Nev Flynn：将艺术表达带入企业级设计

我们已经逐渐远离了过去二三十年里经典的用户界面。有一天，研究团队会训练出一个新的音乐模型，我们就会拥有前所未有的控制音乐的新能力。而你的工作，本质上就是尝试发明一个用户界面，以最连贯的方式来控制它。

We've kind of moved away from classical interfaces that we had for the last 20 or 30 years. One day the research team will train a new music model and we'll have new capabilities for controlling music that we haven't seen before. And your job is to try and invent essentially a UI to control that in the most coherent way possible.

欢迎来到 Dive Club。我是 Rid，这里是设计师永不停歇的学习之地。本周的嘉宾是 11 Labs 的设计负责人 Nev Flynn，我们将深入探讨他如何在当今的设计格局中通过设计实现差异化。他带我们幕后了解了他们所有的内部品牌工具，并详细介绍了他们如何创建一个系统，为所有独特的声音生成超过 4000 万个光球着色器。不用说，这一集充满了艺术表达和创造力。而一个重大的突破是 Nev 最近转向招聘更多专业设计师。

Welcome to Dive Club. My name is Rid and this is where designers never stop learning. This week's episode is with 11 Labs head of design, Nev Flynn, and we're going to do a deep dive into how he thinks about differentiating with design in today's landscape. He gives us a behind the scenes of all their internal brand tools and walks us through how they created a system to generate over 40 million orb shaders for all of their unique voices. Needless to say, there is a ton of artistic expression and creativity in this episode. And a big unlock is Nev's recent shift toward hiring more specialist designers.

我觉得一开始，你知道，其实并没有什么设计的存在。早期的员工必须是通才。比如受过传统训练的品牌设计师和产品设计师。虽然我们仍在招聘这类人才，但过去 6 个月里最有趣的事情是让更多专家加入团队。举几个过去 6 个月的例子，我们聘请了一位不可思议的声音设计师，他负责我们动态设计影片的所有声音工作。他基本上就像一位艺术家和作曲家，他会展示预览和视频站会，我听的时候会起鸡皮疙瘩。就像它是如此复杂和不可思议。还聘请了一位我认为是世界上最好的 WebGL 专业人士之一，这非常非常有趣。我经常说，如果你和他以创意总监和 WebGL 专家的组合方式合作，那么技术上或创意上几乎没有什么是做不到的。所以你可以说，比如一个 3D 场景，像森林，里面有一个光球，它在发光，他会说，是的，我们可以做到。所以是非常有趣的组合，这也让我们能够设计和构建一些非常重要和不可思议的品牌时刻，这些时刻成为了巨大的差异化因素。

I think at the beginning, you know, there wasn't really any design presence. The early hires had to be generalists. So like classically trained brand designers and product designers. And while we're still hiring those, what has been super fun in the last 6 months has been bringing more specialists onto the team. To name a few of the last 6 months, we hired an incredible sound designer who does all of the sound work for our motion design films. And he is basically like an artist and like a composer and he will, you know, show previews and like video standups and I'll get goosebumps listening to them. Like it's so intricate and incredible. hired a who I think is one of the best WebGL professionals in the world which has been so so fun. I often say like if you work with him in like a kind of a creative director WebGL person pairing there's like zero technical or creative limit to what can be done. So you could just be like you know what about like a 3D scene and like forest there's an orb in it and it's glowing and he'll be like yes you know we we can do that. So super fun pairing and that has also allowed us to design and build some like really important and incredible brand moments which serve as these huge differentiators.

作为产品设计师，让自己看起来出色的最简单方法就是有一个专家，能创造出真正美丽的东西，直接放进去，比如一张卡片或一个空状态之类的。这样你所有的作品看起来都会好上 10 倍。

It's the easiest way to look good as a product designer is to have a specialist that can create something really beautiful to just drop in as like a card or an empty state or something. It's like all of your work looks 10 times better.

当然。当然。不，那太棒了。你知道，有很多非常有趣的项目，如果没有他们，我们根本无法完成。而且，在开始招聘这类人时，一开始会有一种紧张感，你能从房间里感觉到，无论是否说出来，人们可能认为我们应该再招一个通才，因为这是显而易见的，但我认为值得游说，值得为这类人争取，因为你可以做出奇妙的事情。

For sure. For sure. No, that's been amazing. And you know there's been so many super fun projects which we just wouldn't have been able to do otherwise. And with these as well, you know, there's a tension at the beginning of of of starting to hire folks like this where you can kind of feel it from the room that whether it's said or unsaid that people maybe think we should hire another generalist, you know, because it's obvious, but I think worth lobbying and worth making the case for these types of people because you can do just you can do marvelous things.

我认为这也变得越来越重要。这很奇怪，因为当 AI 刚出现时，通才具有一种力量。我认为这非常正确，因为 AI 可以推断出你拥有的任何一点技能。然而，每个人的底线都在迅速提高。就像，是的，我觉得真正的设计差异化只存在于专家身上，因为你必须有人能够将其推得比模型帮助其他人达到的开箱即用水平远得多。

I think it's becoming more important too. It's weird because when AI first was on the scene, there is a power to the generalist. And I think that is very much so true because AI can kind of extrapolate any little kernel of a skill set that you have. And yet the floor is raising for everybody so quickly. It's like, yeah, I feel like true design differentiation only exists in the specialist because you got to have people that can just push it so much further than the models are going to help everybody else get to out of the box.

我也这么认为。而且我认为 WebGL 绝对是其中之一。这就像我个人一直以来的兴趣，但也是这样一种情况：因为他加入并做出了这些不可思议的事情，我完全失去了涉足 WebGL 的动力，就像和比你强无限倍的人打网球，突然你就不想打了。但我同意。我认为这是那些看起来非常非常有帮助的领域之一，比如了解它的一切，能够信任你所做的事情，并让事情具有性能，这真是太棒了。

I think so. And I and I think WebGL is definitely is definitely one of these things. It's like always been an interest of mine personally and it's been one of these cases where I've kind of lost all motivation and dabbling in in WebGL because he's joined and done these incredible things like you know playing tennis against someone who's like infinitely better than you and suddenly you don't want us anymore. But I agree. I think that's one of the one of the places where it seems super super helpful to like know everything about it and to be able to trust what you're doing and for things to be performance and it's been it's been amazing.

快速说一条消息，然后我们继续。我想可以肯定地说，这是你早期使用 Paper 的最后机会，感觉每个人都在 Twitter 上谈论它。我的意思是，嘉宾们不断在这个节目中提起它。他们刚刚完成了大规模的 A 轮融资，Paper 很快就成为了那些用智能体交付产品的人的首选设计工具。它真的感觉是专门为这种新的工作方式而构建的，这也是我每天使用它的一个重要原因。所以，如果你还在沉睡，请访问 dive.club/paper。我想你很快就会意识到为什么这么多设计师正在转向它。

Real quick message and then we can jump back into it. I think it's safe to say that this is your last chance to be early on paper and it feels like everybody is talking about it on Twitter. I mean, guests are constantly bringing it up on this show. They just raised a massive series A and Paper's pretty quickly become the design tool of choice for people shipping with agents. It really feels like it was built specifically for this new way of working and that's a big reason why I use it every single day. So, if you're still sleeping, head to dive.club/paper. And I think you're really quickly going to realize why so many designers are making the switch.

所以，我现在正在开发一个新的 Dive 网站。让我告诉你，在 Framer 中拥有我需要的一切是一件大事。它现在不仅仅是一个网站构建器。它确实是一个完整的 AI 设计智能体。我可以在画布上探索想法，构建响应式页面，甚至为所有剧集设置我的 CMS，这非常有帮助。但这不仅仅是便利。Framer 还提高了我能创造的上限。比如，他们刚刚发布了一整套 3D 工具，让智能体可以处理透视、深度、原点，然后我可以直接在画布上微调属性，而无需来回提示。这只是我在 Framer 上构建 Dive 所有内容的另一个原因。所以，请访问 dive.club/framer 今天就开始创作。现在，进入本集。

So, I'm working on a new Dive website right now. And let me tell you, having everything I need within Framer is a pretty big deal. It's so much more than just a website builder now. It really is a full AI design agent. I can explore ideas on a canvas, build out responsive pages, even set up my CMS for all of the episodes, which is incredibly helpful. But it's not just about the convenience. Framer also raises the ceiling for what I can create. Like, they just released a whole set of 3D tooling so agents can work with perspective, depth, origin, and then I can fine-tune properties directly on the canvas without having to prompt back and forth. It's just another reason why I build absolutely everything for Dive on top of Framer. So, head to dive.club/framer to start creating today. Now, on to the episode.

我不骗你，其实我都没想过，但就在这之前，我在玩 11 Labs 的 API 做音效，因为现在产品中的声音设计有如此大的机会。我喜欢有思想的声音设计，而且模型非常擅长实现它。问题是，现在它们非常擅长实现它，每个人都得到完全相同的一套哔哔声，单独听起来不错，但然后你意识到它听起来像其他所有产品。所以我想，天哪，

I kid you not, actually, I didn't even think about it, but like right before this, what I was doing was playing with the 11 Labs API for sound effects because there's such an opportunity right now in sound design in products. I love I love thoughtful sound design and the models are really good at implementing it. The problem is now that they're really good at implementing it, everybody gets the exact same set of beep beep boops that sound good in isolation, but then you realize it sounds like every other product. So I was like, man,

我需要生成东西。我需要推动它。

I need to generate stuff. I need to push it.

而你会说，无意冒犯产品。你知道，产品很棒。我是通过我的智能体来使用它。

And you're like, no offense to the product. You know, it's the product's great. I'm wielding it through my agent.

然而你很快意识到我对真正好的声音设计一无所知。

And yet you pretty quickly realize that I know nothing about actual good sound design.

是的。和这些人一起工作的有趣之处之一是，你知道，我认为在我的位置上，做一个通才是合理的，但当你和真正擅长自己工作的人一起工作时，就像观看他们的工作流程和他们使用的软件，简直令人惊叹。他一直在研究 11 Labs 的声音品牌，这是一件迷人的事情。

Yeah. One of the fun things about working with these people is I, you know, I think in my position I it makes sense for me to be a generalist, but then you work with people who are really good at what they do and just like watching their workflows and the softwares that they use just like mind-blowing. And he's been working on 11 Labs Sonic branding, which is a fascinating thing.

嗯，我想我们对品牌塑造都很熟悉，但声音品牌化真的很有趣。你知道，Netflix 在开头有那种音效，那这些在动态设计影片中出现的时刻都是什么，为什么？所以他在分类、生成、创造和收集这一整套声音，它们都属于同一类别。

Um, I think we're all super familiar with just branding in general, but like Sonic branding is really interesting. What's the equivalent to, you know, Netflix has their like at the at the beginning things like what are all of those moments that feature in in motion design films and and why, you know, so he's kind of categorizing and generating and creating and sourcing this whole collection of sounds that all fall under this this kind of same category.

在 11 LA，我们需要针对不同的用户群体。11 Music 非常面向消费者，我们可以大胆创新。11 Agents 则更偏向企业。所以声音品牌化需要根据谁在另一端观看影片而调整和变化。但如果你花时间在 DAW 这类工具里，看到过为每个动态设计视频制作的复杂编曲，有大量不同的音轨，那就是这样。并不是拿一个生成的音轨当作背景音乐。它完全由成千上万个独立的声音片段组成，这很惊人。

And at 11 LA, there's a bunch of different user groups that we need to kind of target. So 11 music is extremely consumer and we can really push the boat out. 11 Agents leans a bit more enterprise. So that sonic branding needs to kind of like alter and shift depending on who's going to be listening to the film on the other side. But if you spend some time in in in things like Daws and you've seen like really intricate compositions with like a bunch of different tracks for each motion design video, that's what it is. It's not the case of taking one kind of generated track and using it as like background music. It's entirely composed out of like thousands of individual pieces of sound, which is amazing.

我认为聘请这位专家的一个重要理由是我们是 AI 音频专家。你知道，如果要做声音，那就应该是我们。但另一个非常有趣的人一起工作。

I think one of the big justifications for hiring this specialist is that we are AI audio specialists. You know, if if anybody is going to do sound, well, you know, it should be us. But another extremely fun person to work with.

所以，你提到你加入时设计团队还没有那么大的存在感，对吧？那时几乎像个原始的研究实验室。那么有没有例子可以展示你们如何摸索并推动品牌化，以及现在有了这些专家后，你们在尝试什么类型的东西，试图弄清楚 11 Labs 的设计差异化是什么样子。

So, you talked about how you joined design didn't have as much of a presence, right? It's like almost kind of raw research lab at that point. So are there examples that you can show that shine a light on how you all have figured out and pushed the branding and what types of things that you all are experimenting with now that you do have you know these specialists and you're trying to figure out what design differentiation looks like at 11 Labs.

所以，自从我加入以来，其中一个令人惊叹又有趣的项目是，当时我们做大量的文本转语音和声音克隆，现在做得更多，但讨论点总是围绕当我们处理一个声音时，11 Labs 上已经创建了 4000 万个声音。我们如何视觉化，是的，很多，所有这些声音要么是克隆的，要么是根据提示设计的，但声音本质上不是视觉的。所以，我们经常讨论的一件事是如何将声音可视化，因为如果你想想，我们在很多场景中都会谈到生成声音或克隆声音，无论是在户外广告牌上，还是在营销网站上，或者在声音库中，都需要某种方式来视觉化地代表一个声音。而不是随机地做，我们总是想利用声音嵌入，也就是声音背后的数据，是什么让克隆声音成为克隆声音，并利用这些数据来确定视觉输出。

So one of the one of the amazing and and fun ones I think since I had joined we're doing a huge amount of text to speech and voice cloning at the time and we do a lot more now but one of the discussion points was always around when we take a voice there's 40 million voices that have been created on 11 laps. How do we visually yeah it's a lot and all of those voices have either been cloned or they've been designed from a prompt but voice is obviously inherently nonvisual. So, one of the things that we always spoke about was like how can we how can we visualize a voice because if you think about it, we kind of we kind of need to, you know, we'll talk about generated voices or or cloned voices in so many contexts, whether it's on like a an out of home campaign on a billboard or whether it's on the marketing website or in a voice library needs some way to to kind of like visually represent a voice. And rather than doing that just randomly, we we always wanted to take the voice embedding, so like the data behind a voice, what makes a a cloned voice a cloned voice, and use that data to actually determine the the kind of like visual output.

所以我要逐步介绍的是我们的声音球体。你可能在网上见过一些，但这是我们的 WebGL 巫师 Darian 加入后做的第一个项目，事后看来，作为他的第一个项目，这非常雄心勃勃。这个过程是提取那些特征，比如每个声音的嵌入中的声音数据，看起来像这样，将它们转换成我们可以识别的声音特征或特质，然后生成一个完全独特的基础纹理，用于网络上的实时着色器，基本上做 4000 万次，并确保每一个不仅独特，而且当我们看它们时，几乎能在某种程度上听到那个声音会是什么样。所以这就是过程，很多都很有趣，比如调整参数，比如着色器的视觉参数以及基础纹理，找到既符合品牌又美观的东西。我们做了大量的数值微调，直到感觉非常对。

So what I'm going to step through here, um, is our voice orb. And and you may have seen some of these guys on the internet, but this was a the first project that Darian, who's our WebGL wizards, worked on when he joined, which is super ambitious project for his first project in hindsight, which is kind of to kind of step through the process here, extracting those, you know, those features, the the like voice data from voice embeddings for each voice, which which kind of look like this, converting those into things that we might recognize then as kind of like characteristics or traits of a voice, and then generating ing an entirely unique base texture to use as part of a real-time shader on the web and doing that basically 40 million times and ensuring that each of them are not just unique but when we look at them we can we can almost hear what that voice would be in in some capacity. So that was the process and you know lots of this was really fun just kind of like dialing in the parameters like the visual parameters of the shader as well as the base textures to find something that felt you know also really really on brand and like aesthetically aesthetically pleasing. So we did tons of we did tons of just like value nudging just until it felt really right.

你们只是在构建内部工具并分发它们，就像每个人都在玩滑块，跟我谈谈这个。

Were you just building internal tools and what distributing them and like everybody's just playing around with sliders like talk to me a little bit about that.

是的，实际上就是这样。我们创建了一个小的 WebGL 游乐场，有各种不同的基础纹理，然后你可以在像这样的调试器中暴露着色器的参数，如果我们发现喜欢的东西，就截图发给 Darian，他会嵌入它，然后我们反复运行这个过程，直到对输出非常满意。所以这真的非常细致，是极其技术性和极其创造性的美丽结合，身处这样的世界超级有趣。

Yeah, this this is actually what that was. We we created a little WebGL kind of playground with various different base textures and then you could expose the you know the parameters of the the shader in like a debugger like this and if we found something we we liked we'd just like take a screenshot and and and send it to Darian and he'd like embed that and then we we'd run that process over and over again until we were really happy with the output. So it was really really like meticulous and this beautiful kind of combination of something that was like extraordinarily technical but also extraordinarily creative which is super fun world to be in.

所以有一个基础图像，就像每个这些背后的基础纹理，在实时着色器之前。即使有了这些，我们还构建了自定义图像数据集，因为我们需要数百万个符合品牌的东西，我们经历了很多这样的周期。这些我认为不是最终的，但最终生成了大量这些，从中取小切片，用作基础纹理，然后在上面运行着色器。基本上是六层堆叠，赋予它那种超级复杂的外观和感觉。流体模拟、噪声和 FBM，然后是一个羽化的声波，那部分也很重要，它是音频反应性的。所以，如果你在声音库中预览其中一个声音，你希望它真的随着音频舞动和变化，然后是一个简单的胶片颗粒。

So there's a base image that's like the base texture behind each of these before the real-time shader stuff. And even with those built this like custom image data sets because we needed, you know, millions of them needed to be on brand and uh we ran through a bunch of cycles like these. These I don't think these are actually the final ones, but ended up generating tons of these, taking little slices out of them, using them as the base texture, and then running the shader on top. It's basically this stack of six things, which gives it that super intricate look and feel. fluid sims and noises and FBMs and then a feathered soundwave which and that bit was to be also important that it's like audio reactive. So, you know, if you preview one of these voices in the voice library, you want to actually like dance and change to the to the audio and then a simple film grain.

所以，帮我澄清一下。你谈到六样东西。显然，胶片颗粒，我明白。那很容易。其他每一个是单独的着色器还是不同的效果？给我分解一下。

So, clarify this here for me. So, you talk about like the six things. Obviously, the film grain, I get that. That makes it easy. Are each one of these other ones individual shaders or they like different effects? Like, break this down for me.

都在同一个着色器中，但它们基本上是应用于该着色器的不同通道，最终会涌现出非常复杂和有趣的东西。例如，你可以看到这个羽化的声波部分，就是外面波纹的部分。所以，那就是这个，是音频反应的部分。所以这里发生的是，你基本上有所有这些数学着色器技术。你将它们一层层叠加，然后输出是最终的着色器。这样做的原因就是为了让它非常有趣和复杂。

It's all in the same shader, but they're basically different different passes applied to that shader, which ultimately kind of bubble up and give you something that's really intricate and interesting. For instance, to give you an example, you can kind of see this like feathered soundwave piece faces, just the kind of bit that's rippling at the outsides. So that is, you know, that's this one and that's the piece that's like reactive to audio. So what's happening here is you've got basically all of these mathematical shader techniques. You're kind of layering them one on top of the other and then the output is the the final shader. And the reason you do that is just to make it really interesting and and intricate on.

那么你们生成的这些图像，这六个元素分别影响了什么？

So the images that you are generating, what elements of these six are they impacting?

图像基本上作为基础纹理存在，然后这六个元素都负责操纵那个图像。

So the image basically exists as the base texture and then all of these six are responsible for manipulating that image.

好的。它们都叠加在那上面，就像——好吧，我现在明白了。我明白了。

Okay. It's just they're all sitting on top of that's like the Okay, I get it now. I get it now.

没错。没错。

Exactly. Exactly.

谢谢你放慢速度讲解，我只是想确认一下我说对了，因为这真的太酷了。

I appreciate you slowing down and like I just wanted to make sure I I said it cuz it's really really cool.

这很迷人。然后你知道这就是输出。嗯，我们用这个工具来测试很多这样的输出。我们基本上把库里的所有声音都拉进来，想确保分布均匀，我们做了几轮调整，因为觉得某些音高的声音太多了，结果整个球体库都显得太暗了。所以在这个阶段我们也做了很多调整，当我们能够集体查看我提到的那些声音特质和特征时。我们主要抓住的是两个大的维度。一个是音高，另一个是表现力。如果你加更多维度，就会变得有点嘈杂，很难把三个或更多特质与你实际看到的视觉联系起来。音高通常控制明暗，这有点反直觉。你知道，着色器有多亮或多暗。然后表现力，本质上是声音的特征，即声音有多单调或其他。这控制着中心有多少细节、噪点和趣味。你知道，过一段时间，当你习惯看这些东西时，你就能大致判断，哦，那会是一个深沉且富有表现力的声音，或者那会是一个高音调且相当单调的声音。看到我们网站上看到的东西背后有这么多系统，真的很酷。

It's fascinating. And then you know here's the output. Um and we used this tool to kind of test a lot of these outputs. We basically had we pulled in all the voices from the library essentially and we wanted to make sure that there was kind of like an even distribution of things and we did a couple of passes at this because we felt that there were you know there were too many voices of a certain pitch and then as a result the entire library of orbs was feeling too dark. So we did a lot of tweaks at at this stage of the process as well when we were able to like view everything collectively of those voice traits and characteristics that I mentioned. These were kind of like the big two that we that we pulled on that we held on to. One being pitch and then the other being expressiveness. If you kind of add more than that, it starts to just get a bit noisy and it's it's a little bit harder to like associate like you know three or more traits with what you're actually looking at visually. So pitch generally controls the darkness which is kind of in intuitive. you know, how bright or or dark the shader is. And then expressiveness, which is essentially the characteristics of the voice, which is how monotone or otherwise the voice is. That kind of controls how much detail and noise and intrigue there is in the center of it. You know, after a while, when you kind of get used to looking at these things, you can you can kind of gauge like, okay, that's going to be a deep voice that's quite expressive, or that's going to be a high pitch voice that's quite monotone. It's really cool to see how much of a system exists behind the scenes of what we just see on the website, you know.

太神奇了。而且你们还构建了整个自定义工具栈。你知道，两年前，这个概念会非常陌生。比如我们将有数百万的输出，流程的每一部分我们都要构建自定义工具。太神奇了。

It's amazing. And you guys built your entire custom tool stack, too. You know, like two years ago, even this concept would have been so foreign. The idea of like we're going to have millions of output and every piece of the process we're going to build custom tooling for. It's amazing.

但是，你知道，其中一部分我不会深入讲，比如这个的推出。嗯，我觉得我讲不好，但有各种成本考虑，第一次我们考虑推出时，要花费大约五十万美元。嗯，所以有各种优化、性能优化、缓存和迁移等等，最终让我们以很小的成本完成了。所以，这是超级酷的项目。我认为特别酷的是，来自一家在企业领域也快速扩张的公司，看到你们在更具创意艺术性的投资上投入这么多，真的很迷人很酷。

But, you know, part of this which I won't get into is like the rollout of this. Um, I don't think I'd do it justice, but there's all kinds of like cost considerations where the first time we considered rolling this out, it was going to cost like half a million dollars to do that. Um, so it's like all kinds of optimizations, performance optimizations, cachings, and migrations and so on, which um, allowed us to do it for like like a tiny fraction of that in the end. So, it's super cool projects. I think it's particularly cool coming from a company who is rapidly expanding in the enterprise too. To see how much you are investing in this more creative artistic strain of investment is really fascinating and cool to see.

这对我们很重要，你知道，我认为从一开始就是，我认为它创造了一种设计文化，让人们乐在其中。我想我们团队中大多数人进入设计都是因为一些非常有趣的东西，我们经常看到一些早期公司做非常富有表现力和艺术性的事情，然后他们成长为更大的企业公司，可能失去了一些，但我真的不希望我们失去任何这些，我认为如果你能做到，就应该做。所以是的，我们做了各种创意艺术项目，这是从核心品牌的约束中解脱出来的美好休息。你还有其他例子要展示吗？

It's important to us, you know, and I think it has been from the start and I think it creates a design culture where people are kind of having fun. I think all of us got it like most of us on the team got into design with with things that were were really really fun and I think we oftentimes see really early stage companies that do incredible expressive artful things and they kind of grow into like bigger enterprise companies and they maybe lose a little bit of that but I never really want us to lose any of that and I think if you can then then you should so yeah we've done all kinds of creative artful projects which are a lovely break from working within the kind of constraints of the core brand. Do you have any other examples to show? Of

当然有。好的。这有点像是自选旅程，Rid，但嗯，这可能是最近的一个例子，这是我们和网页设计师一起做的，网页设计师很喜欢这个，还有 WebGL 的参与，但你知道，我们为 11 Labs 有一个相当紧凑的核心品牌，但我们总是并且经常允许自己走出那个框架去做其他事情。这对网页设计师或品牌设计师来说是一个很有趣的前景，因为这意味着每隔几个月就有机会为某个东西设计和构建新的身份，这很诱人。11 Lab 峰会就是一个例子。你大概能从这看出来，但我们构建了一个全新的身份，这只是一个占位页面。活动在 11 月 11 日，嗯，年底，我们总是这样做，但我们想围绕 11 的形式构建这个图形系统。嗯，我们之前实际上没有这样做过。我以为这对你来说太明显了，你会对我说出来，但那就是这个背后的灵感。那就是我们想要的技术。但我们也在很多不同的地方举办这个活动，我们想要品牌有一些本地特异性。所以，当我们在二月的伦敦举办时，想做类似的事情，但你知道，像标志性城市时刻可能有点太直白。嗯，比如自由女神像，但当你以不同的方式处理时，它又开始感觉有品味了。所以，很多艺术在其中。然后我想当我在谈峰会时，这不仅仅适用于数字品牌，所以这实际上是二月份在伦敦的 11 Lab 峰会。这里面有很多非常有趣的时刻，我们再次想要非常艺术化。所以有一个巨大的像 11 发光的巨石，很多人在前面拍照。这像一个音频反应装置。我们与一位名叫 Harry F. 的艺术家合作。基本上你坐在这个巨大的高鼓里。每个凳子上有一副耳机。你戴上它们，然后你周围的鼓会对一段

Course I do. Okay. This is a bit of a a choose your own journey here, Rid, but um this is probably the most recent example of this and this is something that we get with like web designers and and web designers love this and the WebGL involvement here as well, but you know, we have a a pretty tight core brand for 11 Labs, but we'll always and often allow ourselves to step outside of that for for other things. And and that's quite a fun prospect for a website designer or a brand designer because it means that every couple of months there's an opportunity to like design and build a new identity for something which is you know enticing. The 11 Lab Summit is an example of that. You know you can probably gauge it from this but we we built a whole new identity and this is just a holding page for it. The event is in on the 11th of the 11th in uh at the end of the year which is we always do that but we kind of wanted to build this graphical system around the kind of 11 forms. Um, we hadn't actually done that before. I thought that was going to be so obvious to you that you were going to say it to me, but that was the that was that was the inspiration behind this one. That was the technique that we wanted. But we also run this event in a bunch of different places and we wanted to have some kind of local specificity to the brand. So, when we run this in London in February, want to do something similar, but you know, like iconic city moments can be a little bit on the nose. um like Statue of Liberty, but when you treat it differently, it starts to feel kind of like tasteful again. So, lots of art in that. And then I guess while I'm on summits, this doesn't just apply to digital branding, but so this is actually an 11 Lab summit from February in London. And there's just lots of really fun moments in this where again we wanted to be really artful. So there was this huge like 11 glowing monolith that a bunch of people took photos in front of. This was like an audio reactive installation. We partnered with an artist named Harry F. And essentially you sat into this huge tall drum. There was a pair of headphones on each stool. You put them on and then the kind of the the drum around you kind of like reacted to a piece of

他创作的很酷的音乐。然后同一个活动，这像一个 1920 年代爵士酒廊灵感的房间，我们在场地建造的，角落里显然有一个点唱机，但点唱机由 11 音乐驱动。所以提示并播放。人们会

cool music that he had created. And then same event, this was like a 1920s jazz lounge inspired room that we that we built at the venue where there's obviously a jukebox in the corner, but the jukebox is powered by 11 music. So prompt and play. People would

这太美了，伙计。这太酷了。多么——这是我在节目中很少遇到这种类型的设计。我笑得合不拢嘴。真的很酷。

This is beautiful, man. This is so cool. What a This is I haven't had a lot of instances of this type of design on the show. And I'm I'm just smiling ear to ear. It's really cool.

我们也喜欢实体东西。嗯，我认为这是一个很好的改变，我们是的，我们非常重视我们的活动，这是一个 amazing 的机会，把一些你知道那些艺术时刻带入现实世界空间，总是感觉很好。

We love the physical stuff as well. It's um it's a nice change I think and we yeah we lean heavily into our into our events and it's like an amazing opportunity to carry some of those you know those artful moments into like real world spaces which always feels so good.

11 Lab 商店也是一个很酷的例子，这个项目分成两半：一半是设计服装系列本身，另一半是设计围绕它的体验。这也是智能体的能力。如果你想看的话，我可以共享我的标签页给你演示一下。但本质上，这些都是超级有创意、做起来非常有趣的项目。但我们希望购物体验本身既有趣又有品味，而且……

The 11 Lab store was also a cool example where a project in two halves really where there's the kind of like designing of the clothing collection itself and then designing of the experience around it. This is also agents powers. I could share my tab and give you a demo if you wanted to see that. But in essence, these were all super creative, really fun to work on. But wanted the actual experience of kind of like shopping to be as kind of like fun and tasteful and...

这是 Roman 吗？

Is this Roman?

Roman 参与了这里。是的。是的，Roman 参与了这里。你知道，仍然有一个网格作为更保守、可预测的备选方案，但我喜欢它。我喜欢这个圆形。

Roman was involved here. Yeah. Yeah, Roman was involved here. You know, there's still a grid for the more conservative, predictable fallback, but I like it. I like the circle.

所以，一堆东西，我是说黑客松网站。我觉得这是第一个，这是一个比较老的项目，但它是第一个允许我们伸展到核心品牌之外、纯粹为了好玩而做的例子，你知道，为什么不呢？那是一个全球黑客松。它在同一个周末在 30 个不同城市连续举行。显然是以开发者为中心，所以是面向开发者的品牌，这里还有我们的朋友 WebGL 和 3JS 场景，然后其余部分都包裹在这种复古风格的 CSS 黑客技术中，但是的，这个项目非常受欢迎，我们后来又举办了黑客松，非常成功，我非常喜欢它，而且你知道，这个网站在互联网上也获得了很多喜爱，它帮助了注册等等，甚至回到你之前关于招聘的想法，它就像一块吸引人才的磁铁。你知道，我觉得有很多关于如何证明这种艺术表达的讨论，但我会把这看作是一个也许是非常有才华的设计工程师可以在网络上创造奇迹的人，就像是的，这就是我想工作的那种地方，你偶尔可以探索一个完全随机的角落，做那些别人可能会看着说‘嗯，那不符合品牌’的事情，但就像谁在乎呢，这很有趣，你知道。

So, a bunch of things that I mean hackathon website. I think this was the first, this is kind of an older one, but it was like the first example of this type of allowing ourselves to stretch outside of the core brand and have fun just because, you know, why why wouldn't we? It was a global hackathon. It was consecutive in like all happening over the same weekend in 30 different cities. Obviously developer focused so kind of developer focused brand with another our friend WebGL and 3JS scene over here and then the rest of it just wrapped in this kind of retro looking CSS hackery of the past but yeah this went down really well we ended up running the hackathon again it was a big success I really liked it and you know the website as well got lots of love on the internet and it's helped with signups and so on even going back to your hiring thoughts earlier like what a magnet for talented people you know Like I think there's a lot of discussion around how do you justify this type of artful expression but I would see this as somebody who maybe is like a really you know talented design engineer can make magic happen on the web and it's like yeah this is the kind of place I would want to work where you just every once in a while get to go explore off in a totally random corner do things that maybe others would look at and be like well that's not on brand it's like who cares it's fun you know.

是的，百分之百。这也是我在招聘对话中会明确对人们说的话，因为我想听到这个。你知道，我想知道我有机会走出核心品牌并享受其中。很多这些都是关于人们作为设计师玩得开心、享受他们的工作，并因为有机会做一些稍微不同的事情而感到充满活力，仅仅因为你可以。

Yeah, 100%. And that is something that I'll explicitly say to folks in hiring conversations because I would like to hear that. You know, I'd like to know that there is an opportunity for me to step outside the core brand and enjoy myself. A lot of this is about people having fun as designers and enjoying their work and feeling energized by the opportunity to do something a little bit different just because you can.

哦，我简直不敢相信我们差点错过了这个。这就像一个愚人节例子。所以，你知道，这里有很多空间和权限来打破品牌。但这里的前提是，Elvin Labs 与 11 Labs 合作了。历史上，Elvin Labs 曾经无所不能，然后他们意识到 11 Labs 拥有令人难以置信的语音技术，他们无法竞争，所以他们能做的就是与我们合作。所以，这里的概念，有一些传说开始，基本上就是我刚才说的。然后实际的时刻是这个 3D 场景。我会暂停一下，但如果你能听到这个，你会得到的是一个智能体体验。所以这是一个 11 智能体体验，你在与这个神谕对话，你知道这个非常有趣的语音设计，神谕就像你可以想象的那样，但这里的场景实际上是一个高斯溅射。你可能在互联网上见过一些。这个令人惊叹的技术。所以这开始只是一个森林场景的静态图像，然后转换成一个高斯溅射，然后你会得到的一些东西是背景中的一些洞，你可以看穿。所以画了一些进去，这样它就像一个全 encapsulated 的 3D 场景。所以如果你在寻找一个打破品牌并享受工作的例子，我认为这差不多是……

Oh, I can't believe we nearly missed this one. This is like an April Fool's example. So, you know, lots of scope and remit to break outside the brand here. But the premise here, Elvin Labs has partnered with 11 Labs. Historically, Elvin Labs was all powerful and then they realized 11 Labs had incredible voice technology and they couldn't compete, so all they could do is partner with us. So, the concept here, there's a bit of a there's some some lore to begin with, which is essentially what I what I just spoke through. And then the actual moment is this this this 3D scene. I'll stop this for a moment, but if you could hear this, what you would have is it's an agent experience. So it's an 11 agents experience where you're speaking to this oracle and you know this like really fun voice design where the oracle is, as you can imagine, but the scene here is actually a gauian splat. You've probably seen some of those on the internet. This this amazing amazing technique. So this began as just like a static of this forest scene and then converted into a into a goian splat and then kind of some of the what you'll get with splats is some of the holes in the background you'll be able to see through. So kind of painted some of those in so that it was like a you know an all encapsulated kind of 3D scene. So if you're looking for an example of breaking outside the brand and having fun with your work I think this is about as...

是的，我认为这算一个。

Yeah, I think this counts.

是的，这算一个。这算一个。

Yeah, this counts. This counts.

好的。所以，让我们拉远一点，更广泛地谈谈设计组织。比如，你们必须如何运作才能为这种表达创造空间，甚至能够腾出时间让这种艺术投资变得合理？

All right. So, let's zoom out a second and talk about the design org more broadly. Like, how do you all have to operate in order to create space for this type of expression and being able to even make time to have this type of like artful investment be justified?

你知道，过去有例子，我们第一次做的时候，它非常受欢迎。比如黑客松网站就是一个有点叛逆的个人任务，我们以前从未真正做过，我花了一个硬核的周末冲刺，它非常受欢迎，获得了大量注册，一切都很棒。你需要这样做并让它成功，这样人们才能理解它值得做。所以是展示而不是说教，然后也要务实。这是我们工作的一小部分。你知道，大多数时候是更经典的产品设计工作、设计工程工作。既然是这样，我们认为绝对可以留出一定的空间给这种真正有创意的工作。最终，我们看到它效果很好。有时它更贴近核心。例如，如果我们做大型模型发布，模型发布对我们来说是一件大事，对任何 AI 公司来说都是一件大事，我们通常会倾向于为它创建一个相当复杂的身份。对我们来说会很有表现力，这绝对没问题。我们知道它有效，既然我们知道它有效，我们就有能力为它腾出空间。

You know, there's examples from the past where we did it for the first time and it was really well-received. So like the hackathon website for instance was a kind of a rogue solo mission where we had never really done this before and I sprinted on it for you know a hard weekend and it was really well-received and got tons of signups and everything was great. You need to do that and have it work in order for people to understand that it's worth doing. So kind of showing not telling and then also look we want to be kind of pragmatic about these things. This is a small portion of our work. You know, most of the time it's more classic product design work, design, engineering work. Since that is the case, it's you know, we think it's absolutely okay to leave a certain amount of room for this type of really creative work. Ultimately, we've seen it work really well. Sometimes it's a bit closer to home. So, for instance, if we're ever doing like a big model release, like model releases are a huge deal for us and they're a huge deal for any AI company, we'll usually lean into creating quite a sophisticated identity for that. will be quite expressive for us and and that's absolutely fine. We know that it works and given that we know that it works, we're kind of afforded to make space for it.

我们能不能把钟摆完全摆回到另一个我觉得有趣的话题，这个话题与你正在做的品牌工具有关，在某种程度上，你能够将专家扩展到他们自身之外。

Can we have the pendulum swing all the way back to this other topic that I find interesting that is related to the brand tools that you're doing in some ways where you're kind of able to scale a specialist beyond themselves.

是的。

Yeah.

并且至少对你们必须覆盖的表面区域有一些了解。我的意思是，这只是一个巨大巨大巨大的产品。提醒我一下，现在产品设计、设计工程团队实际上有多少人？

And having at least some familiarity with the amount of surface area that you all have to account for. I mean it's just a massive massive massive product. And remind me like how many people are actually on the like the product design design engineering team right now?

大概 10 人左右。

It's probably about 10.

是的。所以与你们必须覆盖的表面区域相比，这真的非常小。跟我们谈谈你们是如何做到这一点的，比如需要建立什么样的系统，才能让 10 个人覆盖像 11 Labs 这样规模的产品。

Yeah. So that's like really small compared to the amount of surface area that you have to to make happen. Talk to us a little bit about how you make that work like how what systems need to be put in place for 10 people to be able to account for a product the size of 11 Labs.

设计系统已经成为一个非常有趣的东西。我们是一个相当小的团队，但同时我们越来越多地看到工程师在没有设计输入的情况下使用编码智能体设计东西，这不仅发生在 11 Labs，而是到处都在发生。

The design system has become an extremely interesting thing. We are a pretty small group but also at the same time what we're increasingly seeing is engineers you know designing things without any design input using coding agents and this isn't just happening at 11 Labs is happening everywhere.

我认为作为一个小团队，我们绝对不想成为阻碍进步的人，而是要找到支持他们的方法，确保他们正在构建的任何东西，在没有设计输入的情况下，如何能以真正有效和高质量的方式实现？因为这将会发生并持续发生。而设计系统，我认为如果做得好，就是现代答案之一，这很有趣。它们在大约七八年前非常流行，然后有点没落，现在又流行起来，但原因不同。所以几个月前，我们开始真正大力投入这个，并且正在重做我们的整个设计系统。但思考的方式是，系统的消费者不再是设计师或前端工程师。而是由任何人操作的智能体。那么，你如何制作一个为智能体优化的设计系统？我们一直在思考的几种方式。第一，我们审计了旧系统，发现它非常庞大。非常复杂。它的变化和灵活性远超所需，即使作为设计原则也是如此。举个简单的例子，你的 UI 里有一个按钮，我们有七种尺寸变体，可能并不需要。这是那种情况，是的，我们可能需要所有这些。我们会提供所有范围，但在智能体优化领域，那只是多一个可能出错的变体。所以我认为如果你能将这个原则应用到整个系统，你可以让它范围更紧凑，这是好事。另一件事是关于文档化，当然要记录组件本身，还要记录这些组件应该使用的上下文，几乎就像把你的产品设计原则编码成一个智能体 MD 文件：何时使用这个组件而不是那个组件，针对某些 UI 模式的护栏，我们如何思考产品设计，并将其编码到一个智能体 MD 文件中，然后最终下一步将是应用各种技能等等，并通过以下方式衡量成功：在没有设计师参与的情况下，输出能否与有设计师参与的情况一样好？棘手的事情，但这就是目标。

I think as a small team, we definitely don't want to be the people that stand in the way of progress there, but rather find ways to support them and ensure that whatever it is that they're building, how can that happen in a really effective and high quality way without design input? Because it will happen and continue to happen. And the design system, I think, is one of the modern answers to that if done well, which is interesting. They were really popular 7-8 years ago and then they kind of dipped and now they're popular again, but for different reasons. So a few months ago we started really heavily leaning into this and are in the process of reworking our entire design system. But thinking about it in a way where the consumer of the system is no longer the designer or the front-end engineer. It's the agent being operated by anybody really. So how do you make a design system that's optimized for an agent? And a few ways that we've been thinking about that. One, we audited the old system and found that it was really big. It was really complicated. It had way more variance and flexibility than it needed to have even as a design principle. So a trivial example is you got a button in your UI and we've got like seven size variants of it and probably not needed. It's one of these things where yeah we'll probably need all of those. We'll give all the scope but in agent optimization land that's just like one more variant to get wrong. So I think if you can apply that principle to the whole system, you can make it much more tightly scoped which is a good thing. The other thing is around documenting it and documenting the components themselves of course but also the context in which those components should be used and almost like codifying your own product design principles in the form of an agent MD file: when to use this component instead of this component, guard rails against certain UI patterns, how we think about product design in general and codifying that in an agent MD file and then ultimately the next step will be applying that with various skills and so on and measuring success by: can this output where a designer hasn't been involved be as good as the case where the designer has been involved? Tricky thing but that's the goal.

你能多谈谈你必须走的那条钢丝吗？因为我确信有些听众觉得设计的输出只是一套规则，然后我们就不再需要设计师了，那么平衡在哪里，你如何思考你正在努力实现的理想状态？

Can you talk a little bit more about that tight rope that you kind of have to walk because I'm sure that there are some people listening that feel like the output of design is just a set of rules and then we don't need designers anymore and where is that balance and how do you think about the ideal state that you're working toward?

首先，这将继续不可避免地发生。没有哪个世界我们没有全栈工程师在没有设计师的情况下实施。所以无论如何都会发生。但我看到的积极方式是，很多工作是对事物的迭代更改。某个产品添加了一些小功能，当然我们可以让设计师从他们正在做的大零到一的事情中抽身，或者我们可以自己构建和设计，而且它的范围相当小等等。所以我认为它做的是不取代设计师或让设计师变得不那么有价值。它让每个人都能更快地行动，但同时保持非常高的标准，并防止设计师在上下文之间切换或专注于那些真正独特和新颖且在没有真正有才华的人输入的情况下难以完成的事情。

Firstly, this will continue to happen inevitably. There is no world where we don't have full stack engineers implementing without designers. So it's happening regardless. But the positive way that I see it is a lot of that work is iterative changes to things. There's some small feature that's added to some product and sure we could have pulled the designer off the big zero-to-one thing that they're working on or we could just build it and design it ourselves and it's reasonably small in terms of its scope and so on. So what I think it does is it doesn't replace the designer or make designers less valuable. It allows everybody to move faster but at the same time maintaining really high standards and prevents that designer from context switching around or focusing on the things that are really unique and novel and difficult to do without the input of a really talented person.

是的，这有道理。基本上就是能够将精力集中在产品中可能更重要的领域。不过还有一个后续问题，不是要过于二元化，但当我听到设计系统和智能体 MD 文件时，我的大脑会想到一个非常视觉化的地方。这是我们如何思考令牌。这是我们如何思考间距、理想的高度等。当一群非设计师使用它时，是的，也许他们制作的东西看起来和感觉像是 11 Labs 视觉语言的一部分，但从 UX 的角度来看，它们可能根本上是错误的。试图编码这类东西是不是一个陷阱？你是否在考虑纳入一些底层产品思维的方式？

Yeah, that makes sense. Just basically being able to concentrate efforts to areas of the product that maybe matter more. One more follow-up question though, which is like not to be overly binary, but when I hear design system and like an agent MD file, my brain goes to a very visual place. Here's how we think about tokens. Here's how we think about spacing, ideal like elevation, that type of thing. Where when a bunch of non-designers are wielding that, yeah, maybe they are making things that look and feel like they are a part of the 11 Labs visual language, but they might just be fundamentally wrong from slightly more of like a UX standpoint. Like is that a trap to even try to codify that type of stuff? Are you thinking about ways to incorporate some of the underlying product thinking?

是的。

Yeah.

你如何评估那个变量？

How do you evaluate that variable?

我觉得这是那种你不断调整的项目之一，你知道，你花同样多的时间，组件在那里，你迭代的是 MD 文件或编码产品设计、交互、UX、视觉设计原则，这很有挑战性。我确实认为在一定程度上是可能的，就像我们确实经历这个过程，即使几乎是潜意识地，作为一个设计很长时间的人，我们坐下来设计东西时，我们可能在大脑中运行一套潜意识规则，并落在我们认为合适的东西上。所以我认为在一定程度上可以编码这个，我认为这值得追求，如果我们知道我们不会阻止人们在没有设计师的情况下构建，那么我们能将质量门槛提高到 90% 吗？在我看来那仍然是成功。

I feel like this is one of these projects where you're just constantly tuning it, you know, and you're kind of spending as much time like the components are there and the thing that you're kind of iterating on is the MD file or codifying the product design, interaction, UX, visual design principles, which is challenging. I do think it must be possible to an extent like we do go through this process even if it's almost subconsciously as someone who's been designing for a long time and we sit down and we're designing something we are probably running through a set of subconscious rules in our head and landing on something that we think is appropriate. So I think it's possible to some extent to codify this and I think that's worth reaching for if we know that we're not going to put the brakes on people building without designers then can we get that quality bar to like 90%. That's still a success in my eyes.

不，我喜欢这个。每次我打开 Reddit，我看到 UX 设计子版块前四个帖子中至少有两个是人们在抱怨非设计师在原型设计和构建东西。

No, I like it. Every time I open up Reddit, I see at least two of the top four threads on the UX design subreddit are people complaining that non-designers are prototyping and building things.

就像那将会发生。它就会发生。问题不是，你知道，你应该对此有何感受？问题更像是你如何设置这种不可避免的行为，至少让它不那么糟糕。

And it's like that's going to happen. It's just going to happen. The question is not, you know, how should you feel about it? The question is more like how can you set up this behavior that is inevitable to at least not suck.

完全正确。你知道，我们在品牌设计方面有完全相同的平行情况，你知道，如果我可以转到那里，当 Claw Design 最初出来时，我想增长和上市团队的每个人都像是它来了，你知道，我们可以制作自己的演示文稿，我想我和品牌设计团队的其他人像是哦，天哪，我们能做什么，再次，我们不想成为阻止进步和阻止人们为他们的推介制作演示文稿的人，但我们确实如此，而且我们保持我们的质量和这些标准非常重要。所以我们花了很多时间，就像我们对设计系统所做的那样，围绕在那里做好工作构建基础设施，在 Claw Design 的情况下，尝试使用他们给你的设计系统，并真正深入其中，但成功相当有限。

Totally. And you know this is we have the exact same parallel on the brand design side of the house you know where if I can segue over there when claw design came out initially I think everybody on the growth and go to market team was like it's here you know we can make make our own decks and I think I and others on the brand design team were were like oh oh god what can we do to again we don't want to be the people that like stops the progress and prevents people from making the the deck for their pitch but we do and it's really important that we we we maintain our quality these standards. So we spent a bunch of time in the same kind of light as we did with the design system kind of building the infrastructure around doing a good job over there which was in Claw Design's case trying to work with the the the design system that they give you and really like going deep on that had pretty limited success there.

所以实际上我们最终转向了另一个方向，构建了另一个内部工具，用于制作演示文稿。演示文稿听起来可能很无聊，但我认为这是科技公司和大公司规模扩张中最大的未解决问题之一。它们无处不在。我认为我们用来制作演示文稿的软件已经过时了。作为大公司，我们非常根深蒂固，你知道，我们是一家谷歌公司，很难让人们从那个基础设施迁移到可能是一个全新的、非常紧凑的演示文稿产品，就像一些初创公司那样。所以，我们最终为此构建了一个工具，再次强调，我们现在正在将这些工具交到这些人手中，它将允许他们做他们来这里想做的事情，即快速制作演示文稿，并允许我们做我们想做的事情，即确保这些演示文稿的质量非常非常高。

So actually ended up moving in a different direction where we built another internal tool which is for decks. Which deck sounds boring. I think it's one of the biggest unsolved problems in scaling tech companies and big tech companies. They're everywhere. I think the software we use to put them together is old. I think we're very, as bigger companies, we're very ingrained in, you know, we're a Google company and it's really hard to move people off that infrastructure onto what might be, say, a really tight deck product that's brand new as some startup. So we ended up building a tool for this, which again, we're getting into the hands of these people now, and it'll allow them to do the thing that they came here wanting to do, which was make decks really quickly, and it allow us to do the thing that we want to do, which is ensure that the quality of those decks are really, really high.

有一个问题我无法停止问自己。如果公司申请与你交谈，而不是反过来，会怎样？这个问题是全新 Dive Talent Network 的基础。而且它正在发挥作用。就像现在，我正在帮助我认识的许多最令人兴奋的初创公司，聘请收听这个节目的设计师和构建者。所以，如果你好奇可能有什么机会，也许你想加入我的名单，或者你甚至正在寻找你的下一个设计雇员，请访问 dive.club/talent 今天加入。

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

这真是件迷人的事。这是我现在在节目中最喜欢的话题之一，即高质量内部工具的兴起，这些工具在科技的其他时代基本上被如此低估，而突然间它们变得如此重要，以及能够出现并带来出色的系统思维来弄清楚我们如何奠定有效基础，让 AI 能够以有凝聚力的方式赋予组织中的一群人权力的人的价值。就像突然间，这是一种如此有价值的技能组合，但我们并没有像我认为我们应该的那样多谈论它。我们甚至不把它放在作品集上。我们甚至不为此筛选，但这是我目前觉得如此迷人的特质。

Such a fascinating thing. This is one of my favorite topics on the show right now is the rise of really high-quality internal tools which have been so downplayed at basically every other era of tech and all of a sudden they're so important and the value of somebody that can show up and bring excellent systems thinking to figure out like how can we lay an effective foundation where AI can empower a bunch of people in the org in a cohesive way. Like all of a sudden that's such a valuable skill set, but we don't really talk about it as much as I think we should. We don't really even put it on portfolios. We don't really even screen for it, but it's like this trait that I find so fascinating right now.

我想雇佣一个只专注于这个的人。你知道，我们的许多内部工具是由团队的不同成员完成的，它在我们之间分布。但过去六个月里，我一直对那种生活和呼吸这些东西的人非常感兴趣。我认为那个人的特质，就像你说的，需要是一个真正的系统思考者，然后理想情况下是一个相当自给自足的设计工程师，设计能力与工程能力一样强，他们的工作就是利用所有这些令人难以置信的创意内部工具。我认为这将非常有价值。

And I would like to hire somebody who kind of only focuses on this. You know, a lot of our internal tooling we've done with various members of the team and it's kind of distributed among us. But for the last six months, I've been really interested in somebody who lives and breathes this stuff. And I think that maybe the traits of that person like you say are needs to be a real systems thinker for sure and then ideally a pretty self-sufficient design engineer where the design chops are as strong as the engineering chops and their job is to like rip on all of these incredible creative internal tools. I think it'd be so so valuable.

好的，那么在这个意义上定义设计工程师，比如他们必须看代码吗，还是只是那些非常擅长在云端构建的人？

Okay, define design engineer in this sense though like do they have to be looking at code at all or is it just somebody who's really comfortable building in cloud?

可以是两者之一。我能想到一些工具，看代码会很有用，而其他工具可能不需要。所以我们有很多这样的工具，其中一些是相当棘手的技术问题，比如 WebGL 输出，或者我们使用 WebGL 来获得静态输出或视频输出。所以那些相当棘手。然后还有其他更简单的。所以我认为可以是两者之一。我觉得有趣的是，这是一个令人向往的角色。我甚至在想，我最近在 Lovable 的网站上看到，他们有一个，我想他们称之为品牌和 AI 系统之类的。就像他们没有，只是把一堆关键词组合成职位名称，因为我们甚至不知道该怎么称呼这些人。但历史上我们会划出硬性分界。我们说这是左脑型的人，这是右脑型的人，而现在实际上能够将两者都带到桌面上的人是科技界最热门的角色之一。

Could be either. I can think of some tools where it would be useful to look at the code and others where maybe it isn't. So we have a bunch of these and some of them are pretty gnarly technical problems that are like WebGL outputs or we're using WebGL to get a static output or a video output. So those are pretty gnarly. And then there's others which are much simpler. So I think it could be either. I think it's interesting that it's like a role that is desirable. I'm even thinking I was looking on Lovable's website recently and they have like a I think they called it like brand and AI systems or something like that. Like they didn't it was just like a combination of a bunch of keywords the job title because we don't even know what to call these people. But historically we would have drawn a hard divide. We said like this is the left-brain kind of person and this is the right-brain kind of person and now actually the person that can bring both of those to the table is like one of the hottest roles in all of tech.

100%。是的。100%。其中一个，其中一个，你想看看这些工具吗？

100%. Yeah. 100%. One of these one of these Do you want to see some of these tools?

哦，请。所以这里有一些较小的工具，排版球体，我们最终使用了这个主题，因为你知道我们有这些语音球体。我们有时会以不同的方式使用这些球体的不同表示，这是其中之一。我们有一个巨大的户外广告活动，在旧金山机场的大型广告牌活动。很多广告牌上都是这个工具的直接导出，以视频格式。所以我们有屏幕，它们只是循环播放这些版本，像往常一样，所有这些工具都是超级参数化和可控的等等。所以我们从中获得了很多，这是不断增长的，有峰会网格。这是一个不断增长的列表，再次，你可以投入一些不同的背景图像，然后人们会使用它们用于不同的静态资产和社交媒体等等。这个是针对即将到来的项目品牌着色器。哦，是的，这就像是对我们正在研究的一些静态资产和感觉非常独特的资产的新尝试，因为它们无处不在。就像你身后墙上的三个渐变。嗯，但这是我们发现并喜欢并想使之成为我们自己的技术，你从基础图像开始，几乎像将其挤出屏幕。所以你可以看到这个与一堆不同的样本，你可以上传自己的，它变得非常非常有用，用于快速启动事物。

Oh please. So some of the smaller ones here, typography sphere, we ended up using this motif because you know we have these voice orbs. We kind of sometimes use different representations of those types of spheres in different ways and this is one of them. And we had a huge out of home campaign, big billboard campaign in San Francisco airports. And on a lot of those billboards were like direct exports from this tool in video format. So we had like um screens where they were just like rotating through versions of these and as usual with all of these tools everything is um super like parameterized and controllable and so on. So we get a lot out of these and it's this kind of constantly there's the summit grid. It's this constantly growing list where again with this one you can throw in some different background images and people will then use them for different static assets and socials and so on. This one was for an upcoming project brand shader. Oh yeah, this is like a new take on some we're kind of working on are our kind of static assets and assets that feel really unique to us because they feature everywhere. It's like the three gradients on your wall behind you there. Y um but this was like a technique that that we kind of found and loved and want to make our own where you kind of start with a base image and almost like extrude it off the screen. So you can see this with a bunch of different samples and you can upload your own and just becomes really really useful for for spinning things up quickly.

所以你实际上构建了一个内部主页，你有一堆工具，我想人们只是自己实验，然后你就像推广它们，比如这现在是官方工具包的一部分，或者你甚至如何考虑工具的组织和可发现性？

So you actually built like an internal home and you have a bunch of tools and I would imagine what people are kind of just experimenting on their own and then you're just like promoting them up to like this is now a part of the official toolkit or how do you even think about like the organization and discoverability of tools?

随着我们添加更多这些工具，我们可能需要更多地倾向于这一点。我认为这些只是其中的一些，你知道，你会看到一些标记，比如测试版、进行中和稳定版等等。但就像每周都有一个新的版本在飞行中，团队的不同成员正在研究它们。所以最终，我确信在几个月内会有 20 或 30 个这样的工具，我们将开始对它们进行排序和推广等等。然后这个，我刚才提到了，但这是一个更大的努力。所以就像我们刚刚看到的工具相当轻量级，可以服务于单一轻量级功能，但这是我之前提到的演示文稿工具，这有点迷人。你可能认为我们只是重建了 Google Slides，但嗯，我们没有。

We'll probably need to lean into that a bit more as we add more of these. And I and I think these are just the ones that, you know, you'll see some like tagging on these with like beta and work in progress and stable and so on. But there is like every week there is a new version of one of these in flight and different members of the team are are working on them. So eventually, I'm sure in in a few months time there'll be 20 or 30 of these and we'll start to like order them and kind of promote them and so on. And then this I mentioned this a moment ago but like much bigger effort this. So like the the tools that we just looked at are pretty lightweight can kind of like serve a single lightweight function but this is the deck tool that I was speaking about which is kind of fascinating. You might be thinking we've just rebuilt Google Slides but um we we haven't.

我们真正想要强制的约束是模板，而我们真正想要支持的是能够导出到 Google Slides，因为我们是一家使用 Google Slides 的公司，增长和上市团队都在那里协作，那是他们想用的工具。所以很难也不公平地说，嘿大家，我们都要用这个新产品，从现在起只用它。你需要同时支持两者。所以它的工作方式是，所有这些模板，你可以在这里看到，都是用代码定义的，这给了我们极大的控制权。我们最初在 Figma 中设计模板，然后将其代码化。所以一切都很严格，但在我们希望的方面也很灵活。比如我们拿这个来说，我们以非常非常受控的方式更改参数，这正是我们想要的。而如果这是 Google Slides，假设你发现了这个演示文稿，你意识到你不想要四个这样的，你想要五个，不可避免地，如果你不是设计师，那会非常棘手。

The constraint that we really wanted to enforce was the template, and the thing that we really wanted to support was the fact that we could get over to Google Slides, because we're a Google Slides company and that's where growth go-to-market happens. That's where they collaborate and that's the tool they want to use. So it's really hard and also unfair to say, hey everybody, we're all going to use this new product and that's all we're going to use from now on. You kind of need to support both. So the way this works is all of these templates, which you can see here, are defined in code, which gives us a huge amount of control over them. We designed the templates initially in Figma and then we codified them. So everything is quite rigid but also flexible in ways where we want them to be. So if we take this, we're changing parameters in a very, very controlled way, which is what we want. Whereas if this was Google Slides and say you found this deck and you realized you didn't want four of these, you wanted five of these, inevitably, if you're not a designer, that's going to be really tricky to do.

是的。尤其是在 Google Slides 里，对吧？

Yeah. Especially in Google Slides, right?

如果你是一名设计师，那会很痛苦。我认为在 Google Slides 中设计是设计师可能经历的最令人谦卑的事情。

And that's pain if you're a designer. I think designing in Google Slides is the most humbling thing that can possibly happen to a designer.

是的。是的。我喜欢每月做一次，只是为了让自己脚踏实地。

Yeah. Yeah. I like to do it once a month just to ground myself.

确保你的头不会变得太大。但我们这里的是，无论你需要怎么调整，它都会保持在一个非常符合品牌的世界里。类似地，这里的图片，我们想要建立的是我们知道有效的精选资产，对吧？所以现在我刚把它拖进来。再次，如果这是 Google Slides，你会移动一些东西。它可能不会居中。你知道，它留下了太多——基本上那个软件对非设计师的人来说控制太多了。所以这的美妙之处在于你可以从头开始，或者从提示开始。所以你可以有一个关于你想要创建的演示文稿的相当长的提示，或者你可以通过导入现有的 Google Slides 演示文稿来开始，这非常强大。所以想象一下，有一些销售演示文稿状态不佳。导入它，现在它状态很好，然后在这个过程结束时，你可以导出到 Google Slides。这使用了 Google Slides API，它相当有限。它对我们有效的原因是因为那个 API 基本上接受页面上所有内容的宽度、高度和 XY 坐标。所以我们最终做的是，我们查看这个——克隆我们的 HTML 克隆——我们测量页面上的所有内容，并想象每个东西都有宽度、高度、类型和坐标，然后我们把它放入 Google Slides API，我们就能得到这里内容的完美忠实再现。所以这就像在销售团队需要什么、我们作为品牌团队需要什么之间走钢丝，并以内部工具的形式实现。

Make sure your head's not getting too big. But what we have here is, regardless of how you need to play with it, it's going to stay in a very on-brand world. And similar with images here, the thing that we want to build up is curated assets that we know that they work, right? So now I've just dropped that in. Again, if this was Google Slides, you'd be moving something. It probably wouldn't be centered. There'd be, you know, it leaves too—there's basically too much control in that software for folks that aren't designers. And so the beautiful thing about this then is you can get started from scratch or you can get started from a prompt. So you could have a pretty long prompt about the deck that you want to create, or you can get started by importing an existing Google Slides deck, which is pretty powerful. So imagine there's some sales deck that isn't in amazing shape. Import it and now it's in amazing shape, and then at the end of that process you can export it to Google Slides. And this uses the Google Slides API, which is quite limited. The reason it works for us is because that API essentially accepts widths and heights and XY coordinates of everything that's on the page. So what we end up doing is we look at this—clone our HTML clone—and we measure everything on the page and imagine everything having widths, heights, types, and coordinates, and then we just pop that into the Google Slides API and we get a perfectly faithful recreation of what is in here. So it's just kind of like treading this line between what sales folks need, what we need as a brand team, and doing it in the form of an internal tool.

是的，这真的很酷。我喜欢这次对话的主题，一方面你展示了当你释放设计师并真正发挥创造力时天花板有多高，但另一方面还有所有在幕后发生的自动化，不仅是为了改进输出，这显然是，但我相信有很多设计请求历史上会交给设计师，但现在不再必要了，你可以通过在这些自动化层上投入更多来释放团队。

Yeah, this is really cool. I like the theme of this conversation where on one hand you've showed how high the ceiling is when you can just unleash designers and be really creative, but then also all of the automation that's happening behind the scenes in order to not only improve the output, which obviously it is, but I'm sure there are a lot of design requests that historically would have been placed on designers that are just not necessary anymore, where you can free up the team by investing more in these automation layers.

百分之百。我认为其中一些甚至更有趣，我想深入探讨，那就是在未来一个月左右，我们的品牌设计接收是在一个 Slack 频道中。今天有一个接收表单。发生的事情是有人会请求一个资产，假设为了讨论，它就像是一个用于派对填充或 Luma 活动的一对一封面图像，我们在 Figma 中有这些模板。今天会发生的是，它会进入接收，然后进入分诊，然后制作人将其分配给可用的品牌设计师，然后品牌设计师会去他们自己制作的模板，填入标题、地点和日期，然后他们在 Asana 上更新，然后利益相关者会查看它，说没问题，然后品牌设计师关闭工单。对于一个真正自动化的东西来说，这是一个相当长的过程。

100%. I think some of them are even more interesting that I'd like to get into, which is over the next month or so our intake for brand design is in a Slack channel. There's an intake form today. What happens is someone will request an asset and let's say for argument's sake it's like a one-by-one cover image for like a party fill or like a Luma event, and we have templates for these in Figma. What will happen today is that'll hit the intake, it'll go to a triage, and then the producer will assign it to an available brand designer, and then the brand designer will go to the template that they've made themselves, pop in the title, the location, and the date, and then they will update it back on the Asana, and then the stakeholder will look at it, say it's okay, and then the brand designer will close the ticket. It's a pretty long process for something that's really automated.

是的。

Yeah.

所以我想在接下来的两周内尝试做的是，我认为这很迷人，当那个工单被创建时，我们基本上解析接收和工单中的文本，识别出它是一个 Luma 封面，然后基本上命中一个绘图服务器，创建资产，在线程中回复资产，所以实际上没有任何东西到达制作人、分诊，然后回复一条消息，比如，这看起来好吗？如果他们说是的，那很好，那么它立即关闭工单。

So what I want to try and do over the next two weeks, which I think is fascinating, is when that ticket is created, we basically parse the intake and the text in the ticket, recognize that it is a Luma cover, hit like a drawing server essentially, create the asset, reply in the thread with the asset, so nothing has actually hit the producer, the triage, and then reply with a message which is like, does this look good? And if they say, yep, that's good, then it just closes the ticket immediately.

酷。我认为这可能发生在 30-40% 的品牌设计请求中，这太棒了，因为那是 30-40% 的品牌设计师时间，现在可以投入到真正令人兴奋的前瞻性品牌开发、新身份的开发，他们真正关心并真正想要工作的事情上。

Cool. And I think this could happen for like 30-40% of brand design requests, which is amazing because that's 30-40% of brand designer time that can now go into really exciting forward-looking development of the brand, development of a new identity, things that they really care about and actually want to work on.

对于正在收听并受到你所说的一切和你所建立的文化启发的人，他们可以在作品集或面试过程中做些什么，让你达到这样的程度：是的，我绝对想更多地了解那个人，或者把他们带到团队中，那种真正呼吸和生活这些东西的人，你知道吗？

For someone listening who is inspired by everything that you're saying and the culture that you've been able to establish, what's something that they could do either in their portfolio or maybe in like an interview process that would get you to get to the point where you're just like, yeah, like I definitely want to either learn more about that person or bring them on to the team that kind of live and breathe this stuff, you know?

设计团队中有很多人都是这样的，他们对自己所做的事情充满热情，你可以从他们身上汲取能量，这些人提升了整个设计团队的能量。所以你还可以看出，无论他们在职业生涯的哪个阶段，如果这种特质或特征存在，那么那个人很可能有很强的成长斜率。所以我认为这些就是要点。就作品集而言，是的，非常重要。我希望看到一些感觉完全独特的东西，理想情况下执行得非常好，然后性格方面，那些真正、真正热情、对自己所做的事情充满热情的人。

There's so many people on the design team that are like this, that are so ferociously passionate about what it is that they do, and you can really feed it off them, and those people lift the energy of the whole design team. So you can also tell that regardless of where they are in their career, if that kind of attribute or characteristic exists, then it's quite a high likelihood of there being quite a strong slope there for that person. So I think those would be the things. Like in terms of the portfolio, yes, really important. I would love to see something that feels like an entirely unique thought and ideally really well executed, and then personality-wise, folks that are really, really passionate, enthusiastic about what they do.

让我想起一个例子，我想是上周，我有一位招聘经理给我发了一个作品集，他们说我们想雇佣这样的人，那是一个非常简单的作品集。

Reminds me of an example from I think like last week I had a hiring manager send me a portfolio and they're like we want to hire somebody like this and it was a very simple portfolio.

基本上就是一页文字，然后真的只有一个好的案例研究。他们强调的甚至不是案例研究的内容，而是进入和退出那个案例研究的页面过渡交互。纸张滑过的方式真的很独特，悬停时还有那个小小的窥视效果，就这样。他们看到了那一点，就说这太酷了。就像，是的，作品主体、案例研究都达标了，但那不是突出的地方。真正突出的是那个非常非常独特、执行得很好的东西，那就是信号。所以听你这么说很有趣。

It was basically one page of text, and then there was really just one good case study. And what they highlighted wasn't even in the content of the case study. It was the page transition interaction from getting in and out of that case study. It was really unique how the paper slid over and you had this little peek effect on hover, and that was it. They saw that one thing and they're like, this is sick. Like, yeah, the body of the work, the case study, checks the box, but that wasn't what stood out. It was the one really, really unique, well-executed thing, and that was the signal. So it's interesting to hear you say that.

如果你想一想，当我回顾这些年来我参与的所有个人项目，比如过去 10 年，早期的那些相当糟糕，然后越接近现在，它们就越好。所以对我来说，把所有这些都放进作品集里并不合理。为什么不展示一两个近期的、我作为设计师成长并磨练技艺的项目呢？在我看来重要得多。而且你不想为此付出代价，因为我看了你一个 10 年前的项目。

And if you think about it, when I think about all of the personal projects that I've worked on over the years, over the last say 10 years, the early ones were quite bad and then as we get closer to present day, they get better. So it doesn't really make sense for me to have a portfolio that includes all those things. Why not show the one or two recent ones where I've developed as a designer and I've honed the craft? Much more important in my book. And you don't want to pay off because I've looked at a project of yours that's 10 years old.

是的。是平均值，不是总和。我喜欢这个说法。

Yeah. The average, not the sum. I like that.

那你现在对副业项目的重视程度如何？因为你说的很多内容，我认为可以在你拿报酬的工作之外展示，但显然，你知道，那也是非常不同的评分标准。比如，你如何看待与 W2 工作相关的信号，相对于他们在业余时间捣鼓的东西？

What about the weight that you're placing on side projects today? Because a lot of what you are saying, I think can be exhibited outside of work that you are paid to do and yet obviously, you know, that's a very different grading rubric too. Like how do you think about the signals as it relates to W2 work versus things that they're tinkering with on the side?

我认为这是一个好信号。我觉得它和那个热情的部分紧密相关。你不会发现那些在业余时间认真做事的人对构建、设计或工程没有真正的热情。所以这两者相辅相成。所以我认为这是对那种性格特征的一个好信号。但我在面试中喜欢做的一件事就是直接跳到 Figma 里。我会说，你知道，不是演示，你不需要准备任何东西。就打开你最近的 Figma 工作文件，或者可以是一个原型项目。不一定是 Figma，但本质上是工作文件。跟我讲讲。这些事情可以非常随意、零散、不精致，但你可以从人们谈论他们设计的东西的方式中获得很多信号，我觉得这非常非常有用。所以可以是副业项目，可以是原型，可以是构建的东西，可以是 Figma 文件，可以是他们在工作中做过的东西。从这个意义上说，它是什么并不重要，但肯定仍然有，你知道，那种固有的信号，你可以从那些认真对待副业的人身上得到。他们显然真的很享受他们正在做的事情。

I think it's a good signal. I think it really aligns closely with that enthusiastic piece. You don't find people that are building serious things on the side that aren't really passionate or enthusiastic about like building or designing or engineering. So those two kind of come hand in hand. So it's a good signal for that type of personality trait, I think. But one of the things that I love to do on interviews is just like just jump straight into like Figmas. I'll be like, you know, not like a presentation or you don't need to prepare anything. just like open your most recent Figma working file or could be like a prototype project. Doesn't need to be Figma, but they're working files essentially. Talk me through it. And like these things can be super casual and scrappy and like unpolished, but you kind of tell you can get so much signal off the way people talk about the thing that they've designed, which I find really really useful. So can be a side project, can be a prototype, can be something built, can be a Figma file, can be something that they've worked on at work. doesn't really matter in in that sense what it is, but definitely there are still, you know, kind of inherent signals that you get from the types of people that are taking things on the side really seriously. They they they clearly really enjoy what it is they're doing.

我们今天谈论的很多内容几乎就像是在更设计驱动或艺术兴趣之上添加更多技术技能，你知道。所以，我认为最近我听到很多设计师问的一个问题是，我需要达到的技术门槛是多少，才能有机会在像 11 Labs 这样理想的公司获得职位。所以，你谈到了跳进 Figma 文件，谈到了进入原型。比如，你如何看待技术悟性水平，甚至是对这些新编码工具和工作流的兴趣，与核心产品设计团队相关？

A lot of what we've talked about today has been almost like the slapping on of more technical skill sets on top of more design driven or artistic interests, you know. So, I think there's this question that I've heard a lot of designers ask recently, which is where is the technical threshold that I need to hit in order to even have a chance at getting a role at, you know, a desirable company like 11 Labs. So, you talked about hopping into a Figma file, you talked about getting into a prototype. Like, how do you think about the level of technical savvy or even interest in these new coding tools and workflows as it relates to like the core product design team?

我认为这不一样。我还没有必然地把设计工程师和产品设计师合并起来。我对设计工程师的定义一直是那种在软件和设计上一样熟练的人。要回答你关于产品设计方面的问题，这对我来说发生了变化，我认为几年前人们想要参与编码智能体可以说是可选的加分项，而现在是我的一个偏好。技能水平不如对它的热情重要，我会说。你知道，所以我和很多加入的人聊过，你知道，最近加入的很多人，他们说，我真的很想进入代码库，你知道，就像我已经使用智能体一段时间了。我没有经典的技术背景，但我一直在使用它们。我喜欢它们，我想接近代码，这对我来说就足够了。我认为这些东西的质量会继续提高。在这个世界里，有一种模式效果很好，在 11 Labs，我们有很多我称之为经典训练的全栈工程师，他们经常与产品设计师配对时，会把所有东西连接起来，构建一些东西。他们会把它做得非常接近想象的产品设计最终产品，但还有最后一英里的部分，他们很多时候不想做。不是真的感兴趣，有时也看不到。而最后一英里通常就是标记和交互，以及这里那里的像素微调。对于刚接触技术产品设计师来说，这是相当安全的东西，可以参与进来。所以出于这个原因，我最关心的是有兴趣并想使用工具的人。熟练程度不那么重要。

I think it's different. I haven't necessarily merged the like design engineer and the product designer just yet. And my definition of a design engineer has always been someone who's kind of as proficient at software as they are at at design. To kind of answer your question on the like product design sense, this changed for me where I think a few years ago people wanting to get involved with you know coding agents was kind of like optional bonus I would say whereas now kind of like a preference of mine. The skill level is less important than the enthusiasm around it I would say. you know, so I've spoken to a bunch of people who've joined, you know, a bunch of people joined recently and they were like, I really want to get into the codebase, you know, and it's like I've been using agents for a while. I don't have a a kind of a classical technical background, but I've been using them. I enjoy them, and I want to get close to the code, and that's kind of enough for me. I think the quality of these things will continue to improve. And one of the the kind of models that works well in this world where at 11 Labs we've got a lot of you what I call classically trained full stack engineers who will often times when pairing with a product designer they'll wire everything up and like build something. They'll get it pretty close to the kind of like imagined product design like final products but there is that last mile piece which they don't want to do a lot of the time. Not really interested in doing it and sometimes can't really see. And that last mile is often things just like markup and interaction and like like a pixel nudge here and there. It's pretty safe stuff for like a product designer who's new to being like technical to get in and get involved in. So for that reason, the thing that I care most about is somebody who's interested and wants to use the tools. The proficiency is less important.

还有什么我们没有谈到的，让在像 11 Labs 这样的公司做设计变得独特？

Anything else that we haven't talked about that makes designing at a company like 11 Labs unique?

这也是新的，你知道，就像我们正在训练的研究和模型是新的。所以围绕它们的交互模式也是新的。所以我认为有帮助的一件事是能够想象界面。你知道，我们已经远离了过去 20 或 30 年的经典界面。有一天研究团队会训练一个新的音乐模型，我们将拥有以前从未见过的控制音乐的新能力。你的工作本质上是尝试发明一个 UI，以最连贯的方式控制它。所以能够发明和想象，成为那种产品设计师，这很好，但很有趣，你知道，我认为这也是我要说的一件事，这是一个做产品设计非常有趣的地方，因为问题空间很迷人。你知道，有一天它可能是音乐模型的编辑器。它可能是一个智能体式的线性非线性视频编辑器。就像，你知道，我们如何从提示中设计声音和微调声音？声音克隆流程是什么样子的？我们如何测试对话智能体？这听起来太有趣了。

this is also new, you know, like the the the research and the the models we're training are new. And so the interaction patterns around them are new. So I think one thing that that helps is to be able to kind of like imagine the interface. You know, we've kind of moved away from classical interfaces that we had for the last 20 or 30 years. One day the research team will train a new music model and we'll have new capabilities for controlling music that we haven't seen before. and your job is to try and invent essentially a a UI to control that in the in the most kind of coherent way possible. So kind of like nice to be able to like invent and imagine and and and kind of like be that type of a a product designer, but it's fun, you know, I think that's also one of the things that I'll say like this is a really fun place to do product design because the problem space is fascinating. You know, one day it could be an editor for a music model. It could be a a gentic linear nonlinear video editor. It's like, you know, how do we how do we design voices and fine-tune voices from from prompts? What does a voice cloning flow look like? How do we test conversational agents? That sounds so fun.

是的。那里没有很多先例。这很酷。

Yeah. It's not a lot of precedent there. That's cool.

是的，没错。

Yeah, exactly.

所以，因为它新颖、与众不同，所以很有趣，但要在那里蓬勃发展，要求是能够想象、能够发明，并且能够抛弃任何可能受更经典 UI 技术和模式影响的先入之见。

So, because it's fun because it's new and it's different, but the requirement there to thrive is to be able to imagine, to be able to invent, and to be able to cast aside any priors that you have that are maybe influenced by more classical UI techniques and patterns.

嗯，不，这真的很酷。我也很感激你带来的一切，向我们展示你们在幕后制作的东西。真的非常、非常、非常棒地展现了我认为这类角色的未来，而且听到你如何思考它、你在投资什么。这很愉快。所以，感谢你今天抽出时间。

Well, no, this has been really cool. I also just appreciate you bringing everything to the table and showing us what you all are making behind the scenes. Really, really, really great look at what I think the future of this type of role looks like and, you know, hearing how you're thinking about it, what you're investing in. It's been a joy. So, appreciate you taking the time today.

谢谢你，Red。这太有趣了。

Thank you, Red. It's been so much fun.

嘿，如果你喜欢这一集，别忘了每周四我都会和 Tommy Gio 一起直播 Dive Radio，我们会深入探讨热门话题，并听取业内众多不可思议的设计师的意见。所以，如果你想把它加入日历，只需访问 dive.comclub/radio。我们周四下午 4 点东部时间见。

Hey, if you enjoyed this episode, don't forget that every Thursday I go live with Tommy Gio for Dive Radio, where we deep dive into trending topics and hear from a ton of incredible designers in the industry. So, if you want to get that on your calendar, just go to dive.comclub/radio. And I'll see you on Thursdays at 4:00 p.m. Eastern.