程以获得最佳结果，而不是遵循一个流程。如果任何人都能遵循流程，那么任何人都能做。

And something similar I also tried doing is writing and making fake landing page headlines to imagine how we might package features together and brainstorm those features. This is another riff on the press release that felt faster and more relevant to us at Figma. I love just trying to figure out every time what steps make the most sense for my team and to get the best results. It's actually weirdly fun to me, because I sort of think about the existing design process as like a set of instructions that you get from IKEA furniture that's kind of hard to decipher. And it's just the same steps. You follow them every time and you're supposed to get an outcome, right? But the thing is, when you're designing something, it's kind of different. Like, you actually don't know what the end outcome is. You don't know if it's going to be this Billy bookshelf. It could also be a chair or a lamp or even a hot dog. You just start the process and you actually don't know what you're trying to get at the very end, unlike IKEA furniture. So if you don't know what you're building, how do you know what the steps are, right? How can you use the same steps to get to that result every time? So every time you have to both chart the map and define what the thing is that you're building. That seems hard, right? There's no manual. So here's where that dirty word intuition comes into play again. You have to use another part of your intuition in this new world to know how to get through a project. No one's going to hand you this manual that's going to solve all your problems. So your value as a designer is in honing that process to get to the best results, not following one. If anyone could follow a process, then anyone could do it.

最后，我见过的最后一种做出优秀作品的方式是仅仅为了让人微笑而做某事。如果你玩过 FigJam 里的这些功能，印章、表情、光标聊天，它们是我职业生涯中构建的最喜欢的东西。但老实说，我们制作它们的过程的一部分就是在真实代码中做原型，在我们的内部构建中试用。我们只是出现在人们的会议上，请他们试用。这些是人们在会议上试用原型的真实录音。即使在我们做原型时，也有很多可用性问题和错误。但我们注意到人们在微笑、大笑，他们从会议开始就转变了。所以我们知道那里有好的东西，我们必须继续。最终当我们发布这些功能时，它们为 FigJam 上的许多其他功能设定了先例，也在 Figma 整体上。目标变成了只是让人微笑，我们只是添加了令人愉悦的细节。那是那里每个人都最引以为豪的一些工作。我认为这些不是来自问题陈述。过程中没有为它们设置的步骤。它们只是来自一个非常关心的设计师和工程师团队。

Finally, the last way I've seen great work made is doing something just for the sake of making people smile. So if you've ever played with these features in FigJam, the stamps, the emotes, the cursor chat, they're some of my favorite things that I've ever built in my career. But part of our process for them honestly was just prototyping them in real code, trying them out in our internal builds. And we just showed up to people's meetings and asked them to try them out. Like these are actual recordings of people trying them out in meetings with the prototypes. And even as we were prototyping them, there were a ton of usability issues and things wrong with them. But we noticed that people were smiling, laughing, and they transformed from the beginning of the meeting. And so we knew there was something good there and that we had to keep going. And ultimately when we launched these features, they set the precedence for a ton of other features on FigJam, but also at Figma generally. The goal became to just make people smile, and we just added details that would be delightful. And that was some of the work that everyone was most proud of there. I think these didn't come from a problem statement. There's no step in the process for them. They just came from a team of designers and engineers that cared very deeply.

所以我提到了一个清单。每个人都喜欢总结幻灯片。但不像设计流程，这些不是硬性规则。你可能不应该在任何事情中全部做。你不需要按顺序做。它们甚至不保证好作品。而且可能还有很多我遗漏的其他事情。

So there's a list of things I mentioned. Everyone loves a summary slide. But unlike the design process, these aren't hard and fast rules. You shouldn't probably do all of them in anything. You don't need to do them in order. They don't guarantee good work even. And there are probably also a ton of other things that I've missed.

就像你们可能在团队里做过其他很棒的事情，创造了我这里没有的出色作品。所以我希望我已经向你们展示了这句老话：相信过程。我认为在这个新时代，它已经过时了。我们真的不能再相信过程了。就我个人而言，我从不认为我们真的能相信它。但在这个时代，我们比以往任何时候都不能相信它。

Like you've probably done great other things on your teams that have created great work that I don't have on the here. So I hope I've shown you this old adage like trust the process. I think that's dead in this new era. We can't really trust the process anymore. And personally, I don't ever think we really could have trusted it. But in this era, more than ever, we can't.

我们的角色在改变。我们使用的工具在改变，但同时，强加给我们的设计过程却是一样的。你知道，我们的价值不在于重复一个过程。如果是这样，任何人都能成为设计师。我们现在正在重写这个过程。这相当可怕，因为我们没有先例可循。但我不了解你，对我来说这有点有趣。

Our roles are changing. The tools that we're using are changing, but at the same time, the design process that's dictated to us is the same. You know, our value isn't in repeating a process. If it was, anybody could be a designer. We're rewriting the process right now. This is like pretty scary because we don't have precedent to follow. But I don't know about you, it's like sort of fun for me.

我们需要每次都想清楚我们在构建什么，以及到达那里的地图。但我们的力量在于运用正确的工具，混合搭配，创造出色的结果，无论我们如何到达那里。所以我们不能相信过程。如果我们想在这个时代生存并茁壮成长，我们必须做一些更可怕、更大胆，但同时也更令人兴奋的事情。我认为我们必须再次相信自己，作为设计师、实践者，作为有经验和宝贵技能的人。

We need to both figure out what we're building and the map to getting there every single time. But our power is in wielding the right tools in mixing and matching and creating great results no matter how we got there. So we can't trust the process. If we want to both survive and thrive through this era, we have to do something scarier, bolder, but at the same time more exhilarating. Think we have to trust ourselves again as designers, as practitioners, as people with experience and a valuable skill set.

我们必须相信自己，尝试新事物，运用正确的工具，去感受并知道我们何时设计或构建了伟大的东西。这是我们现在能做的最赋权的事情，但也可能是最脆弱的事情之一。是时候了。谢谢。

We have to trust ourselves to try new things, to wield the right tools, to feel and know when we've designed something or built something great. It's the most empowering, but also probably one of the most vulnerable things we can do right now. It's time. Thank you.