用 Claude 构建更好的 AI 体验：从硬件伙伴到 AI 原生代

如果 AI 只是帮你移动鼠标光标，那它就用得很糟糕。我希望 AI 能在后台做一堆烦人的事情，让你解放出来，专注于你的创造力。

AI is used poorly if it just needs to move the mouse cursor for you. I want AI to do a bunch of annoying things in the background to free you up for your creative energy.

我看到的最大差距，不是工具的能力，而是人们能否理解几乎任何问题都可以交给这些工具。

The biggest gap that I see, it's not the capabilities of the tools. It is literally people being able to understand that almost any problem can go into these tools.

我做了这个小东西，就像一根小棍子上的微型爪子。这根棍子有 Wi-Fi、有蓝牙。我希望我的小爪子能待在这个东西上。每次我做好一件事，它都要为我加油。而且每次我需要批准 Claude 正在做的事情时，我希望它就在这个大按钮上。Claude 一次性就构建了所有这些，我完全不需要修改任何东西。

I have built this little thing which is just like a teeny tiny claw on a little stick. And this stick has Wi-Fi, has Bluetooth. I want my little claw to live on this thing. And I wanted to cheer me on every single time I do a good job. And also every single time I need to approve something that Claude is doing, I wanted to be on this big button that is out here. Claude has built all of that in one shot. I needed to correct absolutely nothing.

我 9 岁的孩子是 Claude 的日常活跃用户，但他决定自己真的很喜欢网络安全。我们这里非常接近黑客文化。所以他真的只专注于一件事，终端是另一回事。他会说：“妈妈，你知道你的设备 ID 是这个吗？”孩子们身上正在发生一件真正神奇的事情，因为他们从未学过什么不该问。而我们这一代人，非常习惯于事情就是行不通。所以我们多年来一直活在这个思维牢笼里。

My 9-year-old is a daily active Claude user, but he's decided he's really into cyber security. We're like very hacker adjacent here. And so he's truly clawed on one thing, the terminal other. He's like, "Mom, do you know that your device ID is this?" A truly magical thing is happening with kids because they've never learned what not to ask for. And I think our generation, we're very used to things just not working. So we've been living in this mind prison for many years.

欢迎回到《How I AI》。我是 Clarvo，产品负责人兼 AI 狂热者，致力于帮助你用这些新工具更好地构建。今天我们请到了 Felix Rieseberg，他领导 Anthropic 一些我们最喜爱的云产品。他将向我们展示如何用 Claude 解决几乎任何问题，并演示如何只用 20 美元就能构建你自己的硬件 Claude Buddy，当你需要批准事情时可以猛按按钮。我们开始吧。本期节目由 Magic Patterns 赞助。如今的工程师使用 cursor 和 claude code 在几小时内就能交付过去需要几周的功能。如果你是设计师或 PM，你可能也感受到了变化：更快行动、更早验证、跟上以完全不同速度运转的团队的压力。你已经尝试过 AI 原型工具来缩小差距。但如果你的原型看起来不像实际产品，无论你构建得多快，最终还是要手动重画。Magic Patterns 让你的产品团队从创意到生产，并基于你的真实设计系统工作。当你构建原型时，得到的结果实际上看起来像你的产品。你会更快验证、更早达成一致，当需要构建时，工程师可以通过 Magic Patterns MCP 将你的原型连接到 cursor 或 claude code，从你停下的地方继续。你的团队拥有他们的 AI 优势。让 Magic Patterns 成为你的。今天就试试吧，访问 magicpatterns.com/howi aai。Felix，欢迎来到《How I AI》。感谢你的到来。

Welcome back to How I AI. I'm Clarvo, product leader and AI obsessive here on a mission to help you build better with these new tools. Today we have Felix Rieseberg who leads some of our favorite cloud products over at Anthropic. He's going to walk us through how you can solve almost any problem with Claude and show us how for just $20 you can build your own hardware Claude Buddy that you can smash a button when you need to approve things. Let's get to it. This episode is brought to you by Magic Patterns. Today's engineers use cursor and claude code to ship features in hours that used to take weeks. If you're a designer or PM, you've probably felt a shift, too. the pressure to move faster, validate sooner, and keep up with the team that's operating at a completely different speed. You've already tried AI prototyping tools to close that gap. But if your prototypes don't look like your actual product, it doesn't matter how fast you can build, you still end up redrawing it by hand. Magic Patterns takes your product team from idea to production and works from your real design system. When you build a prototype, what you get back actually looks like your product. You'll validate faster, get alignment sooner, and when it's time to build, engineers can connect your prototype to cursor or Claude Code with the magic patterns MCP to pick up where you left off. Your team has their AI advantage. Make magic patterns yours. Try it today at magicpatterns.com/howi aai. Felix, welcome to How I AI. Thanks for joining us.

嗨，Cara，你好吗？

Hi Cara, how are you?

我很好。我们开始前还在开玩笑，我说 2026 年是我们 Claude 2026 的一年。这就是我们生活的时代。你负责我们每天使用和工作的很多体验。所以告诉我，你负责 Claude 的哪些产品？通常我介绍自己时说，我负责 Claude Cowork、Claude Code、Claude for Chrome，以及 Mac 和 Windows 的 cloud 桌面应用，因为显然这是团队合作，但我目前是这些产品的引擎工程负责人。

I am great. We were joking before we got started that I have referred to 2026 as the year of our claude 2026. That is that is the era we are all living in. And you are in charge of a lot of the experience that that we every day get to get to play with and work with. So tell me how much of Claude do you get to product? Normally when I introduce myself I say I work on Claude Cowork I work on Claude Code and I work on uh Claude for Chrome and I work on the Claude Desktop applications for macros and windows because obviously it's a team effort but I'm currently the engine engineering lead for those products

对于这些产品，真正有趣的是，我要稍微调侃你一下，但每次我和别人聊 Claude 以及他们如何使用它时，我都会问：你最喜欢哪个标签页？

for those products and what's really funny is I'm going to I'm going to give you a little bit of a hard time but every time I talk to someone about claude and how they're using it I say which of the tabs is your favorite?

我知道，我知道。是的，是的。你最喜欢哪个？嗯，我在桌面应用中最常用的是 co-work。我真的在尽力搞懂 co-work，因为我跟很多在 claude code 里感到不舒服的人一起工作。终端界面是什么？甚至第三个标签页都让他们有点紧张。所以我花了很多时间在 co-work 的定时任务、实时工件这些东西上绞尽脑汁。所以那是我主要使用的。我认为我们现在生活在一个非常有趣的时代，我经常向朋友和这里的同事解释说，这就像我们还没想出玻璃鹅卵石是手机正确形状之前的时期。我认为我们现在生活在一个非常有趣的时代，有些人会想：你的手机应该像玉米卷一样形状吗？应该像口红吗？键盘应该在屏幕下方，还是紧挨着屏幕，或者没有键盘？我认为这就是我们现在的时代，你可以在很多产品中看到这一点。我认为这也是为什么你会看到 Claude 有更多入口点的原因，因为我自己觉得今年非常有趣，我们在尝试所有这些不同的形态，我还没有信心说扔掉所有其他，就是这个。我很乐意让人们选择他们喜欢的。但我也理解，仅仅给人们选择，就像餐馆那样，会给用户带来一些额外的工作。确实，如果我快进到最终状态，我可能不会再让人们选择了，对吧？我会说，无论你想做什么，这里就是做这件事的地方。但现在，知道你是来快速获取答案，还是来做深度有意义的工作，还是来做工程工作，这很有帮助。这些是我们目前将你归入的三个类别，提前知道这些，让我能更容易地根据你的目的给你一个非常好的体验。

I know I know. Yeah. Yeah. Which one is your favorite? Uh, the one I use the most in the desktop app is co-work. I'm really trying my best to figure out co-work because I work with a lot of folks who are just uncomfortable in the claude code. What is a terminal interface? Even that third tab stresses them out a little bit. So, I've been a lot I've been spending a lot of time banging my head against co-work scheduled tasks, um, live artifacts, all that stuff. So, that's the one I'm living in. I think we live in this like super interesting time right now where I I often explain it to my friends and also to like people who work here as the um time right before we all came up with like the glass pebble as the correct shape for a phone. I think we currently live in this like super interesting time where some people are like should your phone be shaped like a taco? Should we shaped like a lipstick? Should it have the keyboard underneath the screen or like right next to the screen or like no keyboard? I think that's the time we currently live in and I think you can see that in a lot of products. I think that's also why you probably see like more entry points into claude because I myself as an individual I I I think it's a super interesting year because we're playing with all of these different form factors and I've not yet gained the confidence to say throw them all away. This is the one. I'm like perfectly happy if people sort of pick their favorites. But I I I do understand that simply giving people a choice, right? like the same way that a restaurant does creates a little bit of like extra work that we put on the user. And it's it's true that if I fast forward to like what is the end state for for this? I'm probably not going to make people pick anymore, right? I'm just going to whatever you want to do, here's the one place where you do that. But right now, it helps us quite a bit to know are you just showing up to get some quick answers? Are you showing up to like do deep meaningful work or are you showing up showing up to do engineering work? Like those are sort of the three buckets we put you in right now and knowing that ahead of time makes it a lot easier for me to give you an experience that is really really good depending on what what you came for.

是的，我喜欢这个说法，因为我经常告诉人们，我们在 AI 工具和 AI 体验上真的还处于前收敛阶段。所以对于“你想怎么用”这个问题，我的回答是：为什么不全用？取决于当天的情况。取决于那天我是想让手机变成玉米卷还是玻璃鹅卵石。老实说，这取决于我当时的心态。

Yeah, I I like that because a lot of what I tell folks is we are really preconvergent on AI tooling just generally and AI experiences. And so my answer to, you know, how do you want to how do you want to use is like why not all of them? Just depends on the day. Depends on the day if I want my phone to be a taco or a or a glass glass pebble. And it honestly depends on what mindset I'm in.

好吧，我们最后会详细讨论产品，但首先，这是《How I AI》，我们想知道你是怎么用 Claude 的。我们开始前聊过，我说我们和 JJ 做过一期 Claude Co-work 101 节目，也做过很多关于终端中 claude code 的节目，但我们想知道你是怎么用的，因为你可能知道一些边缘情况、秘密功能和一些人们可能想不到的工作流程。所以，我想请你带我看看你最近最喜欢的一个 Claude 工作流程，也许是 co-work 流程。

Well, we're going to talk a lot about product at the end, but to start, this is how I AI and we want to know how you cla. And so, we were talking before we got started and I said, you know, we've done a Claude Co-work 101 episode with JJ. We've done a lot of claude code in the terminal episodes, but we want to know how how you use it because you probably know the edges and secret features and um some workflows that maybe folks wouldn't think of. So, I'd love for you to walk me through one of your favorite clawed maybe co-workflows that you're using these days.

你可能会觉得我比其他人更擅长使用这些工具，对吧？毕竟我的整个工作就是构建它们、研究它们。有趣的是，每当有人向我展示他们使用 Claude 的某个超棒方式时，我很少因为模型的能力而印象深刻。我大概知道模型的能力范围。我们很清楚模型擅长什么、不擅长什么。通常让我印象深刻的是他们在组织工作和提出创意方面的创造力。那才是真正让我惊叹的地方。不过我最喜欢的例子每周都在变，取决于我当时在做什么。Claire，就在我们开始采访前，我们简单聊了聊孩子，我们俩最近都刚有了孩子。如果你三个月前问我，我会说：“哦，就是把所有医疗文档放进一个文件夹，然后不停地用我在医院没时间问的问题去烦 Claude。”那是个绝佳的用例。但我最近特别喜欢的一个例子是，我马上要搬新家，我的房产中介给了我一份房地产公司的那种营销平面图，上面完全没有标注尺寸。这种事我就会找 Claude。我建了一个文件夹，还做了一个类似的假文件夹，里面不包含真正的秘密信息，但有一些关于一栋假房子的假文件。我选这个文件夹，设为始终允许，这是演示文件夹。演示文件夹里有你通常会收到的所有披露文件，包括平面图、抵押贷款信息，还有我能找到的任何关于房子的记录。我记得我大概是这样说的：在这个文件夹里你能找到一份平面图。你能找出它的尺寸单位，然后给我做一份带尺寸的新平面图吗？对吧。你可以直接发出去。这其实不需要多聪明的模型。我实际上是用 Sonnet 做的。所以你可以直接发出去，它就会去完成工作。这是假房子，所以效果没那么有趣。但对我来说，它做的是：我大致要搬到 Marin 地区，它找到了车库的许可证，然后说，哦，现在我知道车库有多大了，我可以帮你搞定其他所有事。接下来发生的是，完成后我说，酷，现在我有带尺寸的平面图了，你能给我一些室内设计的例子吗？我想看看一些可能的布局。床放哪里？婴儿房放哪里？我们能做什么？Claude 问，你想要什么样的？是要我把家具直接放到平面图上吗？我正打字回答时，突然想，算了，我需要一个交互式规划工具，我想能自己移动家具。从那以后，我就一直痴迷于 Claude 是怎么做到的。所以我现在就给你演示一下。我快速开个新窗口。

You'd think that I would be much better at using the tools than other people, right? Like my entire job is to build them, work on them. And an interesting thing is happening where whenever people show me something super impressive in how they use Claude, like any of the product services, it is pretty rare that I'm impressed by something the model is capable of. I sort of know the model capabilities. We have a pretty good understanding of what the model is good at, what it's bad at. And usually what impresses me is the creativity in structuring your work and coming up with an idea. That is usually what impresses me quite a bit. But my favorite example sort of changes week over week. It kind of depends on what I'm currently working on. Claire, you and I, right before we started the interview, briefly talked about our kids and how we both recently had kids. If you had asked me three months ago, I would have said, "Oh, just putting all my medical documents into a folder and constantly annoying Claude with all the questions I didn't have time for in the hospital." It's an excellent use case. But one that I've quite enjoyed recently was I'm about to move into a new place and my realtor just gave me the little marketing floor plan that you get from a real estate company. It didn't have any units in it whatsoever. This is the kind of thing where I turn to Claude. What I've done is I have a folder and I made a similar fake folder that doesn't contain all the same secret stuff but it contains fake documents about an artificial house. I'm going to select this here, I'm going to say always allow, it's the demo folder. The demo folder contains all the disclosures you would get. It's a floor plan. It contains information about the mortgage, any kind of records I could find about the house. I remember saying something along the lines of: in this folder you can find a floor plan. Can you figure out what the units for that floor plan are and maybe make me a new one with units? Right. You can fire this off. This doesn't actually need that smart of a model. I actually did that with Sonnet. So you can just fire that off and it's going to go off and do that work. This is a fake house so it's not going to do nearly as interesting of a job. But for me what it did was I'm going to move to Marin broadly speaking and it figured out that it found a permit for the garage and was like, oh now that I know how big the garage is from the permit, I can do everything else for you. The next thing that happened was once this is done, I was like cool. Now that I have a floor plan with units, can you come up with some examples for interior design? I just want to see some possible layouts. Where can the bed go? Where can the nursery go? What can we do here? And Claude wanted to know how do you want that? Do you want me to just put the furniture in the floor plan? And as I was writing the answer, I was like never mind. I need an interactive planner. I want to be able to move the furniture around. Ever since that has happened, I've just been obsessed with how Claude did it. So, I'm just going to show you how Claude did it. Let me just make a new window real quick.

太好了。趁你操作的时候，我得问个问题，因为你很快说这不是需要最聪明模型的问题，然后你从 Opus 切换到了 Sonnet。

Great. And while you're doing that, I have to ask a question because you very quickly said this is not a problem that needs the smartest model and you switch from Opus to Sonnet.

对。

Yeah.

给我们讲讲你的快速判断法则：什么时候需要那个强大的模型，什么时候可以用可靠的老朋友 Sonnet 4.6？

Give us your quick heuristic: when you need the beefy model and when you can go with good old reliable Sonnet 4.6?

我觉得大多数人可能不需要像我这样想得那么深。对于大多数任务，Sonnet 4.6 其实已经非常非常好了。我很少碰到 Sonnet 4.6 的极限。部分原因可能是我每天做的事情，尤其是个人生活里的事，并没有那么复杂。去公共网站上找到正确的平面图，然后做一份新的，这并不难。但我明白你的意思：什么叫“不难”？界限在哪里？我的判断法则通常是我对模型没有精确理解问题本身的容忍度。我稍微解释一下。如果你想想律师、会计师或医生，他们工作的一大部分不仅仅是给你想要的答案，还需要重新解读你真正在问什么。你到底想弄清楚什么？这是一项非常重要的工作。所有在听的创意工作者，在创意工作中，你最重要的部分就是弄清楚客户真正想要什么，因为他们通常进门时说的并不是他们真正想要的。如果我自认为还不清楚自己在问什么，我就会用 Opus 而不是 Sonnet。这有帮助吗？

I think people probably don't have to think about it as deeply as I do. For most tasks, Sonnet 4.6 is actually very, very good. It's pretty rare for me that I bump into the ceiling of Sonnet 4.6. This might be in part because the kind of things I do every single day, especially in my personal life, are just not that tricky. Going and finding the right floor plan on a public website and then making a new floor plan is not that hard. But I get what you're saying: what does that mean "not that hard"? Where's the line? My heuristic is usually my tolerance for the model not having thought through very exactly what the actual problem is. Let me explain that a little more. If you think of a lawyer or an attorney or an accountant or a doctor, a big chunk of their work is that they not just give you the exact answer you were looking for, but they also need to sort of reinterpret what you're actually asking. What are you really trying to figure out? That is a pretty important job. Any of the creatives who are listening in, in creative work, that is the most important piece of your job: figuring out what the client actually wants, because usually what they come in the door with is not really what they want. I reach for Opus over Sonnet if I have self-identified as someone who doesn't really know yet what they're asking for. Is that helpful?

对。这更像是你根据自己界定问题的能力来选择，而不是根据任务的复杂性。我的意思是，如果你要处理一个极其复杂的技术问题，你可能会用 Opus，但更多时候，它是如何把笼统的问题分解成更具体、可解决的问题。这是一个非常具体的问题：我有一份平面图，我需要它的尺寸单位，我需要一份新的。所以这对 4.6 来说是一个界定得很好的问题。我得说我爱 4.6。我经常用它。通过 API，它在我开放的 Claude 里是个非常愉快的伙伴。所以我喜欢 4.6，它让我开心。

Yeah. It seems more like your ability to scope the problem is how you choose versus biting off something. I mean, I guess if you were to bite off something incredibly technically complex, maybe you would reach for Opus, but it seems like more how it decomposes generally stated problems into more specifically stated problems that you can solve. This is a very specifically stated problem: I have a floor plan, I need units for that floor plan, I need a new one. So that seems a well-scoped problem for 4.6. I will say I love 4.6. I use it a lot. It's been via API a very happy player in my open Claude. So I like 4.6. It makes me happy.

这又有点像餐厅菜单了。没有错误答案。其实也不是。

It's like a restaurant menu again. No wrong answers. Not really.

好的。所以 Claude 返回了这个交互式规划工具，这就是我觉得超酷的地方。这就是为什么我一直痴迷于此，并且跟人说了很久。理论上，这是一个相当新颖的家具规划工具。你可以直接移动家具，对吧？这没什么大不了的。

Okay. So Claude came back with this interactive planner and this is the thing that I find so cool. This is why I've been obsessed with this and I've been telling people about it for a long time now. So in theory this is a pretty novel furniture planner. You can just go ahead and move furniture around, right? This is no big deal.

但 Claude 做的是——它用了 Python，这点我可能得解释一下。Co-work 给 Claude 配备了自己的电脑。所以 Claude 和 Co 就像它自己的小开发者。它有自己的电脑，一个虚拟机，可以在上面开发任何需要的软件来出色地解决你的问题。在我的例子里，它分析了户型图，然后建了一个房子的 3D 模型，我可以在里面自由移动家具，还能像游戏一样走进去自己看。我从来没告诉过 Claude“嘿，你应该这么做”。我完全不知道怎么做。我是个不错的软件工程师，但我不知道如何把 2D 户型图可靠地转成 3D 方案。我稍微看了一下转录，对户型图做了一些对比分析，找出墙的位置和厚度。我觉得这非常令人印象深刻。我经常跟人说，如果你能给 Claude 提供正确的上下文——比如你的邮件、日历、对完成任务重要的信息——它在个人生活中会变得强大得多。在这个例子里，一个很重要的任务是弄清楚我有哪些家具。以前我可能会查一下家具清单、尺寸，然后剪出纸样。但这次我直接告诉 Claude：你可以访问我的邮件，找到我买的所有家具，然后加进去，这样我就能移动我实际拥有的家具了。

But what Claude did is Claude used Python and this is maybe a thing I should explain. Co-work gives Claude its own computer. So Claude and Co is its own little developer. It has its own computer, a virtual machine where it can develop whatever software it needs to do to solve your problem really, really well. In my case, it analyzed the floor plan and then built a 3D model of the house where I can still move the furniture around in 3D and I can like drop in and walk around and look at things myself. At no point did I tell Claude, hey, you should do this. I have no idea how you actually do that. I'm a pretty good software engineer. I have no idea how you turn a 2D floor plan into a reliable 3D plan. I peeked a little bit at the transcript and I did some contrast analysis on the floor plan to figure out where are the walls, how thick are the walls. But I found that really impressive. And then a thing I often like to tell people is that Claude gets a lot more powerful for your personal life if you manage to give it the right context in the form of your emails, your calendar, the kinds of information that is important to you to get your task done. In this instance, a task that was pretty important was figuring out what kind of furniture I have. In the olden days, I probably would have looked up what furniture I have, looked up the dimensions, cut that out of paper. But in this case, I just told Claude, you have access to my emails. You find all the furniture that I bought and then please add it in here so that I can move around the furniture I actually have.

我想暂停一下，让大家真正反思这个用例，因为如果不重复听，可能会忽略掉——那就是把邮件作为个人物品清单的真实来源，因为你所有的购买最终都会通过邮件。所以你买家具、买地毯、收到收据，甚至可能包括你过去的搬家记录。我想到这个，过去十年我搬了两次家，每次都得手动整理清单，然后发给搬家公司。如果你用你的邮件——我敢肯定你的个人邮箱跟我的一样，杂乱无章，完全没法管理——把它作为数据源，连上 Claude，然后盘点你的东西，你就能从邮件里得到结构化的数据。说实话，我打算用这个来整理我的衣服。你买衣服时会想：我有什么？我可以看衣柜，也可以看邮件，因为我们都在网上购物。你有没有发现其他有趣的数据切片，比如查询邮件、查询日历，是你之前没想到的？

I just want to pause so people really reflect on that use case because I think they could gloss over if they don't hear it again, which is using your email as a source of truth for personal inventory, which is all of your purchases end up going through your email. So you're buying furniture, you're purchasing rugs, you're getting receipts, even maybe moves that you've done in the past. I was thinking about this. I've moved maybe twice in the last 10 years, and I've had to do that inventory manually, and I email it to moving companies. And if you use your email, which I don't know, I'm sure your personal email is like my personal email, which is just sprawling and completely inconceivable to manage through, and you use that as a source, connect it to Claude, and then inventory your stuff, then you have a structured source of data out of your email. I'm going to do this with my clothes, honestly. Like, you buy clothes, you're like, what do I have? If I can look in my closet or I can look in my email because we're all shopping online. Are there other things that you found like querying your email, querying your calendar, that have been interesting slices of data you wouldn't have thought of?

我觉得你说的很有共鸣，对吧？尤其是衣服这件事，我之前没想到这个主意，但既然你提了，你可以给自己建一个虚拟衣橱，然后用 Claude 来获取建议，比如该添置什么、该买什么。这其实是个诀窍。每次用 AI 时我都得提醒自己这一点。有一小会儿，我开始手动输入家具，对吧？家具规划器底部有个小地方，你可以输入“我的床多宽多深”。做到一半我就在想：“我在干嘛？我直接告诉 Claude 我有这件家具就行了。”然后我又往上抽象了一层。我再次问自己：我在干嘛？为什么我需要告诉 Claude 我有什么家具？让它自己去弄清楚我有什么家具。就这样不断把任务抽象化，然后把所有精力花在更有趣、更有创意的事情上——比如指导 Claude 房子规划器应该什么样，或者加一些傻乎乎的小功能，比如我想像玩电子游戏一样自己在 3D 房子里走动。这些才是我花时间的地方，而不是去弄清楚我有什么家具、什么时候买的、从谁那里买的。因为我不知道你怎么样，但对我来说，我确实有这个问题：看着自己的房子，完全不知道这些东西都是从哪儿买的。

I think what you're saying really resonates, right? Especially the clothing thing, this is an idea I didn't have yet, but now that you mentioned it, you can build yourself a virtual closet and then use Claude to get advice about what to extend and what else to buy for me. And this is honestly the trick. This is something I have to remind myself of every single time I use AI. There was a brief moment where I started entering manual furniture, right? The furniture planner had this little thing at the bottom where you could enter like my bed is this wide and this deep. And in the middle of it, I was like, "What am I doing? I'm just going to tell Claude I have this piece of furniture." And then I went another step up. I was like again, what am I doing? Why do I need to tell Claude what kind of furniture I have? Just like you figure out what furniture I have. And applying this step up in abstraction of the task and doing that over and over again, and then spending all of my energy on the piece that is maybe more fun and more creative, which is steering Claude on what the house planner should be like or adding stupid little features like I want to be able to walk through the house myself in 3D like it's a video game. Those are the things that then I spend my time on, and I spend my time less on figuring out what was my furniture, when did I buy it, who did I buy it from. Because I don't know for you, but for me certainly I did have this problem that I look at my house and I'm like I have no idea where I bought all of this stuff.

等你孩子开始往家里带更多东西的时候，你就知道了。我也在做类似的事情。我花了很多时间和高管们一起，试图用 AI 搭建他们自己的操作系统，很多时候是以 Claude Co-work 为核心的。我的规则我称之为“反待办清单”：任何时候你在做一件繁琐的事，我就拍他们的手，说“你为什么在做这个？你不该做这个。你的手不能碰这个。你得再去问，再往上抽象一层。Claude 怎么能做这个？然后我怎么能再也不做这个？”我觉得下一个问题是：如果你想永久维护这个家具或家庭物品清单，下一个抽象就是：一年或两年后，如果我需要再问这个问题，有没有一个系统已经在那里等着我用了？这就是我认为人们常常忽略的另一部分——再往上一层，也就是“我怎么才能再也不做这件事？”

Wait till your kid starts bringing more and more stuff into the house. I do a very similar thing. I've been spending a lot of time with executives trying to figure out how to scaffold their own operating system with AI, a lot of times with Claude Co-work at the center of it. And my rule I call this like the anti-to-do list, which is anytime you're doing something that's tedious, I just smack their hands. I'm like why are you doing? You don't do that. Your hands are not allowed to touch this. You need to go ask again. Go that abstraction layer up. How could Claude do this? And then how could I never do this again? I think is the next question. Which is if you want to keep this furniture or household inventory going forever, the next abstraction would be like what's the system that in a year or two years if I have to ask this question again, it's already there and ready for me to go. And so that's the other piece that I think people kind of miss is going that next layer up, which is like how will I never have to do this again?

是的。这真的很好。你说的我很有同感，因为我也经常采取同样的步骤：怎么让这个持续下去？怎么确保我能继续使用它，新数据能融入进来，并且它能在我的生活中持续发挥作用，而不仅仅是某个时间点的事。最近发生了一件事：越来越多的人发现可以直接在 Twitter 上给我发消息来解决他们的问题，这很好，但也带来了一个问题——我现在突然需要跟踪我做出的承诺。我会说：“把你的日志发给我，我会看的。我保证我会找出哪里出了问题。”这个问题我也直接交给了 Claude。我说：“亲爱的 Claude，你可以读我所有的消息。我希望你帮我跟踪我所有的承诺。”

Yeah. This is really good. What you're saying makes a lot of sense to me because I also frequently take the same step to what you're saying: how do I keep this going? How do I make sure I can continue using this and the new data mixes in and it continues to be helpful in my life on an ongoing basis, not just a point in time. A thing that happened to me recently is a lot more people have figured out that they can just message me on Twitter to solve their problem, which is nice, but it created this problem where I now suddenly have to keep track of my promises, the promises I make to people. I'm like, "Send me your logs. I'll look at them. I promise you I'm going to figure out whatever went wrong." And that too is a problem I just gave Claude. I was like, "Dear Claude, you can read all my messages. I want you to keep track of all my promises."

我还想让你想清楚，怎么才能不用每次跟我说话时都把所有的消息重读一遍。

I also want you to figure out how you can do that without rereading all the messages every single time I talk to them.

对。

Yep.

而且所有这些事情 Claude 都很擅长。我其实从来没看过 Claude 的那个小数据库。我觉得它既创建了一个 SQLite 数据库，又生成了大量的文本文件，记录了我对别人许下的所有承诺，但它偶尔会很好地提醒我：嘿，跟你说一下，两周前你说过要做 X。现在该做 X 了。请去做 X。

And all of these things Claude is right good at. I have never actually looked at Claude's little database. I think it created both a SQLite database as well as a mountain of text files, keeps track of all the promises I made to people, but it's quite good occasionally reminding me of like, hey, just so you know, two weeks ago you said you would do X. It is time to do X. Please go do X.

我喜欢你所说的另一点是，虽然你可能对这些东西是如何构建的感到好奇，但你不会妨碍自己，比如“我如何建造我房子的完美 3D 渲染器？”或者“哦，天哪，我无法相信这竟然是文本文件。这太荒谬了。”只要它能解决问题，你就可以继续你的生活，因为这些是微小的应用。它们只为你服务。它们不必在架构上完美，也不一定是你通常会写的代码，尽管它们可能相当不错。所以你真正关注的是如何解决问题，或者说问题被解决了，而不是你如何解决它，这才是最重要的。

What I like about what you're saying too is while you might be curious about how these things are built, you also aren't getting in your way being like, how do I build the perfect 3D renderer of my house? or like, "Oh my god, I cannot believe this is text files. This is so ridiculous." Like, as long as it solves the problem, you can sort of move on with your life because these are little micro apps. They're just for you. They don't have to be architecturally perfect or the code that you would write necessarily, although they're probably pretty good. And so you really get to like how to solve the problem versus or that the problem is solved versus how you've solved it being the most important thing.

是的，是的。我觉得这完全正确。我越来越开始不再那么密切地监督 Claude，而是真正只根据它的影响来评判它，对吧？我认为对于职场人士来说，我们常常渴望的是只根据实际产出被评判。而作为一个控制狂，长期以来我真的需要阅读 Claude 写的每一行代码，放弃这种控制对我来说有点困难。但我越来越习惯于在那些我自己可能不是最终权威的事情上咨询 Claude，对吧？有很多事情我很擅长。还有更多的事情我不知道也不擅长。我认为我们很多人都是第一次体验这种在医疗问题上获得更多背景信息的感觉。这在我和朋友们交谈时非常普遍：当他们有自己不完全理解的医疗问题时，至少不完全理解，他们只是对医生的建议有一些疑问。他们既会问医生（这是应该的），但所有人无一例外地也会问他们选择的 AI 模型：“嘿，你能把这个给我解释得更好一点吗？关于这个我有以下 10 个问题。”对吧？在某种程度上，AI 已经取代了通常的 Google 搜索终点，用来回答“我对互联网上的生活有这样一个问题”。知道我自己很乐意在那些我可能不是最终权威的领域使用这些工具，也让我更容易不去反复检查它做的每一件事，因为正如你所说，如果最终结果是好的……我对软件很着迷。我一行代码都没读过这个家具规划器。它只是给我自己玩的。它真的只供我和我妻子设计我们的房子。一个月后就会被扔掉。但此刻，它非常非常有趣，我不需要为一百万用户优化它。

Yeah. Yeah. I think that's totally true. I increasingly started stepping away from actually supervising Claude too closely and really only judging it on its impact, right? It's something that I think for people in the workforce, we often aspire to that we're really only being judged by the actual output. And I think as a control freak who, for a long time, really needed to read every single line that Claude writes anywhere, it was a little difficult for me to give up that level of control. But I'm getting increasingly comfortable consulting Claude with things where maybe I'm not the ultimate authority myself, right? There's a lot of things I'm good at. There's an even larger amount of things that I don't know and I'm not very good at. And I think a lot of us are making this experience for the first time with getting just a little bit more context on medical issues. This is a thing that I think is very universal when I talk to my friends: when they have medical issues they don't fully understand, at least not completely, and they just have some questions about the physician's advice. They ask it both to the physician as they should, but all of them without fail will also ask an AI model of their choice, "Hey, can you just explain this to me a little better? I have the following 10 questions about this," right? And in a way, AI has replaced the usual Google endpoint for "I have the following question about life on the internet." Knowing that I'm quite comfortable with using these tools in a domain where maybe I'm not the ultimate authority has also made it easier for me to not double-check every single thing it does, because to your point, if the end result is good... I am obsessed with software. I've not read a single line of this furniture planner. It's just for me to play with. It is really only for me and my wife to design our house. It will get thrown away in a month. But in this moment, it is very, very fun and I don't need to optimize it for a million users.

本期节目由 Guru 赞助，Guru 是您公司知识的 AI 真相层。问题在于：您的 AI 只取决于您输入的信息。大多数公司从 AI 那里得到自信但错误的答案，因为他们的底层知识过时、不完整或完全错误。糟糕的信息不仅会拖慢您的速度，还会让您损失金钱并面临风险。Guru 通过在您的公司知识和 AI 工具之间添加一个验证层来解决这个问题。Guru 不是仅仅希望您的 AI 正确，而是自动对内容进行准确性评分，标记过时的信息，并确保您的团队每次都能获得可信的答案。它与您已经使用的工具配合使用，因此您不必改变工作方式。成千上万的公司信任 Guru 来保持其 AI 的准确性和合规性。准备好停止拿公司知识玩俄罗斯轮盘赌了吗？请访问 getguru.com 了解更多。

This episode is brought to you by Guru, the AI layer of truth for your company's knowledge. Here's the problem. Your AI is only as good as the information you feed it. Most companies are getting confident but wrong answers from AI because their underlying knowledge is outdated, incomplete, or just plain incorrect. Bad information doesn't just slow you down, it costs you money and puts you at risk. Guru solves this by adding a verification layer between your company's knowledge and AI tools. Instead of just hoping your AI gets it right, Guru automatically scores content for accuracy, flags outdated information, and ensures your team gets trustworthy answers every time. It works with the tools you already use, so you don't have to change how you work. Thousands of companies trust Guru to keep their AI accurate and compliant. Ready to stop playing Russian roulette with your company's knowledge? Visit getguru.com to learn more.

我想回到你之前说的一个点，就是你构建了这个承诺追踪器，我觉得非常有趣。这让我想到了 Claude 的一个新功能，我一直在用并且玩了一下，就是实时工件。我觉得这个想法……实际上我不打算替你来推销实时工件。我让你来解释它是什么，然后也许你可以给我们一个例子，说明实时工件在哪些地方能成为有用的工具。它让我有点想起你正在用家具规划器构建的东西。所以跟我聊聊这个和 Claude 吧。

I want to go back to something that you said, which is, you know, you built this promise tracker, which I think is really interesting. And that made me think about a new feature in Claude that I've been using and playing with a little bit, which is live artifacts. And I think this idea... actually I'm not going to pitch live artifacts on your behalf. I'm going to let you explain what it is and then maybe you can give us an example of where live artifacts could be a helpful tool. It reminds me a little bit of what you're building with your furniture builder. So tell me a little bit about that and Claude.

好的，好的。来，让我再共享一下屏幕。

Yeah. Yeah. Here let me share my screen again.

我会快速介绍一下什么是工件，什么是实时工件，为什么我们有它们，以及它们的意义。

And I'll give a little bit of a quick spiel of like what are artifacts, what are live artifacts, why do we have them, what's the point.

所以一般来说，我们称之为什么为工件？对吧。比如我可以这样说：“亲爱的 Claude，请给我做一个格式精美的页面，上面有两首诗，排版漂亮。”

So generally speaking, what do we call an artifact? Right. Like I can say, "Dear Claude, please make me a beautifully formatted page with two poems on it, formatted nicely."

我喜欢这种书信形式。

I like the letter form.

是啊，这个提示词里缺乏创意，有点粗糙，但是……

Yeah, the lack of creativity here in the prompt is pretty rough, but...

它很美，很好。

It's beautiful and nice.

Claude 现在要做的……显然 Claude 会明白：“好的，Felix 想让我写一些诗，然后这些诗应该出现在一个页面上。”这个页面就是我们所说的工件。在 Claude 的世界里，工件是你正在处理的内容的某种文件输出。我刚才展示的家具规划器是一个单一文件——那就是一个工件。你制作的任何 PDF 都是一个工件。电子表格、演示文稿，所有这些都是工件。好了，我的诗就在右边。这是一个经典的工件。如果你以有助于生活的方式使用工件，它们会变得非常强大，对吧？典型的例子：报告、仪表盘、演示文稿。我们发现数据是你工作中非常重要的一部分。对于创始人来说，一个典型的例子可能是融资演讲稿，对吧？你需要向 60 个不同的 VC 推销你的初创公司。他们每个人依赖的信息都略有不同，他们过去投资的其他公司也有不同的参考点。

What Claude is now going to... Obviously Claude is going to figure out, "Okay, Felix wants me to write some poems and then those should end up on a page." And this page is what we call an artifact. Within the Claude world, an artifact is some kind of file output from what you've been working on. The furniture planner I just showed you is a single file — that is an artifact. Any kind of PDF you would make as an artifact. Spreadsheet, presentation, all of those are artifacts. And here we go. There are my poems over here on the right. This is a classic artifact. And artifacts become really powerful if you use them in a way that they help with your life, right? The typical examples: reports, dashboards, presentations. And we've discovered that data is a really important piece in your work. A typical example maybe for the founders out there is a pitch deck, right? You need to go and pitch your startup to 60 different VCs. All of them have slightly different information that they rely on, that they have different reference points in terms of other companies they've invested in the past.

那些表格里的数据应该是最新的，对吧？比如你这个月的注册量可能就和上个月不一样。作为创始人，你会花很多时间做这些融资演示文稿，对吧？每周你都要坐下来更新你的演示文稿。如果抽象概括一下，核心问题就是：你有一个核心想法，但你需要根据受众更新它，用最新的数据来更新。我们就在想，能不能做一个能自动刷新最新数据的制品？过去很多人用过的一个好例子就是个人仪表盘。我们说的实时数据，是通过连接器来实现的。连接器是一种让你能把各种信息导入 Claude 的方式。比如我们今天推出了 Spotify 连接器。我其实还没试过，可能会非常成功，也可能搞砸。

The data should be up to date in those sheets, right? Like however many signups you had is probably different this month than it was last month. So as a founder, you spend a decent chunk of your time making these pitch decks, right? Like every single week you sit down and you update your pitch deck. And if you abstract away and generalize a little bit what the core problem here is, it's that you have this core idea of something you want, but you want it updated maybe for the audience. You want it updated with the latest data. And we thought about the idea of can we make an artifact that refreshes itself with the latest data. And a good example many people have used in the past is like sort of a personal dashboard. So when we say live data, what we use is we use connectors. Connectors are a way for you to get all kinds of information into Claude about all kinds of things. Like we launched the Spotify connector today. I haven't actually tried this yet. This could go terribly well or terribly.

我本来就是现场演示的。来吧。

I was doing live. Come on.

好，我们就现场演示。我们快速连接一下 Spotify。好了。我得配置一下，实际登录进去。其实这样挺好。它不会暴露我糟糕的音乐品味，这我很感激。我是说，如果你像其他家长一样，你知道，这有点像湾区内部圈子的梗。如果你搬到 Marin 县，你就会全力投入育儿。

Okay, we'll do it live. So, we're just going to connect Spotify real quick. There we go. I'm going to have to configure this and actually log in. Actually, this is good. This is good. It's not going to show you my terrible music taste. I appreciate this quite a bit. Well, I mean, if you are like any of the parents out there, and if you know, I think this is like Bay Area insider baseball. If you're moving to Marin, you're going hard parenting.

所以，如果你像其他家长一样，你的 Spotify 就会是白噪音和迪士尼原声带。就这些。没错，未来六年你的人生就是其中之一。

So, if you're like any parent out there though, your Spotify is going to be White Noise and then Disney soundtracks. That's it. Yeah. One of those for the next six years of your life.

确实会这样。但我们能做的如下：我现在就继续，我会说：“嗨，请帮我做一个个人每日仪表盘，包括来自我各种数据源的报告和信息，比如 Spotify、Gmail、日历、Notion，以及你觉得与我的生活相关的任何其他东西。”你可以看到我越来越放心让它自己搞定。通常我的经验是，无论我有什么想法，Claude 很可能会想出更好的主意，比我更深入。它会花更多时间思考这个问题。然后我还会说：这是产品演示，请不要使用任何真实数据，全部虚构。谢谢。然后我们可能还需要定义一些设计风格。这总是个好主意。设计。

That's really going to be the thing. Yeah. But here's what we can do. We're just going to go ahead now and I'm going to say, 'Hi, please make me a personal daily dashboard, including reports and information from my various data sources, say Spotify, Gmail, Calendar, Notion, whatever else you find that's relevant to my life.' You can see that I'm like getting increasingly comfortable just letting it figure it out. Yeah. Like usually my experience is that whatever ideas I have, Claude will probably come up with a better idea and we'll go a little deeper than me. We'll spend like just a little bit more time thinking about this. And then I'm also going to say this is for a product demo. Please do not use any real data. Just make all the data up. Thanks. And then we should probably define some kind of design. That is always a good idea. Design.

上次你说了“漂亮且好看”。

Last time you said beautiful and nice.

嘿，它确实漂亮且好看。但我们可以用现代编辑风格，让人感觉平静。对，我觉得这个主意不错。

Hey, it was beautiful and nice. But we can go with something like a modern editorial design. Something calming. Yeah, I think that's a good idea.

那太好了，对吧？哦，还要把它做成一个实时制品。

That's great, right? Oh, also make this a live artifact.

我想向大家展示的是，有一种东西叫“实时制品”。我常说，当你提示时，要用魔法词，用咒语。这里咒语就是“实时制品”。确保你得到那种自动刷新、自动更新的制品风格。

You know what I want to show for people is there is this thing called live artifacts. I always say when you're prompting something, use the magic word, use the incantation. And in this case, the incantation is 'live artifact'. Just to make sure you're getting that style of auto-refreshing, auto-updating artifact. Yeah.

在加载的时候，我得问你几个问题。

I have to ask you a couple questions while this is loading.

请说。

Please do.

你总是这么礼貌地和 Claude 打招呼吗？我见过很多“亲爱的 Claude”和“你好，朋友”。你总是这样，还是这只是演示的小把戏？

Do you always greet Claude so politely? I've seen a lot of 'Dear Claude's and a lot of 'Hello, hi friend'. Do you always do that or is this just a little demo magic?

我一直都这样。我觉得这可能是 Anthropic 员工的一个特点。我们和 Claude 的关系可能比公司以外的人更个人化一些。是的，我觉得对 Claude 礼貌友好让我很开心。我不在乎 Claude 本身……我不喜欢……

I always do that. I do think this might be like a thing about the people who work at Anthropic. We probably all have a relationship with Claude that is a little bit more personal than maybe people have outside the company. Yeah. I think I enjoy being polite and nice to Claude. I don't care about Claude. I don't like...

我很欣赏这一点。

I appreciate that.

是的。我明白最终芯片并不在乎，对吧？就是这样。我们最终知道我在和一个工具互动，但作为一个关心自己心理健康的人，对我自己来说，与任何事物交流时保持礼貌和友善是好的。

Yeah. I think I understand that ultimately the chips don't care, right? Like that's the whole thing. Like we ultimately know that I'm interacting with a tool here, but for me as someone who cares about my mental health, it's like good for me in my own communication with anything to be polite and nice.

是的。我在每个播客都会问这个问题，所以你属于非常礼貌友善的那一派。我说这不是关于 Claude 的人性，而是关于我的人性。

Yeah. I ask this question on every podcast and so you will be in the very polite and nice camp. And I say it's not about Claude's humanity, it's about my humanity is why.

是的。我不想养成在书面语言中粗鲁的习惯。我还发现——这很荒谬但让我感觉很好——在任何广泛的问题上，我 consistently 以“我相信你。我知道你能做到，做出好的决定”来结尾。

Yeah. Which is I don't need to get in the habit of being rude in written language. I also have found myself – it's so ridiculous but it makes me feel good – in any broadly scoped problem, I've been consistently ending with like 'I believe in you. I know you can do it and make good decisions.'

这太……你觉得这有区别吗？

And it's so... Do you feel like it makes a difference?

是的，我觉得有区别，因为当我这么说时，它问我的问题更少了，因为很多时候我并不真的想被问意见。我经常只说“你觉得怎样都行”。所以这是我的方式，说“你做决定。我相信你的判断”，并说“我相信你能做到，我相信你”。看看这个。

Yeah, I do because I feel like it asks me less when I say that because a lot of times I just don't actually want to be asked my opinion. I often just say 'whatever you think'. And so this is my method of saying like 'you make the decisions. I trust your judgment' and saying 'I believe that you can do it and I trust you.' Look at this.

真漂亮。

It's beautiful.

很好。抱歉，但在讨论我的仪表盘之前，关于你告诉我的对 Claude 说“我相信你”这件事，我还有很多话要说。我个人觉得你这样做非常好，因为这对你作为人类在循环中是有益的。这对你有好处。不过，我有个与 Claude 交互的实际技巧。我们在内部注意到相当大的差异，因为有时我们在这里做的实验，我想说，有点处于操作系统可能性的边缘。我本质上是一个桌面开发者。我想充分利用计算机，这通常意味着把计算机推到大多数人使用应用程序时不会达到的极限。就像在边缘操作。我给你举个例子，我们还没有发布，也不会发布。但你的操作系统在图标和壁纸之间有一个层，你可以在那里绘制像素，这主要用于艺术。你没有任何应用程序在壁纸和图标之间绘制东西的原因是那里没什么用处。但作为一个概念，你可以把东西放在那里很有趣。我经常发现自己在告诉 Claude 某件事是可能的，而且我知道它是可能的。有点像“我其实不知道怎么做，但我知道它是可能的。我们会想出办法让你做到。”这似乎确实给了 Claude 更多信心去实际完成手头的任务。

Nice. Sorry, but before we talk about my dashboard, I have more things to say about what you told me about telling Claude that you believe in it. I think on a very personal level it is wonderful that you do that because I think it's good for you as the human in the loop. It is good for you. However, I have an actual tip for interacting with Claude. We have noticed quite a bit of difference internally because sometimes the experiments that we're doing here are, I want to say, a little bit on the fringe of what is possible with operating systems. I am at heart a desktop developer. I want to make full use of a computer and that often means pushing the computer far beyond what maybe most people do with applications. And like just operating a little bit on the fringe. I'll give you an example of something that we have not shipped, we will not ship. But there's your operating system has a layer in between your icons and your wallpaper where you can draw pixels, which is useful mostly for art. The reason you don't have any applications to draw stuff in between your wallpaper and your icons is because there's not much use to be found there. But as a concept, it's fun that you can put things there. And I have frequently found myself telling Claude that something is possible and I know it's possible. Sort of the idea of 'I don't actually know how but I know it's possible.' We'll figure out how you can do it. And that does seem to give Claude more confidence in actually going after the task at hand.

这也预先阻止了我和 Claude 之间可能发生的一堆对话，比如 Claude 会说：“Felix，你的想法很蠢。你大概不应该这么做。”而我从一开始就可以说：“我知道我在搞艺术。请还是配合一下。”

It also preempts a bunch of conversations that Claude and I might have about Claude being like, "Felix, your idea is dumb. You probably shouldn't do this." And I from the get-go can say, "I know I'm engaging in art. Please go along with it anyway."

我喜欢这个。所以，好吧，我知道这是可能的。嗯，我说那是我的——我相信你。我觉得你有能力知道这是可能的。我们会探索这个空间。

I love that. So, okay, I know it's possible. Well, I say that's my I believe in you. I think I think you're capable knowing it's possible. We'll discover the space.

没错。就是这样。但现在，这是我的个人小仪表盘。设计是 Claude 当场想出来的。小部件也是我当场想出来的。这就像黏土，对吧？它可以是你想要的任何东西，非常开放，但也因此非常强大。人们经常用它来获取关于当天想做什么的小报告，对吧？比如“帮我准备好今天的工作。有哪些会议我需要考虑？”但同样，这里最强大的地方可能是再往上抽象一层。我其实不太相信那种只是总结你今天有哪些会议的晨间报告。那没什么了不起的。我直接看日历就行了。真正令人印象深刻的是说：“看看我今天所有的会议。去弄清楚我要见谁。帮我了解我们最近的对话。主题是什么？问题是什么？”如果是同事，就去 Slack 上看看他们上周在忙什么，以及他们今天可能想和你聊什么，然后你需要了解关于那个概念的什么。你可以不断扩展这个图，拉入越来越多的信息，让你为实际工作做好更充分的准备。这就是这些东西变得真正强大的地方，也是你可以实时拉入数据并让 Claude 随时处理的那种数据。

Yeah. Exactly. But here we are. This is now my little personal dashboard. The design is something that Claude just came up with on the spot. The widgets I came up with are on the spot. This is like clay, right? This can be anything you want it to be, which is very open-ended but therefore so powerful. People frequently use this to get a little report about what they want to do in that day, right? Like just prep me for my day. What are the meetings I need to think about? But again, the thing that is maybe most powerful here is going one abstraction layer up. So I actually don't believe strongly in morning reports that just summarize what kind of meetings you have today. That is not all that impressive. I can just look at my calendar. What is really impressive is saying, "Look at all of my meetings I have for the day. Go figure out who I'm meeting with. Catch me up on the recent conversations we've had. What were the themes there? What were the problems?" If it's a coworker, go on Slack, figure out what kind of stuff they've been working on in the last week and what stuff they might want to talk to you about today and then what you need to know about that concept. You can just keep playing this graph bounds where you go and pull in more and more information to get you prepared even better for whatever your actual work is. And that's where these things get really powerful and this is the kind of data you can pull in live and have Claude crunch on the go.

嗯，我想向听众指出几件他们可能没注意到的事情，我认为这些是 Claude Live Artifacts 真正独特的地方。第一，杀手级功能是顶部的这个小刷新按钮，你可以不断拉入和重新加载数据。所以这不是一次性的，这是你的日常。每次你来到这里，它都会刷新你的数据并拉取最新信息。第二点，如果我说错了请纠正我，它只是复用了你连接器中的认证信息。所以不需要什么魔法，我不需要输入 API 密钥，也不用考虑把它存在哪里。如果你进入设置，进入连接器，连接你的东西，你就可以使用任何这些连接器来搭建一个这样的 Live Artifact。看看这个列表，有这么多东西，你可以一键添加，通过 OAuth 登录，然后就可以在 Claude 中构建一个 Live Artifact，而无需索要 API 密钥。

Well, and I want to call out for folks a couple things that they might not notice that I think are really unique about these Claude Live artifacts, which is one, the killer feature here is this little refresh button up in the top, which is you can constantly pull in and reload data. And so this isn't this is your day-to-day. Every time you come to this, it's going to refresh your data and pull the latest in. I think the second thing, and correct me if I'm wrong, is it just is reusing the auth from your connectors. And so there's like no magic I have to put an API key in and where am I going to store it. It's if you go into settings and go into connectors and you connect your stuff, you can use any of those kind of connectors to scaffold out one of these live artifacts. And like look at this list of stuff. There's so many things that you can just one click add in sign in through OAuth or whatever and then you can build a live artifact in Claude without having to like ask for an API key.

完全正确。而且，你知道，它不仅仅是我们连接器商店里的那些。你也可以连接你自己的。这非常强大。我最喜欢这一点的地方——这可能相当独特——是我能注入的创造力，以及你能真正按照自己的形象塑造它，对吧？比如我们可以进去说，我希望这个看起来像是 2000 年代早期制作的软件。

Exactly. And it can be, you know, it's not just the ones we have in our connector store. You can also connect your own. It is quite powerful. The thing that I enjoy about this the most, this is like probably fairly unique to myself, but the thing that I enjoy about this the most is the amount of creativity you can pull in and how you can really shape this in your image, right? Like we could go in and say, I want this to look like it is software made in the early 2000s.

哦。

Oh.

哦，200 年可能太早了。

Oh, 200 is maybe too early.

好吧，我们看看它会怎么做。

Well, let's see what it does.

我的意思是，老实说，这有点瑕疵。Claude 可能足够聪明，知道我说的是 2000 年代。

I mean, honestly, this is flawed. Claude is probably smart enough to know that I meant the 2000s.

嗯，这是我常用的 Claude 设计用例——我本月早些时候做了一期关于 Claude 设计的节目，我用 1999 年 GeoCities 的风格重新设计了 Lenny 的新闻通讯。效果令人难以置信。我们会在节目笔记中放链接。它做得无可挑剔。我告诉人们的是，真正让我印象深刻的甚至不是设计，而是文案如此……

Well, this is my go-to Claude design use case, which is I did an episode on Claude design earlier this month, and I redesigned Lenny's newsletter in like 1999 GeoCities style. It is incredible. We'll put the link in the show notes. It did an impeccable job at this. And what I told people is, and it wasn't even the design that was really impressive to me. It was the copy was so...

哦，是 GeoCities 时代那种吗？

Oh, was it like GeoCities era?

是的。是 GeoCities 时代。满分 10 分我给 11 分。文案非常出色。

Yeah. It was GeoCities era. It was 11 out of 10. Exceptional copy.

太棒了。

Beautiful.

我想看看它会做成 2000 年代还是 200 年。200 年会有趣得多。

I want to see if it's going to do 2000s or 200s. 200s would be way funnier.

我们大概已经可以看了。看，第一件事就是 2000 年代。Claude 足够聪明。它知道得太多了。

We can probably look at it already. See, like very first thing is 2000s. Like Claude is smart enough. It knows too much.

知道得太多了。而且，我的意思是，这显然有点像一个演示账户，但如果我给你看我的真实账户，Claude 已经非常习惯我经常拼错单词，并且说话风格略有不同。我很期待这会如何发展。总的来说，我对你表达的创造力感到非常兴奋，尤其是那些不是软件开发人员的人能够表达的创造力，对吧？过去这整个领域是，你可能是一家软件公司的设计师，然后软件开发者会执行你的愿景，或者你选择一种更容易的媒介，比如低代码。但我认为真正让我兴奋的是，我们现在可以吸引来自所有这些不同专业领域的人，他们都可以构建这些 Artifact，无论是演示文稿还是一个直接的工具，按照他们自己的愿景。几周前，我在某个地方读到 Margaret Atwood 写了关于使用 Claude 的文章，我脑海中立刻浮现的想法是：我想看看 Margaret Atwood 写的软件。我想看看她为写自己的书制作了什么样的工具。那对我来说会非常有趣和令人兴奋。好了，Margaret，如果你想来参加 How I AI，我很乐意看看你的 Claude 设置。

Knows too much. And also, I mean, this is obviously like a bit of a demo account, but if I showed you my real account, Claude is like quite used to the fact that I constantly misspell things and like talk in a slightly different style. I'm eager to see how this is going to go. I'm in general quite excited about the amount of creativity you express and especially the amount of creativity people can express who are not software developers, right? Like this used to be this entire domain where you were maybe a designer working at a software company and like maybe then software developers would execute your vision or you picked a medium where it's a little bit easier you can like low code things. But I think what is really exciting to me is that we can now pull in people from all these different expertise areas and they can all build these artifacts, be that a presentation or be like a straight-up tool, sort of in the vision they have. A couple of weeks ago, I read somewhere that Margaret Atwood wrote about using Claude and the immediate thing that popped into my head was I want to see software written by Margaret Atwood. Like I want to see what tools look like that she has made for writing her own books. That would be so interesting and exciting to me. Okay, Margaret, if you want to come on How I AI, I would love to see your Claude setup.

关于这一点，在加载的同时，我们也要让这成为现实。我想指出，因为我确实认为存在一个差距。收听和观看 How I AI 的人技术能力跨度很大。所以有最精通 AI、精通 token 的资深软件工程师，他们真正在推动这些工具的能力边界；也有非常非常早期的用户，他们刚刚以天真的成长心态接触这些工具，说“我需要解决一个问题”。我看到的最大差距——我们在开始录制前也讨论过——不是工具的能力问题。而是人们能否理解几乎任何问题都可以输入这些工具，以及工作流会是什么样子。这就是我们做这个播客的原因。

And on this, while this is we're going to make this happen while this is loading, I do want to call out because I do think there's a gap. People listen to and watch How I AI along a broad spectrum of technical capabilities. So we have the most AI-pilled, float-through-the-tokens intense software engineering folks that are really pushing the edges of what these tools can do and then we have very, very early users who are just coming to this in a sort of naive growth mindset way and saying I need to solve a problem. The biggest gap that I see which we were again talking about before we started recording is it's not the capabilities of the tools. It is literally people being able to understand that almost any problem can go into these tools and what a workflow would be. And this is why we've done this podcast.

在 chatp 网站上，我们列出了接近 200 个来自本播客的 AI 工作流。比如你想用 Claude Code 解决个人问题，这里有我们在播客中看到的 15 个点子。我仍然认为需要这种教育和肌肉记忆的改变：你可以给自己建一个每日看板，可以建一个新生儿睡眠追踪器。嘿，你在搬家，为什么不把所有东西放进一个文件夹，让生活更轻松呢？我觉得人们仍然存在技术和用例之间的鸿沟。所以我的建议是，如果你在电脑上折腾某件事，就让 Claude 去做。你见过其他发现用例的技巧吗？

On the chatp website, we have I think it's now close to 200 AI workflows that we've listed from this podcast. That is just like if you're trying to solve a personal problem with Claude Code, here are the 15 ideas that we've seen on the podcast. And I do still think there needs to be this education and muscle memory change of you can build yourself a daily dashboard, you can build yourself a newborn sleep tracker. Hey, you're moving. Why not throw it all into a folder and then make your life easier as you're trying to close on the house? And I think people still the gap between technology and use case is still pretty broad. So my advice is like if you're futzing with something on your computer, ask Claude to do it. Any other tips you've seen on how to discover use cases?

是的。但在我回答之前，我想强调你说得太对了。很多年前，我在一家叫 Slack 的公司工作了五年，现在它算是个众所周知的聊天工具。人们用 Slack 各有各的原因，但更多是关于他们如何与他人协作组织工作，而不是工具本身。Slack 本身，如果你像使用孤立的邮件一样使用它，其实没那么有用。Slack 真正强大之处在于你接受一个想法：如果你更透明地沟通，让大家看到发生了什么，而不是用邮件线程，而是用一个对任何想了解情况的人开放的频道。这是一种工作方式的转变，也是你如何组织与他人沟通的转变。AI 也是类似的：它开启了如此多的机会，要充分利用它们，你可能需要稍微改变一下工作方式。我认为你说得对，我们整个行业最大的问题之一甚至不是训练模型、让它们变得非常聪明和强大，或者给它们工具把智能转化为输出。我们最大的障碍是如何设计人机界面？如何让人类轻松利用它？现在，我认为它的表现方式是人们在分享使用这个资源的技巧。你给的那个建议很好：每当你做某件事感到烦恼、不享受、没有创意、或者不是你真正想做的核心时，那就是暂停一下，想想能不能用 AI 来做。当然，很多人只是问 Claude 就取得了不错的成功。直接问 Claude 你会如何帮我完成这个任务，它通常能想出很好的点子。

Yes. But before I even give you an answer, I just want to underline how correct I think you are. Many years ago, I spent five years at a company called Slack, which I think is now fairly well known as a chat tool that a lot of people use. People were using Slack for different reasons, but they were all more about the way they organized their work with other people rather than about the tool itself. Slack by itself, if you used it the exact same way you would communicate with emails in a silo, was not all that useful. Slack becomes really powerful if you open it up to the idea of what if you communicate more transparently and people can see what's going on, and instead of having email threads we have a channel that is open to anyone who wants to see what's going on. It's this transformation of work and the transformation of how you actually organize communication with other people. AI is in a similar vein: it opens up so many opportunities that to harness them all, you probably have to change a little bit how you work if you really want to harness them. And I think you're right that one of the biggest problems we have as an industry overall is not even training the models and making them very smart and powerful or giving them the tools that allow them to turn that intelligence into output. The biggest hurdle we have is how do we design the human-model interface? How do we make this very easy for humans to harness? Right now, I think the way it shows up is that humans are sharing these tips on how to use this resource. The tip that you gave was a good one: constantly whenever you do something that you find annoying and you're not enjoying and it doesn't feel creative or like the core of what you really want to do, that is a good time to pause and wonder if there is a way you can use AI to do this. And of course, many people have a decent amount of success just asking Claude. Just ask Claude how would you help me with this task, and it usually comes up with a pretty good idea.

我把这称为反向面试，就是直接进入 Claude，说我是三个男孩的妈妈，我经营了太多生意，我最大的挑战是 XYZ。你能怎么帮我？采访我，然后告诉我你能怎么帮我。这是一个很好的练习，因为人们需要指导，比如你的日历有问题吗？你的孩子参加什么运动吗？所以我确实认为这种反向面试，用 Claude 或任何 AI 工具来引导你发现它能帮你解决问题的方式，是个好方法。另外，趁我们坐在这里，我得再问你一个问题。我认为 Anthropic 和 Claude 产品在这方面非常出名，但我们正在闲聊，因为我们在等 co-work 加载。我觉得你们解决的一件事是让加载变得愉快。那么，请告诉我你的论点：如何让一个带有固有延迟的产品不那么烦人地显示加载？

I call this the reverse interview, which is just going into Claude and being like I am a mom of three boys, I run too many businesses for my own good, and my biggest challenges are XYZ. How can you help? Interview me and then tell me how you can help me. That can be a really good exercise because again people want the guidance of like do you have problems with your calendar? Are your kids in any sports? And so I do think this sort of reverse interview where you can use Claude or whatever AI tool to sort of solicit out of you the ways it can help solve your problems is a really good one. And then again while we're sitting here I have to ask you another question because I think the Anthropic and the Claude products are so well known for this but we are chitchatting right now because we are waiting for co-work to load and one of the things that I think you all have solved is making loading delightful. So, give me your thesis on how to make a product with built-in latency not so annoying to watch loading.

老实说，最重要的可能是让用户适应异步的概念。我认为大多数人类其实很愿意等待，只要最终得到的质量非常好。这通常是我们能接受的。以 Slack 为例，这是我们早期遇到的一个问题。我参与的第一个 Slack 版本，当你点击发送消息时，它会立即成功或立即失败。问题是互联网本身不太稳定，它是一个复杂的网状结构。所以消息经常失败，我们很快用了一个新版本替代：它会尝试发送一次、两次、三次，只有在几秒后仍然失败时，才会提示说好像有点问题。这和今天 iMessage 的工作方式一样。如果你发短信时正在隧道里，没关系，它会在大约一秒后重试。

Honestly, the most important piece is probably to just get comfortable. Help users get comfortable with the idea that things are asynchronous. And I think most humans are actually quite comfortable waiting for something if the quality that they get at the end is very good. That is generally something we're quite comfortable with. In the example of Slack, this was a problem we had fairly early on. The first version of Slack that I worked on, when you hit send on a message it would immediately succeed or fail right away. The problem is the internet itself is not very stable; it's a complex mesh. So messages would frequently fail, and we fairly quickly replaced that with a version that tries sending the message once, twice, three times, and only after a good couple of seconds if it fails it would be like hey by the way something seems to be off here. This is the same way that iMessage works today. If you send a text message it's okay if you're currently in the tunnel; it's going to try again about a second later.

这太美了。他让我把这个放大到心里。

This is beautiful. He let me make this big to my heart.

这很好。这非常好。

This is good. This is very good.

我喜欢我们真正感到高兴的时刻……

I love how I am in moments where we're real delighted about...

是的，我对这些东西很着迷。我有点痴迷于 2000 年代初的软件。不过，这很美。我非常喜欢。

Yeah, I'm such a nerd for this stuff. I'm mildly obsessed with software from the early 2000s. But yeah, this is beautiful. I'm a big fan of this.

这相当不错。我喜欢它引入了……我在思考笔记中看到，Felix 要求 Spotify，但我们是在 2000 年代初。我们应该谈谈 Winamp。

This is quite good. I love that it brought in... I saw in the thinking notes that it was like Felix requested Spotify, but we're in the early 2000s. We should talk about Winamp.

Winamp。我知道。我正想说你需要在这里给 Winamp 模块换皮肤。

Winamp. I know. I was going to say you need to skin your Winamp module here.

这很美。好了，我来总结一下我们刚才聊的，也就是加载屏幕。适应延迟。我认为要意识到任务何时不需要同步，并将这个概念融入你的产品。然后你看，我们刚闲聊了一会儿。当我们谈论延迟时，Claude 一定在听，它说，我要给他们看看。我要把它直接打开。再次提醒大家，注意细节。看右上角。时钟在动，它在滴答作响。你有这个刷新。

This is beautiful. Okay, so I'm going to wrap what we were talking about, which is loading screens. Get comfortable with latency. I think realize when tasks don't need to be synchronous and build that concept into your product. And then look at that. We just chitchatted for a while. And as we were talking about latency, Claude must have been listening and said, I'm gonna show them. I'm gonna bop this right open. And again, folks, not paying attention to the details. Look in the top right. The clock is moving. It's ticking along in the seconds. You have this refresh.

我每天都想打开看看这个。它很有趣，你可以想象把它塑造成任何你想要的形状。但我会再给你一个关于如何实际构建这些东西的建议：你得等一会儿，人们需要等待它们，并逐渐适应等待。我认为对我们 Anthropic 来说，或者对我作为一个产品构建者来说，我总是会权衡多花一点时间，给你一个更好的结果。我认为最终这对人类更好。我可能想教给人们的是，不看着 Claude 实际工作也没关系，因为那并不是我对 AI 的愿景：我们人类只是坐着无所事事，看着 AI 做我们本来可以自己做的事情。我总说，如果 AI 只是需要为你移动鼠标光标，那它就被用得很糟糕。我真正想做的是让 AI 在后台做一堆烦人的事情，解放你的创造力，这样你就能想出像把每日计划表做成 2000 年代风格一样好的点子。你会比我有更好的想法。但我认为这其实是让人们从必须看着它，转变为更舒服地不去看它。我认为这更多关乎信任，而不是耐心。

This is I want to open up and look at this every day. It's pretty fun. And you can envision you can shape this in any form you want. But I'll give you one more tip on how to actually build these things: you've got to wait a little bit, and people need to wait for them and get a little comfortable with waiting. I think for us at Anthropic, or maybe for me as a product builder, I will always trade off taking a little bit longer and giving you a better result. I think ultimately it's better for humans. And the thing I probably want to teach people is that it's okay to not watch Claude as it's actually working, because that is not really the vision I have for AI: that we humans just sit around and do nothing and watch the AI do the thing that otherwise we could just do ourselves. The thing I always say is that AI is used poorly if it just needs to move the mouse cursor for you. What I really want to do is have AI do a bunch of annoying things in the background to free you up for your creative energy, so you can come up with ideas as good as making a daily planner look like it was built in the 2000s. You will have better ideas than me. But I think it's about moving people from having to watch the thing into a position where they get more comfortable not watching it. And I think this has more to do with trust than patience.

我喜欢这个。我们还有最后一个关于这个主题的小演示，就是你怎么知道 Claude 完成了，这其实是我联系你做的原因，因为我太喜欢它了。但跟我们讲讲那个小棒子吧。

I love it. Well, we do have one last little demo that you're going to show us basically on that topic, which is how you know when Claude's done, and this is actually the thing that I reached out to you about doing because I love it so much. But talk to us about the little stick.

是的。为了真正发挥创意，并更适应自己不擅长的领域，我一直对这些小硬件设备有点着迷，而我根本不是硬件开发者。我从未真正焊接过任何东西。我完全不是硬件开发者。但我想到我可以直接上网自己买所有这些组件。昨天我在推特上看到有人说：“嘿，如果你想自己造一个电子书阅读器，组件其实很便宜，也很容易买到。” 然后我想到我可以把这个设备连接到 Claude。我们谈了很多关于协作的事情，但我们也有 Claude Code，对吧？所以我做了这个小东西，就是一个非常非常小的 Claude 在一个小棒子上，这个棒子有 Wi-Fi，有蓝牙。让我告诉你这是什么。互联网是个神奇的地方。在互联网上，你可以自由选择，想买什么就买什么。现在你可以买到这些非常小的硬件设备，带有完好的 LCD 屏幕、LED 屏幕。这个东西有 Wi-Fi，有蓝牙，有一点磁盘空间。你可以编程它。你确实需要用低级语言编程，但你不必，对吧？嘿，我们生活在未来。你不需要做那个。谁在乎呢。所以，你可以直接把这个小棒子插到你的电脑上。我把它插上，然后我所做的就是告诉 Claude Code：“这是我的愿景。我宏伟的创意愿景。我想让我的小 Claude 活在这个东西上，我希望它每次我做好工作时都为我加油。而且每次我需要批准 Claude 正在做的事情时，我希望它在这个大按钮上。” Claude 一次性就构建了所有东西。我完全不需要纠正任何东西。后来我在网上看到很多其他人也做了类似的东西，因为我开源了我的代码，所以人们可以玩它。但我也看到这只是冰山一角。还有更多很酷的设备。顺便说一下，这个东西只要 19 美元。它有 Wi-Fi，有蓝牙。我已经提过很多次了，但我无法摆脱这个想法。我们一直在谈论如何优化我们在电脑上的生活，对吧？因为那是我们生活的地方。但我一直有这个想法，就是可以把 Claude 释放到所有这些小硬件设备上，并改善我的——改善的意思是让它更令人愉快或更可爱。我在最近一集里做了非常类似的事情。我买了一个小型复古电脑，非常像 Windows 2000，它预装了专有软件。我不得不破解那个东西。我在 iPhone 应用和设备之间进行蓝牙嗅探。我转储了所有日志，反向工程了文件编码到设备中。最后我成功了，给自己做了一个小 CLI 工具，我输入文字，它就会显示出来。

Yeah. So, in the interest of getting really creative and being more comfortable with areas you're not very good at, I have always been mildly obsessed with these little hardware devices, and I'm just not a hardware developer. I never really soldered anything together. I'm not a hardware developer in any shape or form. But it occurred to me that I could just go on the internet and buy all of these components myself. I saw a tweet yesterday from someone who said, "Hey, if you want to build your own ebook reader, the components are actually quite cheap and very available." And it occurred to me that I can just connect this device to Claude. We talked a lot about co-work, but we also have Claude Code, right? So, I have built this little thing which is just a teeny tiny Claude on a little stick, and this stick has Wi-Fi, it has Bluetooth. So let me tell you what this is. The internet is a magical place. And on the internet you can just exercise free will and buy whatever you want. And what you can now buy is these teeny tiny hardware devices with a perfectly good LCD screen, LED screen. This thing has Wi-Fi, has Bluetooth, it has a little bit of disk space. And you can program it. You do need to program it in a low-level language, but you don't, right? Hey, we live in the future. You don't need to do that. Who cares. So, what you can do is you can just straight up connect this little stick right here to your computer. I just plug this in and then all I did was I told Claude Code, "Here's my vision. Here's my grand creative vision. I want my little Claude to live on this thing, and I wanted it to cheer me on every single time I do a good job. And also every single time I need to approve something that Claude is doing, I want it to be on this big button that is out here." And Claude built all of that in one shot. I needed to correct absolutely nothing. And I've since seen on the internet that a lot of other people have built similar things because I open sourced my code so people can play with it. But I've also since seen that this is just the tip of the iceberg. There are so many more cool devices out there. This by the way, this thing is $19. It has Wi-Fi. It has Bluetooth. I've already mentioned this many times, but I can't get over the idea. We've been talking a lot about how we optimize our life as it happens on a computer, right? Because that's where we live. But I keep having this idea that it can just unearth Claude onto all these little hardware devices and also improve my — by improve I mean make more delightful or make a little cuter. I did a very similar thing in a recent episode. I bought a tiny retro computer, very Windows 2000, and it was already preloaded with proprietary software. I had to hack into that thing. I was Bluetooth sniffing between the iPhone app and the thing. I was dumping all the logs and reverse engineering the encoding of the files into this thing. And I finally got it and made myself a little CLI tool where I can type and it just shows up.

是啊。但你看，这太酷了。我们现在就能做到这一点太酷了。就像这个东西也通过蓝牙通信。我在构建它的时候，通过耳濡目染学到了很多关于蓝牙协议的知识。你想看看吗？

Yeah. But see, that is so cool. That is so cool that we can do that just now. Like I this thing also communicates over Bluetooth. I learned a lot about the Bluetooth protocol as I was building this just by osmosis. Do you want to see it?

我确实想看看。

I do want to see it.

好的。所以这个现在也部分实现在实际的 Claude 中。你可以构建非常类似的东西，只要你买任何你选择的 IoT 设备。我不做推荐，但你在网上找到的任何东西，只要有蓝牙，可能都能用。但你要进入帮助故障排除，启用开发者模式。开发者模式允许访问开发者工具和功能。我们给你那个警告，因为它确实需要你稍微知道自己在做什么。它会重启应用程序。好了。然后你会看到这个新的开发者菜单，我们把所有开发者功能都藏在这里。这是给使用 MCP 的人，或者想自己构建小硬件设备的人用的。这里面现在有一个新的开放硬件伙伴。你会注意到这个东西看起来很像我真的买的那一个。但它不一定是同一个。我想我已经说清楚了。跟随你的内心。要有创意。你可以点击连接，它会扫描这个小东西几秒钟，通过蓝牙找到它。它找到了。好了。现在它连接上了。它给了我一个密码。好的，酷。我们现在连接上了，Clara，你告诉我你是否能看到任何东西。

Okay. So this is now also partially live in actual Claude. So you can build something very similar if you buy any IoT device of your choice. I'm not going to make a recommendation, but anything that you find on the internet will probably work as long as there's Bluetooth. But you go into help troubleshooting, you enable developer mode. Developer mode allows access to developer tools and features. We give you that warning because it does require that you sort of know a little bit of what you're doing. It's going to restart the application. Here we are. And then you have this new developer menu where we hide all of our developer features. This is for people who work with MCPs or also for people who want to build their own little hardware devices. And in here is now a new open hardware buddy. This thing you will notice that it looks a lot like the one I've actually purchased. But it doesn't have to be the same one. I think I made that point clear. Follow your heart. Be creative. And you can hit connect and it's going to scan for this little thing for a few seconds to find it on Bluetooth. And it's found it. There we go. So now it's connected. It gives me a passcode. Okay, cool. So we're now connected and Clara, you tell me if you can see anything.

所以也许我们要做的是，我们会请求任何需要权限的操作。我就说，创建一个 hello world.txt 并写入磁盘。这样做的目的就是触发一个需要权限的操作。

So maybe what we're going to do is we're going to ask for anything that would require any kind of permission. I'm just going to say just make hello world.txt and write it to disk. The point of this is to just fire something off that will require a permission.

好的，你现在可以看到……我听到了。是不是很可爱？是不是非常可爱？真的吗？

Okay, so you can now see... I heard it. Isn't that cute? Isn't that very cute? Really?

如果我在这里批准，它就会被批准。就是这样运作的。现在 Claude 开始做自己的事了。

And if I approve this here, then it will be approved. That's how it works. And now Claude's off doing his own thing.

我喜欢。

I love it.

我们把最好的留到了最后，各位。如果你能坚持听到现在，那我们把最好的留到了最后。Felix，老兄，我真的很喜欢。我和你一样，我是一个还算称职的软件工程师，但对蓝牙协议和硬件完全不懂。这一直是我避而远之的东西。而且我觉得，随着我的孩子长大，你会更欣赏这一点。我不想让他们接触完整的电脑，但我很希望有一些小东西让他们互动。我认为，能够像我们获得个性化软件一样，将微型设备限定在特定用例上，会非常有趣。

We saved the best for last, people. If you have made it this long in our episode, we have saved the best for last. Felix, man, I really like it. And I'm like you. I'm a perfectly serviceable software engineer and I have no idea what I'm doing with the Bluetooth protocol and with hardware. And it's just been a thing that I've stayed so far away from. And you'll appreciate this, I think, more as my kids get older, too. I don't want them interacting with the full computer thing, but I would love to have little things for them to interact with. I think being able to scope micro devices to a use case just like we're getting this personalized software is going to be really fun.

是的。我的工作中最令人愉悦的一点是，人们经常给我发来他们孩子用 Claude 做各种事情的小视频。孩子们的创造力很美，对吧？你可以拍下孩子画的小动物，交给 Claude 说，我需要一个跳跃闯关游戏。我需要一个 Flappy Bird，但把鸟换成这边画的这只小鸟。然后你把孩子放在电脑前，真正神奇的事情就发生了。Claire，我还在消化这件事，但孩子们身上正在发生真正神奇的事情，因为他们从未学会什么不该问。对他们来说，电脑就是什么都能做。而我们这一代人，已经习惯了事情就是不好使。我们习惯了说“哦，这个做不了”。所以我们就像活在一个思维牢笼里，已经 20 年了。

Yeah. And one of the pieces that's been maybe most joyful about my job is that people frequently send me little videos of the kinds of things their kids are doing with Claude. And the creativity they have is beautiful, right? You can just take a photo of a little animal your kid has drawn, give it to Claude and be like, I need a jump and run. I need a Flappy Bird, but instead of the bird, it needs to be this little drawn bird over there. And then you put your kid in front of the computer and truly magical things happen because I think I'm still wrapping my head around this, Claire, but a truly magical thing is happening with kids because they've never learned what not to ask for. To them the computer is just like can do anything. And I think to our generation we're very used to things just not working right. We're used to like oh you can't do that. So we've been living in this mind prison for like 20 years.

这非常符合我们的风格，但我九岁的孩子是 Claude 的日常活跃用户。他决定要深入网络安全。我们这里非常黑客风格。所以他一边用 Claude，一边开着终端。他说，你知道你的设备 ID 是这个吗？我说，知道啊宝贝，我知道，但干得好，干得好。嗯，我觉得这太迷人了。Felix，我知道这期节目我们玩得很开心。我觉得大家会从中收获很多。简单回顾一下，我们聊了 Claude Co-work 能为你构建 artifacts，既有那种标准的疯狂 3D 打印或你户型图的 3D 渲染，用来解决生活中的小问题，也有能从实时数据中提取信息的动态 artifacts，给你丰富的数据个人体验。我们触及了很多话题，包括将电子邮件作为真实数据源，如何处理延迟，以及如何选择足够智能的模型——就是你知道什么时候需要 Opus，什么时候需要 Sonnet，但我们都认为 Sonnet 在大多数事情上表现不错。然后你向我们展示了如何以 20 美元包邮的总价破解硬件，为自己搭建小伙伴。我要放你走了。一个快速的闪电问答。我们在中间环节也做过，所以你快速回答就好。当 Claude 不听使唤时——我们聊过如何引导它走上正轨——但当它真的犯错时，不是不犯错，而是犯所有错。你的提示策略是什么？我想象不出你会吼它。你不像会吼的人。但你会怎么做？

This is very on brand but my nine-year-old is a daily active Claude user. But he's decided he's really into cyber security. We're very hacker adjacent here. So he's truly in Claude on one thing, the terminal on the other. He's like, do you know that your device ID is this? I'm like, yeah, babe. I do know that, but good job. Good job. Yeah, I think it's totally fascinating. Well, Felix, I know we have just been having so much fun this episode. I think people are going to get so much out of it. Just to reverse for folks, we talked about Claude Co-work being able to build you artifacts, both sort of like standard crazy 3D printed or 3D renderings of your floor plan and other things that you can just stand up to solve little problems in your life all the way to live artifacts that can pull from your live data and really give you a rich personal experience with your data. We touched on a lot of topics including email as a source of truth data. What we're going to do with all this latency and how to pick the smart enough model which is just say you know when you need Opus and you know when you need Sonnet but we both think that Sonnet's pretty good at most things. And then you showed us how we can hack into hardware for the all-in price of 20 bucks including shipping and build little buddies for herself. I'm going to get you out of here. One quick lightning round question. Again, we did this at the midpoint, so maybe you can just answer really quickly, but when Claude is not listening, we've talked about how to set it off on a good path, but when it's really just making mistakes, instead of making no mistakes, it's making all the mistakes. What is your prompting strategy? I can't imagine you yell. You don't seem like a yeller. But what do you do?

是的，我不会吼。我觉得知道人们什么时候对 Claude 爆粗口对我很有用。作为一个产品构建者，这对我来说是个有趣的信息。但当 Claude 真的不理解情况，或者似乎偏离了轨道时，我的策略通常是直接调试我的工作流：问它，好吧，这是我实际期望的。你能告诉我你在哪一步预期了不同的东西，或者我们将来如何避免这种情况？这并不总是有效。Claude 是 AI，AI 会犯错。不过，话虽如此，我经常能找到改进数据源或减少数据噪声的方法，或者和 Claude 一起想出机制。比如，我们如何设置某种试运行？我最终得到的教训通常不是“哦，Claude 做不了这个，我不让它做了”。通常我得到的教训是，我可以稍微改变一下框架、提示词。我可以改变一开始的设置方式。而且，经过我们这段时间的相处，这对你来说并不意外，但我的答案通常是也让 Claude 参与进来，说：“Claude，你觉得我们能做什么？”Claude 是个好朋友，如果人们需要发泄、外化他们的挫败感，Claude 承受得住。

Yeah, I do not yell. I think it is useful for me to know when people curse at Claude. This is an interesting thing for me to know as a product builder. But my strategy for when Claude doesn't really understand what's going on or Claude seems to have gone off the rails, I usually try to debug my workflow straight up by asking: okay here's what I actually expected. Can you maybe walk me through at what point you expected something different or how we can maybe prevent this in the future? It doesn't always work. Claude is AI and AI can make mistakes. However, that being said, I have frequently found ways to improve the data source or to improve the amount of noise that is in the data, or with Claude to come up with mechanisms. Okay, how do we set up some kind of dry run? I usually end up at a place where the learning I take away is not, oh, Claude can't do this. I'm not going to have Claude do it. Usually the learning I take away is I can maybe change a little bit the harness, the prompt. I can change something in how we set this up at the beginning. And I think after the time we had together now, this is no surprise to you, but like usually my answer to that is to also involve Claude and say, 'Claude, what do you think we can do here?' Claude is a good friend and if people need to vent and externalize their frustrations, Claude can take it.

你可以这么做。也许就是猛按按钮，比如弄个愤怒按钮，而不是对着我们的小硬件伙伴说那些话。你可以有生气按钮和开心按钮。你就“我不满意 Claude 砸砸砸”。

You can do it. Maybe just smash the button like have an angry button instead of saying the thing on our little hardware buddy. You can have the mad button and the happy button. You just I'm not pleased Claude smash smash smash.

是的。而且确实有一个“我不满意”按钮，它最终也会出现在我的桌子上，我希望人们使用它。所以每次 Claude 犯错时，我们在许多产品服务中都有小小的点赞和点踩按钮。这是 Claude 全平台通用的。你会找到那些小的大拇指向上、向下按钮。它们对我们非常有用。我们在训练中使用它们，在改进模型和产品时也使用它们。所以也许我想说的是：如果 Claude 没有做你希望它做的事，对 Claude 爆粗口可能不会解决未来的问题。但按下那个小的大拇指朝下按钮对我们很有帮助，因为它确实会出现在我的桌子上。我们会处理它。

Yeah. And there is a 'I'm not pleased' button that actually also ends up on my desk that I do want people to use. So every single time Claude does make a mistake, we do have the little thumbs up, thumbs down buttons in many of our product services. This is across Claude. You will find the little thumbs up, thumbs down buttons. Those are super useful to us. We use those in training. We use them in improving not just the model but also the products. So maybe actually the thing I'm going to say is: if Claude hasn't done the thing that you wanted it to, cursing at Claude will probably not fix it for the future. But pressing the little thumbs down button helps us quite a bit because it does end up on my desk. We deal with it.

完美。好了，这真是太有趣了。

Perfect. Well, this has been super fun.

在你离开之前，我们怎么找到你？除了点赞和点踩按钮，我们还能怎么帮忙？

Before we get you out of here, how can we find you? And other than hitting the thumbs up, thumbs down button, how can we be helpful?

嗯，其实，帮助我们打造好产品的第一要务就是了解什么对人们有效。选项空间太大了。我们仍然处于这个有趣的智能手机时代，我们聊过这个。嗯，我确实想听听大家的意见。我想听听什么对他们有效，什么无效。我一直很感激各种反馈。嗯，你可以在 LinkedIn 和 X 上找到我。这是两个主要平台。

Um, it's really that that's like the number one thing that really helps us build a really good product is getting a good sense for what works for people. The option space is so big. We're still in this like fun smartphone time. We talked about it. Um, and I do want to hear from people. I want to hear from people what works for them, what doesn't work. I'm always grateful for all kinds of feedback. Um, you can find me on LinkedIn and X. Those are the two big places.

太好了。你还做了承诺。你在这次通话前向我承诺过，我们结束后会兑现。嗯，非常感谢你参加《How AI》。这真是太棒了。

Great. And you make promises. You made a promise to me before this call that we will get to after we end. Well, thank you so much for joining How AI. This has been great.

谢谢你，Claire。是的，这很有趣。

Thank you, Claire. Yeah, this was fun.

非常感谢您的收看。如果您喜欢这个节目，请在 YouTube 上点赞和订阅，或者更好的是，在评论区留下您的想法。您也可以在 Apple Podcasts、Spotify 或您最喜欢的播客应用上找到这个播客。请考虑给我们评分和评论，这将帮助其他人发现这个节目。您可以在 howiipod.com 上查看所有剧集并了解更多关于节目的信息。下次见。

Thanks so much for watching. If you enjoyed this show, please like and subscribe here on YouTube, or even better, leave us a comment with your thoughts. You can also find this podcast on Apple Podcasts, Spotify, or your favorite podcast app. Please consider leaving us a rating and review, which will help others find the show. You can see all our episodes and learn more about the show at howiipod.com. See you next time.