Granola 联合创始人如何用 AI 驾驭会议

大家好。今天我非常激动地欢迎 Sam，Granola 的联合创始人。Granola 最初是一个 AI 会议工具，现在他们估值 15 亿美元。所以我真的很想让 Sam 给我们展示他个人如何使用 Granola，不仅仅是会议笔记，还有 Granola 内部如何使用智能体运行。欢迎 Sam。

Hey everyone. Today I'm really excited to welcome Sam, the co-founder of Granola. Granola started as an AI meeting tool and now they're worth 1.5 billion. So I actually really want to get Sam to show us how he personally uses Granola beyond meeting notes and also how Granola runs internally using agents. So welcome Sam.

谢谢 Peter。很高兴来到这里。我一直很期待。

Thank you Peter. Nice to be here. I've been looking forward to this.

我们直接开始吧？我想大多数人使用 Granola 只是为了用 AI 记录会议，但我很好奇你个人如何使用 Granola。你能给我们展示一下你个人的绿灯设置吗？

Why don't we get right into it? I think most people use Granola just to take meetings with AI, but I'm curious how you personally use Granola. Can you show us your personal green light setup?

我变得非常依赖这些会前简报。我觉得这个最有用。我有时会为与新人（比如新候选人）的会议准备不足。如果我在招聘过程中做很多文化面试，而如果我对正在交谈的候选人不够了解，这能多次避免我陷入尴尬。

I've become very dependent on these pre-meeting briefs. I find this most useful. I've kind of underprepared for a meeting with somebody new like often like a new candidate. If I do a lot of cultural interviews as part of the hiring process and if I'm not up to speed on who the candidate is I'm talking to, this saves my ass an embarrassing amount of times.

不错。

Nice.

而且，是的，要让它在你进入会议时提供恰到好处的信息需要很多工作。但如果没有它们，我会感觉有点赤裸，你知道，现在如果我要见新人。

And yeah, like it was a bunch of work to make this like just the right information that you need right as you're going into the meeting. But I feel kind of naked without them, you know, now if I'm meeting something new.

我认为如果 Granola 能真正帮助发现人们没有表达出来的分歧，然后像在会议中那样说“实际上 Sam 和 Peter，你们真的同意这件事吗？”，那将会是下一个层次。

I think it'll be next level if like Granola can actually help suss out like disagreements that people are not expressing and then be like actually Sam and Peter like do you guys actually agree on this thing like during the meeting?

所以，我们的一位投资者有一个非常好的配方，设计为在会议后运行，而不是在会议中，但他经常在会议后使用这个来回顾并进行赛后分析，或者弄清楚“哦，我错过了什么，我应该问什么，我如何在对话中更深入？”是的，我认为这种事情非常强大，就像销售团队在销售电话中也会这样做，你知道，会议后他们会想“我本应该做得更好？”那就是 Granola，你知道，它有你之前销售电话的历史记录，所以有时它能给出相当聪明的答案。

So, one of our investors has a really good recipe which designed to run like after the meeting not really during the meeting but like he uses this a lot I think after meetings to kind of like go back and do like a post-match analysis or and figure out like oh what did I miss what should I have asked how could I have gone deeper during that conversation yeah I think that that kind of thing is really powerful like a sales team will do this as well on a sales call you know they after the meeting they'll be like What should I have done better? That's ang granola and you know it has the history of previous sales calls you've done and so it can sometimes give a pretty smart answer there.

明白了。你是不是只是在这里记一些笔记，然后 Granola 填充其余的笔记，然后你之后应用一个配方，或者应用某种模板？

Got it. Do you just like you know take some notes here and then granola fills in the rest of the notes and then you apply a recipe to it afterwards or like you apply some sort of template to it afterwards.

对我来说，自动笔记在 95% 的情况下都能解决我的问题。我会使用配方之类的东西，比如我经常使用这个“我错过了什么”功能，如果我在会议中 multitasking 或者走神想其他事情，我会用它来回到正轨。我们正在讨论将其产品化。比如你这里有这个 nump 东西，它有实时转录，我想让它在你悬停时立即告诉你错过了什么，因为这是最常用的配方，人们一直在用这个。

For me like the automatic notes sorts me out like 95% of the time. The things I will use like recipes and stuff for is I use this what did I miss thing a lot like if I'm yeah I shouldn't be but if I'm end up multitasking during the meeting or you know zone out thinking about something else I'll be reaching for this to to get me back on track. We yeah, we're talking about adding like making this more of a more of a productized feature. Like you have this nump thing over here which has a live transcript and I kind of I kind of want to just make it so that you hover over this we we tell you what you missed like straight away cuz it's it's by far the most used recipe like people cook on this thing all the time.

是的。我不知道我是否应该承认这一点，但有时我在面试产品岗位的候选人，他们给我一个答案，然后我可能不应该承认这一点，但我喜欢用 Granola 双重检查，比如“嘿，这是一个好答案吗？”

Yeah. I don't know if I should admit this, but like sometimes I'm interviewing a candidate for like product too and um they're giving me an answer and and then I probably should not admit this, but like I like to double check with granola like hey, you know, is is this a good answer?

比如他们在说什么？

Like what are they saying?

是的，特别是如果你事先给它一个很好的模板，比如你试图评估的评分标准。

Yeah, especially if you like give it like a pretty good template beforehand of the rubric that you're trying to assess.

实际上它相当不错。

Like it's it's actually pretty pretty good.

我认为对于有更多结构化会议的人来说，他们非常依赖模板。就像你知道当你有一些特定问题要问时，他们会确保笔记以结构化的方式呈现。

I think for folks who have more kind of structured types of meetings, um they lean on the templates a lot for this. like it's a you know when you when you know you have the certain questions you want to ask they'll be like making sure the notes come out in a structured way and

是的，能够做可重复的事情很重要。

yeah being able to do a repeatable thing is important.

嗯，你能谈谈一些，因为你是设计师出身，对吧？还是我错了？

Um, can you talk about some of the cuz you're a designer by trade, right? Or am I wrong or?

是的。是的。

Yeah. Yeah.

所以，你们在构建这个应用时有一些产品原则、设计原则吗？因为它看起来非常简单，我相信其中有很多权衡，对吧？

So, do you guys have some like product principle, design principles when building this app? Cuz it's like very simple and like there's I'm sure there's a lot of trade-offs that went into this, right?

我们，我不知道。我们大约一个月前写下了这些，所以它们还很新鲜。

We like I don't know. We we wrote these down like a month ago, so that they're pretty fresh.

好的。

Okay.

就像编纂成文，你知道我们有更多新人加入。我们试图为每个新加入的工程师明确，在他们思考 Granola 时，应该记住哪些关键点。我们最终确定了三条。我认为我们希望人们在使用 Granola 时感受到的第一情绪或感觉是平静。Granola 是为生活中极其忙碌的时刻设计的。我们在思考产品时，主要关注的是连续会议之间的混乱间隙。你知道，你可能会在会议室之间匆忙赶路，可能迟到了，可能忘记了要跟谁说话。你可能还有一些快速需要处理的事情，然后就要进入下一个会议。Granola 试图插入那个时刻，那个非常混乱的时刻。所以，我们真的试图坚持这个理念：Granola 应该降低你的血压，而不是增加混乱。它应该感觉像是在帮助你。是的，我的意思是，这样一个空白画布很大程度上体现了这一点。我认为我们在这里做出了一些激进的设计决策。比如转录被隐藏得如此之深，很多人觉得这很荒谬，老实说，我们收到很多投诉，人们甚至没有意识到转录在那里。我们为让它这么小而付出了代价。但我认为我们真的想剥离一切不是你在会议中需要放在眼前的东西。

Like codify like you know we have more new people joining. we were trying to cook by like what um for every new engineer that joins like what what few things should they have in their head you know as we like think about granola and um these are the three we got to I think the number one emotion or feeling we want people to feel is calm when when they use granola we design granola is designed for like incredibly busy point like moment in your life I think the main time we hold in our heads as we're thinking about the product is like the messy in between moments between back-to-back meetings you know where you're like rather your windows or might even be rushing between rooms you're probably running late, probably forgotten who you're going to be speaking to. Uh you might also have a couple like quick things you need to get on top of before you jump into the next meeting and granola trying to like insert itself into that moment like really chaotic moment. And so yeah, we really like trying to hold ourselves to this idea of like granola should looking at granola should lower your blood pressure. It should not add to this chaos. It should feel like it's uh it's helping you there. Yeah, I mean the fact that such a blank canvas is like a um a lot a lot of that and like the I think we've we've made quite a few aggressive design decisions here. I think the fact that the transcript is so hidden a lot of people find kind of ridiculous and honestly we get a lot of complaints that people don't even realize the transcript is there like we pay a price for having be that small. Um but I think uh we really really want to like just strip away everything that's not what you need to have in front of you during the meeting.

是的，我觉得这很棒，因为开会时你想专注于对方，对吧？你不想盯着实时转录。

And yeah, I think that's super because when you're having a meeting, you want to focus on the other person, right? You don't want to be looking at a live transcript.

是的，我们有个比喻，Granola 应该像手写一样——这是描述 Granola 在你生活中角色的一种方式。它绝对不是主角。你才是主角，努力完成工作，专注于会议，决定需要决定的事情，或者讨论你要讨论的事情。Granola 真的就在旁边支持你，帮助你变得更好。

Yeah, we had this metaphor that Granola should feel like a handwriting—it's one way of talking about Granola's role in your life. It's definitely not the main character. You're the main character, trying to get your work done and be present in this meeting, deciding the things you need to decide or talking about the things you're going to talk about. Granola is really there on the side to support you and help you be better.

是的，我真的很喜欢这个比喻。它让你保持谦逊。

And yeah, I really like this metaphor. It keeps you humble.

是的，是的。不，这太棒了。我完全同意。比如连续开会，你试图在最后五分钟记笔记，在 Slack 上发送一些要点，就像疯狂的冲刺，对吧？然后你必须切换上下文去参加另一个会议。所以摘要和高级简报对此非常有帮助。

Yeah, yeah. No, it's fantastic. And I totally agree. Like back-to-back meetings, you're trying to jot down notes in the last five minutes, trying to send some takeaways on Slack, and it's just like a mad rush, right? Then you have to context switch to the other meeting. So the summaries and the premium briefings are super helpful for that.

是的，是的。

Yeah, yeah.

本集由 Google AI Studio 赞助，这是构建并与真实用户分享想法的最简单、最快速的方式。例如，我使用 AI Studio 为这家面包店创建了三个不同的着陆页，使用了 Google 内置的 nano banana 模型生成的图像。一个突出新鲜面包，另一个突出美味的蛋糕，最后一个突出面包店的故事。我可以轻松发布网站并获得自定义 URL，在这个例子中是 corgi confections.ai.studio，一键与潜在客户分享所有三个变体以获取反馈。当我还是产品经理时，Google AI Studio 是我构建此类原型的首选工具。它确实是一种简单且免费的方式，可以与你的团队、客户和其他所有人分享你的想法。请在 astudio.google.com 查看。现在，回到我们的节目。

This episode is brought to you by Google AI Studio, the easiest and fastest way to build and share your idea with real users. For example, I use AI Studio to create three different landing pages for this bakery using images generated by Google's built-in nano banana model. One highlights fresh bread, another highlights awesome cakes, and the last highlights the bakery story. I can publish the website easily and get a custom URL, in this case corgi confections.ai.studio, to share all three variations with potential customers for feedback in one tap. When I was a product manager, Google AI Studio was my go-to tool for building prototypes like this. And it really is an easy and free way to share your ideas with your team, customers, and everyone else. Check it out at astudio.google.com. Now, back to our episode.

另一件我觉得很有趣的事情是，你们刚开始时，只是做会议记录，对吧？但现在，我记得 Chris 之前说过，会议是公司拥有的最有用、最新的背景信息，以某种形式，对吧？所以现在我觉得这里有一个巨大的机会，可以演变成一个完整的——我不知道——像维基百科或整个公司的上下文平台。

Another thing I think is really interesting is when you guys started, it was just like taking meeting notes, right? But now, and I think Chris said this before, meetings are the most useful and up-to-date context that a company has in some shapes, right? So now I feel like there's a big opportunity here to just evolve into a whole—I don't know—like a wiki or a contextual platform for the whole company.

是的，是的。

Yeah, yeah.

这似乎是你们正在走的方向，对吧？或者你能多谈谈这个吗？

And that seems like the direction you guys are going down, right? Or can you talk more about that?

是的，是的，当然。我觉得这有点——我们觉得这对我们来说是一个机会，但我不认为我们真正意识到它会变得多大。但我认为，当你日复一日地使用 Granola 处理所有会议时，它会积累大量关于你是谁、你生活中的人是谁、你在做什么项目、你的公司正在发生什么的上下文。而且不仅仅是高层次的要点，而是真正的细节。Granola 可以告诉我哪些项目可能很快会出问题，或者我没有给予足够关注的事情，而且它在转录中有足够的细微差别和细节，能够给出聪明的答案。

Yeah, yeah, for sure. I think this is a bit of a—we feel this would be an opportunity for us, but I don't think we really realized how big of a thing it would be. But I think so much of when you use Granola day in and day out for all your meetings, it just accumulates so much context about who you are, who the people in your life are, what projects you're working on, what's happening in your company. And not just at high-level bullet points, but real nitty-gritty details. And Granola can tell me what projects are likely to be going off the rails soon or what things I'm not giving enough attention to, and it has enough nuance and detail in the transcripts to be able to give smart answers to that.

我会设置像会前、会中、会后这样的流程。你经常运行一些配方吗？或者你通常用所有这些上下文检查哪些东西？

I'll set up like just like the pre-meeting, meeting, and post-meeting. Do you run some recipes like on a regular basis or like what are some stuff that you kind of like, you know, check with all this context?

是的，我会用配方来做非常具体的事情。所以，是的，如果我要——举个好例子——但比如，如果我只是增强笔记——

I do. I will use recipes for quite specific things. So, yeah, if I'm gonna—let's get a good example—but like, let's say if I just enhance the notes—

试图内部做一个关于我们设计原则的演示。你能为我做一个幻灯片大纲吗？

Trying to make a presentation internally about our design principles. Can you make a slide deck outline for me?

我经常使用对话主体来起草东西。今天早些时候我还在为一个新职位写职位描述。我用过这个——博客文章的方式——比如新功能公告的博客文章，我们所有人开会讨论我们认为博客文章中需要包含什么，然后 Granola 有这些，但它也有整个项目的历史可以借鉴。所以我觉得这对那种事情非常有用。

I'll often use the conversation body as a way to draft things. I've been doing this with a job description just earlier today for a new role we're looking for. I've used this—what are the ways a blog post—like blog posts with new feature announcements have been written by us all having a meeting about what we think needs to go in the blog post, and then okay, Granola has that, but it also has the whole history of the project to draw from as well. So I find it very useful for that kind of thing.

我明白了。那么，如果我想以我的格式写一篇博客文章，我可以制作自己的配方或把一些提示保存在某个地方吗？

I see. And like, if I want to write a blog post in my format, like I can make my own recipe or save some prompt somewhere?

是的，是的。是的。

Yeah, yeah. Yeah.

我这里有一个配方。

I have a recipe on here.

是的，你有。

Yeah, you do.

相当疯狂。

Quite insane.

让我们看看团队中其他人做了什么。

Let's see what other folks on the team have done.

哦，这是来自团队的。不错。这个魔鬼代言人。

Oh, this is from the team. Nice. This devil's advocate.

是的，这些是来自团队成员的。是的。

Yeah, these are from folks on the team. Yeah.

有没有像 Sam 审查配方，比如 Sam 会怎么想或 Chris 会怎么想？

Is there like a Sam review recipe like what would Sam think or what would Chris think?

哦，这是关于头脑风暴产品想法的。明白了。

Oh, it's about brainstorming product ideas. Got it.

是的。非常有趣。

Yeah. Very interesting.

哦，这很有趣。这相当准确。

Oh, this is funny. This is pretty accurate.

这是我们可以展示的最简单版本。我们如何衡量它是否有效？观察双向门。

It's the simplest version we can show. How do we measure it's working? Watch the two-way door.

是的。不错。不错。

Yeah. Nice. Nice.

你把这个开源了。

You got open source this.

是的。是的。

Yeah. Yeah.

明白了。

Got it.

但是你怎么考虑——好吧，Granola 真的让我更能自我意识到我在会议中的表现，对吧？但是你怎么考虑团队方面，比如，我猜有隐私问题，但我怎么知道——比如如果你和我在一个团队，我们一起开很多会，或者你和别人开会，我怎么知道 Sam 是如何运作的？或者也许我是你的下属，你想知道我在会议中如何表现。那里有机会吗？

But how do you think about like—okay, so Granola is really good for me to be a little bit more self-aware about how I show up in meetings and stuff, right? But how do you think about for the team, like, I guess there's kind of privacy issues, but how do I know—like if you and I were on a team and we have a bunch of meetings together, or you have meetings with other people, how do I know how Sam operates? Or maybe I'm your report and you want to know how I show up in meetings. Is there like opportunity there or something?

是的。是的。有——我们显然——我认为 Granola 最令人惊讶的困难之一是在每一刻保护你的隐私。所以以适当的方式。所以 Granola 绝对不能去查看我下属的会议并告诉我他们在做什么，或者他们应该怎么做。

Yeah. Yeah. There's—we obviously—I think one of the surprisingly hard things about Granola is preserving your privacy at every moment. So in the appropriate way. So Granola definitely can't go and look into the meetings of my reports and tell me what's going on in them, or how they should be.

对，但至少不是所有私密内容。但你肯定可以问这类东西，任何被分享出来的内容。我们有团队空间，里面有很多很多文件夹，都是人们的团队笔记。你可以看到站会记录。

Yeah, but not all the private ones at least. But you definitely can ask stuff like that, anything that's been shared. We have our team space, and there are tons and tons of folders in here of people's team notes. You can see standups.

我明白了。好。所以有一个团队，比如设计频道，然后人们主动在这里分享东西，然后任何人都可以查找。是那种——

I see. Okay. So there's like a team, so there's like a design channel and then people proactively share stuff in here and then anyone can look stuff up. Is that kind of—

对。对。这个用得不多。我猜像是设计评审，设计团队用的。销售文件夹每天有很多很多通话。

Yeah. Yeah. So this one isn't used very much. I guess it's like design critiquing, the design team. Like the sales folder has many many calls a day.

好。对。这我相信。

Okay. Yeah. I can believe that.

对。嗯。这个我用得很多。通常你其实不需要去这个文件夹，因为聊天可以直接访问所有内容。但我会去——

Yeah. Mhm. And yeah, this one I use a lot. Usually you actually don't really need to go to this folder because the chat can just access all of it. But I go to like—

因为销售实际上就是客户对话，对吧？所以你可以寻找洞察和趋势之类的。

Because the sales is actually the customer conversations, right? So you can look for insights and trends and stuff.

对，没错。对，我可以问客户在说什么。对，这会查看——不只是我的会议，而是团队最近所有的会议。

Yeah, exactly. Yeah, I can be like what things are customers saying. Yeah, this is going to look across—not just my meetings, but across everything that the team's been having recently.

好，明白了。那有没有办法——也许以后会有——有没有办法设置自动化，让它每周一搜索一次之类的？

Okay, got it. And is there a way to—or maybe this is coming at some point—is there a way to set up automations so that it does a search every Monday or something?

我们——对，我们——我希望很快。我们原型设计了很多这类东西。我觉得我们做到一般水平很容易，但要做好很难。我们目前的总结就是让它开箱即用就很好，输出质量非常高。所以对，我肯定觉得我们会在某个时候实现。对，我们有——因为 API 和 MCP，我们现在有很多用户在用它们创建每周报告之类的。

We—yeah, we—I hope so soon. We prototyped a lot of these things. I think it would be easy for us to do averagely, and I think quite hard to do well. Our summary so far has just been making it really good out of the box, making the quality of the output very good. So yeah, I definitely think we'll get there at some point, someday. Yeah, we have—because of the API and the MCP, we have a bunch of users right now using those to create a weekly report of things like this.

好。

Okay.

对。对。其实我自己就这么做。我用 Granola MCP，它从 Granola 获取信息，也从我的其他来源获取，然后给我发一封邮件总结正在发生的事情。对。

Yeah. Yeah. I do it for myself actually. I use Granola MCP and then it gets the information from Granola but also from my other sources and it sends me an email summary of what's going on. Yeah.

不错。不错。超级有用。

Nice. Nice. Super helpful.

对。对。对。喜欢从人们如何使用 MCP 中学习，我们很多——现在可能只需要弄清楚人们最常用 MCP 做什么，然后把它产品化回应用里。你觉得呢？

Yeah. Yeah. Yeah. Like love to learn from how people use the MCP and a bunch of our—at this point could probably just be like figure out what are people using MCP for the most and productize it back into the app. What do you think?

其实我作为设计师有个问题问你。因为人们只是在使用这些——人们只是让他们的智能体使用这些产品，对吧？就用 MCP 或用 API。然后你怎么看？那么也许你漂亮的 UI 甚至不会展示给人们。你怎么想？作为设计师，你是不是改为设计 API？

I actually have a question for you as a designer. Because people are just using these—people just gain their agents to use these products, right? Just use the MCP or use the API. And then how do you think of that? So then maybe your beautiful UI won't even be shown to people. Like how do you think about that? As the designer, do you design API instead?

想到这个，我作为 UI 设计师的那部分确实会有一丝伤感。肯定的。对。对。老实说——我觉得正因为这个原因，我们抗拒了一段时间。我们当时觉得不行，我们真的需要让所有人都在 Granola UI 里做所有事。我觉得这真的——与越来越大的企业客户合作真的很开放。因为我觉得一旦你成为足够大的公司，你不可避免地会有所有这些内部工具，如果 Granola 不能和这些内部工具很好地配合，你就出局了。不会采用——对，我觉得我们基本上——我觉得现在我们已经接受了这个事实：对于很多工作流，使用 Granola 的最佳方式是摄取上下文并在会议中使用，但之后它会在各种其他应用中驱动工作流，因为它连接到你的内部智能体或代码收费之类的。

There's definitely a pang of sadness in the UI designer part of me when I think about that. For sure. Yeah. Yeah. We honestly—I think because of that reason we fought it for a while. We were kind of like no, we really need to keep everybody in the Granola UI for everything. I think it's been really—it's been really open working with bigger and bigger enterprise customers. Because I think just as soon as you become a big enough company, it's inevitable you have all of this internal tooling and if Granola doesn't play nice with that internal tooling, you're a non-starter. Not going to take—and yeah, I think we basically just—I think at this point we're at peace with the fact that for a bunch of workflows the best way to use Granola is to use it to ingest the context and use it during the meetings, but then it's powering workflows off in all kinds of other apps because it's connected to your internal agent or code for charge or whatever.

对，没错。对。我是说，总会有更普通的用户只用 UI，对吧？但即使对我这样的个体创业者来说，Granola 有超级有用的上下文，但我在 Google Docs 和其他来源里也有很多上下文。所以力量在于把所有这些东西结合起来，帮助 AI 理解一切，你懂的？

Yeah, exactly. Yeah. I mean, there's always going to be more normal users who just use the UI, right? But even as a solopreneur for me, Granola has super useful context, but I also have a bunch of context in Google Docs and other sources. So the power is combining all that stuff together, helping the AI understand everything, you know?

对。对。对。是的。是的。它的强大取决于你给的上下文和你拥有的连接数量之类。对。对。对，我很兴奋能做更多这方面的事。会前简报对我们来说很棒，因为它迫使我们构建了一堆底层基础设施，让 Granola 擅长筛选邮件找到正确的东西。而且对，很多构建方式让我们也能为其他工具做到这一点——Slack、CRM，以及你公司拥有知识世界的所有其他地方，我们都可以查看。

Yeah. Yeah. Yeah. It is. It is. It's powerful as the context you give and the amount of connections and stuff you have. Yeah. Yeah. Yeah, I'm excited to do a lot more of that. The pre-meeting briefs was great for us because it forced us to build a bunch of infrastructure under the hood to make Granola good at sifting through emails to find the right thing. And yeah, a lot of that's been built in a way so that we can make that work for other tools too—Slack and CRM and everywhere else your company has a world of knowledge that we could be looking at.

对，老兄。机会太多了，伙计。你可以做每周简报，比如每周会议简报，或者你可以做会后洞察简报。你可以做太多了——太多简报了。

Yeah, dude. There's so much opportunity, man. You can do a weekly brief like a weekly meeting brief or you can do a post-meeting insight brief. There's so much you can do—so many briefs.

对。会很棒的。

Yeah. It'll be amazing.

对。简报有点像显而易见的第一个用例之类的，你懂的。所以我觉得我们有个风险——如果你是那种去尝试 AI 工具的人，那么你会从四面八方收到简报。这不是常见的首要问题，就像——

Yeah. The briefs is kind of like the obvious first use case or whatever, you know. So I feel like there's a risk of us—if you're the kind of person to go try out the AI tools, then you brief from everybody left, right, and center. That's not a regular first problem like—

对。我觉得——

Yeah. I think—

在我们继续谈产品之前，你知道我最终的——我最终的功能请求是什么吗？

Before we move on to the product, you know what would be the ultimate feature—the ultimate feature request from me?

我最终的功能请求是——这可能不太好，但你和我各自做一个——我有个 Peter 智能体，你有个 Sammy 智能体，然后我们派我们的智能体先开个会对齐，如果它们无法对齐，我们再介入。

The ultimate feature request from me is—this is probably bad, but you and I just make our own—I'll have a Peter agent and you have a Sammy agent and then we send our agents to have a meeting first to align and if they can't align then we come in.

那对你来说太棒了，老兄。

That would be awesome for you, dude.

我知道。人们有个大需求——很多 CEO，我们有很多 CEO 问这类东西，比如你能成为我应对所有问题的第一道防线吗？

I know. People have a major thing—a lot of CEOs, we get a lot of CEOs asking for that kind of thing, like can you maybe be my first line of defense from all the questions I get?

是个很酷的想法。对。

It's a cool idea. Yeah.

对。有点东西。

Yeah. There's something there.

对。好。那么现在让我们换个话题，谈谈你们内部是如何工作的。

Yeah. All right. Well, now let's kind of switch gears and talk about how you guys work internally.

对。

Yeah.

我对这种 AI 狂热有点悲观。感觉很多 AI 原生公司就是 996，不停歇地干活，老兄。作为一个家长，我真的想要那样吗？但我觉得 Granola，也许你们对这件事有不一样的理念。

I'm kind of a little bit pessimistic about this whole AI psychosis thing. It feels like a lot of AI native companies are just working 996 and just working nonstop, dude. And I feel like as a parent, do I really want that? But I think Granola, maybe you guys have a different philosophy on how this stuff works.

是的。说实话，在这方面我真的很喜欢待在伦敦。我觉得团队工作很多、非常努力，他们投入了很多自己。但我认为你应该有工作之外的生活，生活里远不止写代码和做软件。我觉得这非常重要——如果你有一些平衡，你做的工作会更好。

Yeah. I think on that front I really enjoy being in London, honestly. I think the team works a lot and very hard, they pour a lot of themselves into it. But I think you should have a life outside of work, and there's just so much more to life than writing code and building software. And I think it's so important — the work you do is better if you have some balance.

不过这挺难的。我感觉自己完全迷上了编程智能体这台老虎机。每次我打开笔记本，就像，哦，再来一个提示，再来一个提示，再来一个提示。

Been hard though. I feel like I'm fully addicted to the slot machine of the coding agents. Every time I'm on my laptop, it's like, oh, one more prompt, one more prompt, one more prompt.

是的。

Yeah.

是啊，其实是因为我们总说 Granola 帮助在会议之间切换上下文，但实际上在不同智能体线程之间切换上下文更难。你简直每五分钟就要切换一次对话。

Yeah, it's actually because we talk about Granola helping with context switching between meetings, but actually context switching between the different agent threads is even harder. You literally switch conversation like every five minutes.

是的。是的。

Yeah. Yeah.

是啊。所以这挺累人的。

Yeah. So it gets pretty exhausting.

这是我们在设计 Granola 时正在努力解决的问题，因为我们想让 Granola 在会议之外也对你更有用。你能用聊天做的事算是这方面的一点暗示，但我认为我们还能做很多，尤其是围绕会议的后续工作，一旦接入工具之后。但真正我们关注的是：如果一场普通会议会产生三件你可能需要去做或思考的事，而你一天有五场会议，一个很大的立方体就变成了一大堆待办事项。

It's a thing we're grappling with at the point, I think, with designing Granola, because we want to help Granola be better for you outside of meetings. And the stuff you can use chat for is a little hint of that, but I think there's a lot more we could do, especially when you follow up work around meetings, once connected tools. But really our thing there is: if you have three things that come out of an average meeting that you might need to go away and do or think about, and then you have five meetings during the day, a very big cube becomes a big stack of things to do.

是的。

Yeah.

以及你如何把它呈现给用户，或者 Granola 如何帮你处理完这一堆事情，让方式感觉是让你的生活更美好、更好、更平静，而不是只是增加压力，或者在你头上堆越来越多的待办事项。这是件很难想清楚的事。

And how you present that to the user, or how Granola can help you get through that stack of things in a way that feels like it makes your life nicer and better and more calming, rather than just adding stress or pressure and more and more to-dos on top of you. It's a hard thing to figure out.

是的。是的。是的。没错，因为如果 Granola 每场会议都有三个要点，到一天结束就会有 20 个要点。那就太多了。

Yeah. Yeah. Yeah. That's right, because if Granola has like three takeaways from each meeting, by the end of the day there'd be like 20 takeaways. It'd be too much.

是的。有意思。

Yeah. Interesting.

我们想成为你生活中一种让人平静的存在，而不是相反。

We want to be a calming presence in your life, not the opposite.

那你们——因为我觉得现在任何工程师都能直接发布，发布这件事不再难了，或者比以前容易了。所以在某种程度上，公司里几乎每个人基本上都能发布。那么现在，作为一名设计师，你如何让工艺和质量依然保持良好？你如何避免一堆垃圾一直上到生产环境？

And how do you guys — because I feel like nowadays any engineer can just start shipping, the shipping stuff is not hard anymore, or it's easier than before. So in some ways almost everybody in the company can ship basically. And now, as a designer, how do you make the craft and the quality still good? How do you avoid a bunch of slop going to prod all the time?

是的。这很难，因为让 Granola 的每个人都能以某种方式写代码、发布代码有很多好处。很多是内部工具，不进生产环境，但令人惊讶的是也有相当一部分进了生产环境。

Yeah. It's hard, because there's a lot of benefit to letting everybody at Granola write code, ship code in one way or another. A lot of it's for internal tooling, not into production, but a surprising amount is into production as well.

是的。

Yeah.

让任何人都能去做这件事有很大的力量，我想鼓励这一点，而不是让设计成为它的瓶颈。我想我们做过一些事，我觉得在那里真的帮了忙：我们在二月做的品牌重塑，做了很多工作来彻底改造应用的外观和感觉，让它与品牌一致。但在做的过程中，我们也把它建立在一个新的设计系统上。所以幕后一切都是 token 和组件。而且它都是可复用的，不仅对人，对智能体也是。到处教它们如何使用设计系统的 MD 文件也少了。这对那些不是特别重设计、或者不需要特别费心思考的东西帮助很大——它通常出来看着还不错。

There's so much power in having anybody be able to go and do that, and I want to encourage that and not make design a bottleneck for that. I guess some things we've done that I think really helped there: with the rebrand we did in February, we did a bunch of work to overhaul the look and feel of the app, make it feel in line with the brand. But in doing that, we also built that out on a new design system. So everything behind the scenes is tokens and components. And it's all reusable, not just for humans, but for agents too. There's less MD files all over the place to teach them how to use the design system. That's helped a lot for stuff that's not super design heavy or doesn't need thinking about super hard — it usually comes out looking decent.

好的。

Okay.

就靠智能体写代码。很多——我不知道我们是不是在设置菜单里做企业功能。比如你可以直接让 Granola，你直接让编程智能体，你知道，我需要设置里一个新页面，做 X、Y 和 Z，当这个开关关掉时，这样，并且做所有正确的事——它就用那些组件做出来了。

Just from the agent writing the code. A lot of the — I don't know if we're building enterprise features in the settings menu. Like you can just ask Granola, you just ask the coding agent, you know, I need a new page in settings that does X, Y and Z, and when this toggle's turned off, this and does all the right things — it comes out with the components.

是的。明白了。

Yeah. Got it.

是的。是的。

Yeah. Yeah.

是的。对设计师来说，我觉得这是个有意思的时代。Figma 曾完全掌控整个设计行业，你知道，有充分的理由——它就是个很棒的工具。我不知道——我一天 80% 的时间都花在 Figma 里，既画想法草图、迭代不同方案，也做精修和交付设计。而我觉得这一切在过去一年左右被彻底打碎了。老实说现在——我不知道——所有设计师整天用各种各样的工具。比如我们是一家线下办公的公司，我们很多初步思考是在房间里一起用白板做的。我的意思是，现在有时一个白板草图就够交给工程师去构建了，因为有了设计系统，你可以去做那件事，你知道，做正确的事。

Yeah. For the designers, I feel like it's an interesting time. Figma had a complete hold on the whole design industry, you know, for good reasons — it's just a great tool. I don't know where — 80% of my day was spent in Figma, both sketching ideas, iterating on different approaches, and also refining and handing off designs. And I feel like that's just all been blown apart in the last year or something. And it's honestly now — I don't know — all the designers use all kinds of different tools all day long. Like we're an in-person company, we do a lot of initial thinking on whiteboards together in rooms. I mean sometimes a whiteboard sketch is enough now to give to an engineer and go build it, because with the design system you can go and do the thing, you know, do the right things.

是的。然后——我不知道——其实我有个例子。我们有一些稿子之类的，premium briefs，所以我可以讲讲那个。

Yeah. And then — I don't know — I have an example actually. We have some mocks and stuff, premium briefs, so I could talk through that.

是的。我是说，你现在用 Figma 用多少？是 10% 左右，还是你已经不用了？

Yeah. I mean, how much do you use Figma now? Is it like 10%, or you just don't use it anymore?

也许 20%。是的，差不多那样。会有变化。团队里有些设计师会用得多得多，我觉得这取决于项目。比如 Tanya 在做付费墙的时候，你知道，订阅流程、把你升级到商业版之类的。那全都是——那里所有难的设计问题就是，这个流程说得通吗，文案有没有在正确的时刻把想法放进你脑子里？而只靠在画布上看全部内容来迭代要快得多。而且它不重交互细节，也没有太多状态要担心。它就是一个线性流程，难的是文案。而比如 premium briefs 就互动性强得多，也更融入核心流程，所以对 premium briefs 来说，Figma 用得少得多，在代码里做原型多得多。

20% maybe. Yeah, something like that. It varies. Some designers on the team will be a lot more than that, and I think it depends on the project. Like when Tanya was working on the paywall, you know, subscription flows and upselling you to the business tier and stuff like that. That's all — all of the hard design stuff there is like, does this flow make sense, and does the copywriting put the ideas in your head at the right moments? And it's far faster to iterate on that by just looking at it all on canvas. And it's not heavy on interaction detail, or there's not too many states to worry about. It's just a linear flow with copywriting being the hard thing. Whereas premium briefs, for example, was much more interactive and integrated in the core flow, and so it was a lot less Figma and a lot more prototyping in code for premium briefs, for example.

对，这说得通，因为就是几个要点而已，对吧？你得确保这些要点确实写得好。

Yeah, it makes sense because it's just a couple bullet points, right? You got to make sure the bullet points are actually good.

对。

Yeah.

那你们把它放在应用的哪个位置展示？

So where do you show them in the app?

对。这是 Tanya 早期做原型探索的其中一版。我们基本上就是从在房间里对着白板聊这个 brief 该放在哪儿，直接跳到了她做出这个原型。这一版是设想：如果它出现在通知里，内容就从这里来。

Yeah. So this was one of Tanya's early prototyping explorations. I think we basically went from chatting about where the brief could live in a room on a whiteboard or something to her mocking this up. So this was like, what if it's in the notification and stuff comes from here.

你可以让 Cursor 去做，比如说：嘿，我有一段 brief 文本。你能不能把它放进会议前出现在屏幕右上角的通知里？能不能也放到屏幕底部？能不能再放到屏幕侧边那个悬浮的转录指示器里？把这三件事都做了。然后你立刻就能获得多得多的信号，判断这个东西到底会不会让人觉得舒服。

You can ask Cursor to be like, hey, I have this brief text. Can you go put this in the notification that appears in the top right of the screen before a meeting, and can you also please go put it in the bottom of the screen, and can you also go put it in the floating transcription indicator on the side of the screen? Do all three of those things now, and you instantly get so much more signal on whether this is going to feel good or not.

对。然后整个项目就在这样的原型和 Figma 之间来回切换，又回到原型，再回到 Figma。

Yeah. And then this whole project flip-flopped between prototypes like this and then back into Figma and prototypes again, back into Figma.

对。整个过程真的完全不是线性的。

Yeah. It was really not at all linear.

我猜你是想尽可能贴近真实的终端用户体验，对吧？所以这个比静态稿更有互动感。

I guess you want to get as close to the actual end user experience as possible, right? So this feels more interactive than static mocks.

对，对，完全正确。我觉得尤其是像这种东西，它其实很简单，就是一堆文字。至于做得多显眼、放在屏幕哪个位置之类的，有很多细微差别，但你真正要设计的是：当你压力很大、快迟到了、正要进这个会议、还忘了对方是谁的时候，以这种方式接收到这些文字，会是什么感觉。

Yeah. Yeah. Exactly. And I think especially with something like this, it's pretty simple. It's just a bunch of text. And there's a bunch of nuance in how loud you make it and where you put it on the screen and whatever, but really the thing you're designing for is how does it feel when you're stressed and you're late and you're arriving to this meeting and you've forgotten who the person is, to receive these words in this way.

光看别人 brief 的稿子，或者一个编出来的 brief，很难获得那种真实的感觉。真正重要的是，你看的是你自己的数据、你自己的东西，这样才能判断它到底行不行。

It's very hard to get a realistic feeling of that from looking at a mockup of somebody else's brief, or a made-up brief. It really matters that it's your data and your stuff you're looking at to make a judgment call on whether it works or not.

对，这很有道理。

Yeah. That makes a lot of sense.

对。而且 brief 本身写得好不好也很重要，对吧？关键不在 UI，而在于这个 brief 到底好不好。

Yeah. And also how good the brief is also matters, right? It's not about the UI. It's like how good is the brief?

对，完全正确。

Yeah. Exactly.

你们是不是也做另一件事，既能让人快速发布、又能保持更高质量？就是在内部大量吃自己的狗粮，给 beta 用户看、在发布产品前收集他们的反馈，对吧？我猜你们也这么做。

Do you guys do another thing I found to still let people ship but still have higher quality? Just a ton of dogfooding internally and showing to beta users and getting their feedback before you ship the product, right? I assume you guys do that too.

对，我们确实这么做。我们真的很努力地推动，在项目一开始就尽快在应用里做出一个很糙但端到端能跑的版本。这个 brief 我们在项目启动后一周、一周半内就在应用里上线了一版。

Yeah. We do. We really try and push to get, when a project starts, a shitty end-to-end working version of it in the app as quickly as possible. We had a version of this brief in the app within a week, a week and a half of starting the project.

而且它丑得不行，就是放在笔记顶部的一个大框里。数据也不好，写得也不行。但我觉得你就是想让大家日复一日地感受到这东西不够好带来的痛苦。这是一种非常有效的方式，逼着你在意把它做好。

And it was incredibly ugly. It was just in a big box at the top of the notes. The data wasn't good and it wasn't well written. But I think you want people to be feeling the pain of this thing not being good day in and day out. It's a very effective way of making you give a shit about being good.

对，你们大概有一堆内部 Slack 频道在给反馈之类的，对吧？

Yeah, you probably have a bunch of internal Slack channels giving feedback and stuff like that, right?

对，即时反馈。

Yeah. Instant.

好。那我想也许你可以再给我们展示一个东西。Claude 最近推出了 tags，大家都在说，哦，我们从 IDE 转向应用，现在又要转向人们直接在 Slack 里给智能体下命令。我记得你们有一堆内部智能体，对吧？好像有个叫 Nacho 的，或者 Honeybot 之类的。能给我们展示一下吗？

Cool. Well, I guess maybe you can show us one more thing. Claude launched tags recently and people are saying, oh, we're moving from the IDE to the app, and now we're moving to just people giving commands to agents in Slack. I think you guys have a bunch of internal agents, right? There's like Nacho or something, or like Honeybot. Can you show us some of this?

好。先说 Nacho。我们有这么一整套内部仪表盘。这里面有大家构建的各种功能、仪表盘、内部工具，各种各样的东西。我真的很为它骄傲。我觉得它的搭建方式让团队里每个人都能往里面加东西、一起构建。

Yeah. So Nacho. We have this whole internal dashboard thing. This is all kinds of features people have built, dashboard internal tools, all kinds of things. I'm really proud of it. I think it's set up in a way where everyone on the team is adding stuff in here and building it.

你可以看到我们有个小小的“用 Cursor 编辑”功能。如果我打开它，然后说，把这个改成红色。你基本上可以指指点点，说：嘿，我需要改一下这个内部工具，帮我做这件事。它就会去启动一个 Cursor 智能体来执行。所以我们真的在努力降低做内部工具的门槛。

You can see we have this little 'edit with Cursor' do that. If I turn this on and I'm like, make this red. You can basically point and click and be like, hey, I need to change the internal tool to go do this thing for me. And it's got to go spin up a Cursor agent to go do it. So we've really tried to lower the bar to making internal tools.

Nacho 是我们目前使用量遥遥领先的内部工具。它这里是个聊天机器人。大体上说，它连接了我们能给它接上的尽可能多的东西——Slack、Granola、Amplitude、我们所有的日志、网站分析，能接的都接上。然后大家就用它。它主要在 Slack 里使用。

Nacho is by far and away the most used internal tool we have. It's a chatbot here. At a high level, it's connected to as many things as we can give it — Slack, Granola, Amplitude, all of our logs, website analytics, as many things as we can. And then people use it. It's mostly used in Slack.

举几个希望可以安全分享的例子：看看代码，搞清楚代码是怎么运作的，给我们一个聪明的答案。这种事你自己去深挖要花很长时间。

Some examples that hopefully are safe to share: look at the code, figure out how the code works, give us a smart answer. This will take so long to go actually dig into yourself.

这是一次小事故——我记得是摘要生成短暂地掉了一下。频道里一开始是 Toby 问发生了什么。所以它这是在查日志，我猜，去搞清楚这些错误在哪儿冒出来、到底怎么回事。然后它还能拿这些去和代码做交叉验证，搞清楚为什么会发生这个错误。

This was a minor incident — I think summary generation had a little dip for a second. The channel starts out, Toby asks what's going on. So this is looking at logs, I think, to figure out where these errors are popping up, what's going on there. It's then able to sense-check that against the code and figure out why that error would be happening.

我明白了。所以基本上你就是把它接入了公司内部的多个数据源，听起来它还能选你想用的模型，然后它先做第一轮回复，对吧，你还可以跟它对话。

I see. So basically you just linked it up to multiple internal company sources, and it sounds like it can pick the model that you want to use, but then it kind of does the first-pass response, right, and you can talk to it.

对。而且它在 SQL 查询这类事情上非常可靠。所以它能相当——

Yeah. And it's really reliable at SQL queries and things like that. So it can pretty—

你可以比人类通过仪表盘思考更快地提取数据和调查问题。

You can pull data and investigate issues way faster than a human thinking through dashboards would be able to.

是的。

Yeah.

那你们有专人负责这个吗？比如内部工具团队之类的？

And do you have like some guy or some person working on that? So like an internal tools team or something?

呃，没有专人负责。比如 JK，我们的一位工程师，我想他是利用周末或晚上开始做这个项目的。

Uh, nobody owns it. Like, JK, one of our engineers, I think started it as a weekend or an evening project.

嗯。

Yeah.

不，大家只是时不时地往里加东西。比如 Nacho 非常受欢迎，所以如果你能让 Nacho 更好，你就会成为大家最好的朋友。所以，呃，有很强的动力，你知道，如果你有 10 分钟，就试着想办法让 Nacho 对所有人都更强大。

And no, people just add things here and there to it. Like, Nacho is incredibly popular and so if you make Nacho better, you'll be everybody's best friend. So, uh, there's a strong incentive to, you know, if you have 10 minutes to try and figure out a way to make Nacho powerful for everybody.

是的。我觉得这是正确的做法，老兄。因为我觉得，嗯，比如你当然也可以把 Claude 或者 Cursor 引进来，但我觉得公司应该为此构建自己的内部工具。我觉得这并不太难。而且你有更多灵活性，比如如果某个模型变得更好，或者价格太高之类的，你可以随时切换，对吧？

Yeah. I feel like this is the right way to do things, man. Because I feel like, um, like I guess you can also just bring Claude or you know like Cursor in here too, but like I feel like companies should just build their own internal thing for this. It's not too hard to build I feel like. And like you have more flexibility like if someone's model becomes better or like the price gets too high or something you can just switch, right?

是的。灵活性。

Yeah. Flexibility.

是的。而且，我的意思是，我们有，嗯，我们有，我想 Nacho 可以创建 Cursor 智能体。所以经常，你知道，比如做调查往往会导向写一个 PR 来修复问题，我觉得自从我们加入 Cursor 并把它和另一个工具整合后，这个过程顺畅多了。

Yeah. And it's I mean we have um we have I think Nacho can create cursor uh agents. So like it will often you know like an doing an investigation will often lead into a PR being written to fix the thing and I think that's got way smoother since we added cursor and made that not two to two different tools.

是的。所以我想，呃，就像，我想总结一下，你知道如何做到 AI 原生，呃，同时还能让人们有生活，就是，呃，有点像任何人都可以构建，但你有设计系统和组件来确保至少 UI 看起来不错，来自任何工程师，然后还有一堆内部智能体连接到内部系统，让人们可以提问，对吧，那大概就是

Yeah. So I guess uh like I guess just to recap like you know how do you be AI native uh and and still give people a life is is like uh it's kind of like anyone can build but you have like design system and components to make sure that at least the the UI is like looks good like from any engineer and then also having a bunch of internal agents hooked up to internal systems for people to ask questions right that that's kind of

是的，我觉得是这样，是的，然后，呃，是的，是的，不要，信任人们按照他们，你知道，按照他们，嗯，按照他们希望的方式使用时间，就像，而且

Yeah I think so yeah and then uh yeah yeah don't and trust people to use their time as they as you know like as they um as they wish like a and

是的，我的意思是，是的，你不是想确保人们每周工作八小时或一天之类的。你只是想看产出，对吧？看产出。

Yeah I mean yeah like you're not trying to you're not trying to make sure people are working like you know eight hours a week or like a day or something. You're just trying to see the output, right? Look at the output.

是的。

Yeah.

是的。是的。我真的觉得，嗯，尤其是现在，可能比过去更甚，嗯，制造东西太容易了，你知道现在这是陈词滥调了，但制造东西太容易了。嗯，难的是知道该制造什么，而这种思考不是来自一直拼命苦干。可能所有有趣的东西都来自，我不知道，散步、休假，或者，嗯，有空间去思考一些事情。所以

Yeah. Yeah. I really think like um especially maybe now like more than in the old days like um it's so easy to make stuff that you know it's a cliche now but like it's so easy to make stuff. Um the hard thing is knowing what to make and like that kind of thinking comes from like doesn't come from just like grinding as hard as you can the whole time. like the the all probably all the interesting stuff comes from I don't know going for a walk or taking your time off or or um space to go and think about something. So

是的。

Yeah.

是的，那部分仍然非常人性化，老兄。因为比如我试着用 Codex 之类的头脑风暴该做什么，然后 AI 还是太顺从了，老兄。它们就像，你知道，这里有些想法，这个好吗？这是个好主意。你确定这是个好主意吗？不，这不是个好主意。就像

Yeah, that that part is still very human, dude. Because like I I I try to brainstorm like what to make with like you know code codex or something and then like the AI is still too agreeable, man. They just like you know here's some ideas and like is this good? It's a great idea. Is it are you sure it's a great idea? No, it's not a great idea. It's just like

这和与人交谈不一样。

It's not the same as talking to a human.

是的。是的。不一样。

Yeah. Yeah. Not same.

是的。是的。是的。嗯，嗯，Sam，这真是荣幸，老兄。你有没有什么结束建议给 AI 构建者或试图构建，呃，比如 AI 工具或公司的人？我想这是关于弄清楚你想构建什么，也许

Yeah. Yeah. Yeah. Well, well, Sam, this this been a pleasure, man. Do do you have any like closing words of advice for AI builders or people trying to build uh just like AI tools or companies? I guess it is about like trying to figure out what you want to build and maybe

我，我觉得，我的意思是，是的，我真正欣赏我们这里的一点，我主要归功于我的联合创始人 Chris，他灌输给我和公司里每个人的是，嗯，就是，你知道，现在制造东西太容易了。第一，你得制造有趣的东西。你知道，很多困难在于弄清楚什么会是有趣的东西，但第二，我认为大多数人在构建东西时停得太早了，比如他们，你知道，他们做到第一版，就把它推出去发布了。我认为制造东西越容易，就越重要的是真正注重细节，多走一英里，做出对某人真正有用的东西。让自己负责，不仅仅是发布东西，而是这个人是否真的从中获得日常价值。嗯，并且直到实现那个目标才停止，而不是仅仅发布目标。

I I think that I mean, yeah, the thing I the thing I really like appreciate about what we have here, which I mostly credit Chris, my co-founder, with instilling in me and and everyone else in the company is like um it's it's just, you know, it's so easy to build things nowadays. Like one, you got to build something interesting. you know, a lot of the hard think figuring out what's going to be the interesting thing, but two, like most people stop way too early, I think, when in building stuff, like they, you know, they get to the version one and they kind of push it out the door and ship it. And I think the easier it gets to build stuff, the more it matters to like really sweat the details and go the extra mile to make something that's actually going to be really useful to somebody. To hold yourself accountable, not just to shipping something, but to like does the person actually get like daily value out of it. Um, and to not stop until you achieve that that goal rather than just the shipping goal.

我明白了。是的。是的。是的。因为你想发布一些东西来获得市场反馈。你不想只是永远计划，对吧？发布。

I I see. Yeah. Yeah. Yeah. Because you you want to ship some stuff to get some market feedback. You don't want to just like plan in forever, right? Ship.

是的。是的。

Yeah. Yeah.

我们被鼓励尽快在内部发布东西。但是，嗯，我们，你知道，并且向小群朋友和早期用户等非常非常早地发布，但我们确实，我想我们设定了一个非常高的标准，比如这个东西是否真的对人们有用，你知道，在我们经常发布东西之前。

We're encouraged to like ship we ship stuff internally like as quickly as we can. But, um, we, you know, and and to like small groups of friends and baby users and things like that very very early, but we do like I think we bowl like a a very high bar for like is this thing actually working for people, you know, before we put stuff out often.

明白了。嗯，嗯，呃，把我加到那些粗糙的内部构建里吧，老兄。我，我，我觉得我现在用的产品太精致了。把我加到粗糙的构建里。是的。

Got it. Well, well, uh, add me to the janky internal builds, man. I I I think the product I'm using is too polished right now. Add me to the janky builds. Yeah.

是的。是的。是的。是的。酷。嗯，非常感谢，老兄。我觉得从这次聊天中学到了很多。谢谢。

Yeah. Yeah. Yeah. Yeah. Cool. Well, thanks so much, man. I feel like I learned a lot from the chat. Thanks.

这真的很有趣。谢谢。

This is really fun. Thanks.