15项产品经理技能排行：哪些能经受AI冲击，哪些会被颠覆 | 余楠（Linear）

如今你写的一切，真正的受众已不再是人类。因为即便是会阅读你内容的人，他们也在通过AI来阅读。你完全可以一键全选，然后交给AI生成。如果你想的话，甚至可以同时生成100份这样的内容。

Everything you write, the main audience is not really people anymore. Because even the people that are gonna read your stuff, they're reading it using an AI. You can literally just go, like, command a, select all, and then delegate it to. Gen. You can just spawn off 100 of these simultaneously if that's what you want to do.

要知道，我们所谓的战略思维、优先级排序这类工作，往往需要绘制大量二维矩阵，通过逻辑推演得出结论——而这正是AI的强项。不管你怎么评价AI，至少在情感层面，它们目前确实还不太擅长。

You know, a lot of our sort of what we call strategic thinking and prioritization and things like that is, we have to, like, draw a bunch of two by twos and and and really figure out how to logic our way through it. Like that is something that they are excellent at. Say what you want about AIs. They're not really quite, you know, good at the emotional stuff, like, yet.

我们在采访前制作了一个幻灯片。重点是列出仍然适用的技能、正在或已被颠覆的技能，以及产品经理应该掌握的新技能。不如我们就逐项讨论吧？好的，欢迎大家。

You and I made a slide before the interview. The point is to, list all the skills that are still relevant, skills that are being disrupted and maybe already disrupted, and new skills that PMs should build. Why don't we just talk about them one by one? Okay. Welcome, everyone.

今天的嘉宾是Linear产品负责人Nan。要知道Linear只有两名产品经理，却服务着超过15,000家企业，包括OpenAI、Ramp和Mercury等知名公司。今天我将邀请Nan和我进行一个有趣的练习，为AI时代仍然重要的产品经理技能排序。Nan还会演示他最爱的AI工作流和Linear的新AI智能体。欢迎你，Nan。

My guest today is Nan, head of product at Linear. You know, Linear only has two PMs, but it's scaled to over 15,000 companies, including OpenAI, Ramp, and Mercury and other great companies. And today, I'm gonna ask Nan to do a fun exercise with me to to rank what PM skills will still matter in the AI era. And and Nhan will also demo his favorite AI workflows and Linear's new AI agents. So what welcome, Nhan.

是的，很荣幸来到这里。

Yeah. Great to be here.

那么先从这个问题开始：为什么你们只有两名产品经理？你认为整个行业的产品经理岗位会大幅减少吗？

So let's start with this question. Why do you only have two PMs? And and do do you think we'll see, like, far fewer PM openings just in the in industry?

我认为这个岗位的性质会发生很大变化。我们PM人数少有多方面原因，部分算是运气——我们的工程师和设计师非常优秀，能承担传统上由PM负责的需求发现、商业案例推演等工作。但我们也是刻意保持缓慢招聘。高速发展的公司常会因事务繁杂而快速扩充PM团队。

I think the the nature of the job is gonna change quite a lot. Like we have very few PMs for a variety of reasons, some of which are kind of luck, right? Like we have very skilled engineers and designers who can do a lot of what's traditionally been kind of done by PMs in terms of discovery and thinking through business cases and stuff like that. But we also wanted to hire very slowly on purpose. I think a lot of times when you have a high growth company, people hire really, really quickly on the PM side because there's just too much stuff to keep track of.

这些新增岗位往往不是为了创造价值，而是处理机械性工作：维护太多产品需求文档、撰写过多更新说明——这类需求最终总是落在PM肩上，因为我们就像粘合剂。而我们不愿为这种需求专门招人。

It's not necessarily even to build. It's like specifically to like, just need more hands to maintain things like, oh, there's too many PRDs to maintain. There's too many updates to write. These very mechanical types of needs that always ends up falling into the laps of PMs because we're just the glue that holds things together. And we didn't want to hire people specifically because of that need.

因为早在三年前，我们就预见到这类工作将被AI取代。我们不想陷入这样的困境：招了一堆人做这些事，结果这些工作突然可以自动化了，又得为他们寻找新定位。我们更倾向招聘能长期深耕某个领域的人才。

Because that's something where even three years ago, we saw that type of work being largely overtaken by AI. And we didn't wanna be stuck in a position where like, hey, had a bunch of people to do this stuff, now this stuff is kind of free. And then we'd have to find new things from them too. Would rather hire people for things that we think that they would be able to kind of own and and kinda grow with for for a long period of time.

明白了。那我们实际点，把之前做的幻灯片调出来吧。

Got it. So let let's get, like, a very practical here. Like, let's let's pull up that slide that we made.

好的。当然。

Yeah. Sure.

所以我想我们在面试前做过一张幻灯片。重点是列出所有仍然相关的技能、正在被颠覆或可能已被颠覆的技能，以及产品经理应该培养的新技能。我们何不逐个讨论这些呢？对吧？那就从产品品味开始吧。

So I I think you and I made a slide before the interview. And the point is to list all the skills that are still relevant, skills that are being disrupted and maybe already disrupted, and new skills that PMs should build. And and why don't why don't we just talk about them one by one? Right? So let's start with product taste.

产品品味到底指什么？什么是产品品味？

What does product taste actually mean? What is product taste?

是啊，我知道这概念有点玄乎对吧？因为一旦开始描述它，你就会陷入各种细节，试图剖析它的微观层面。而它真正的含义在于，你是否能对特定产品或交互的好坏产生直觉感受。

Yeah. I know. It's weird, right? Because if you start describing it, then you're going to go into all these detail and try to figure out the minutiae of it. And then the real kind of meaning behind it is that like, do you, you know, can you like get a feeling for how like good or bad specific products or interactions just feel.

你从中获得的感受，是否能映射到足够大的受众群体——你要么能代表他们，要么至少在脑海中模拟：'好吧，这个人处在这种情况下会有什么感觉'。这很大程度上像是爬虫脑的本能反应，没有太多理性分析，就是试图理解不同交互方式和功能在你内心唤起的情感。

The feeling that you receive from it, does that map to a large enough audience where you can either be representative of them or at least run a simulation in your head about like, okay, here's how this person would feel in this situation. And a lot of it is very, it's very kind of like lizard brain. There's not a lot of reasoning around it. It's just like, you're kind of, trying to understand like what kinds of emotions that, different, types of interactions and features kind of conjure up within you.

这有点像流程吗？对我来说，流程就是你必须了解客户，看过足够多的产品后自然就能分辨优劣。

Is it kind of like process? And to me, process is like, you just have to understand the customers. You have to see enough products that you have a taste of what is good or bad, like that kind of stuff.

没错。很多时候你会先有感觉后知原因。看到某个设计时你会想'用户肯定会讨厌/喜欢这个'，有了这种直觉后，你需要逆向推导出：'该如何向别人解释他们产生强烈情绪反应的实际原因？'

Yeah. And a lot of times you'll have the feeling before you know why. You'll see a thing you're like, Oh, they're going to hate that or they're going to love that. You're going to have that feeling and then you have to kind of work backwards from that and figure out, Okay, what is the actual reasoning that I can explain to somebody about why they would have such a strong emotional reaction?

明白了。第二点是品牌建设。这是指个人品牌吗？你说的品牌具体指什么？

Okay, got it. And number two is branding. So is this like personal branding? What do you mean by branding?

我认为两者都是。产品品牌当然重要——比如命名就要在极简信息中传递全部内涵。人们会从最微小的信息点形成产品印象，你的工作就是利用这些接触点让他们做好接受产品核心信息的准备。但个人品牌同样关键。

I think it's both, right? It's certainly like product branding, you know, like just even things like choosing the name of the thing in a way where like it communicates all of the information in a really compact package. You know, it's just like people will have an impression of your product from the very, very smallest piece of information. So part of your job is to figure out how do you leverage those kinds of small touch points in order to get them to have to be primed, right, to really receive what your product's message is or what you're trying to do. But also personal brand.

现在很多公司都是靠创始人的网红属性带动产品市场匹配。看看Cluly这些AI领域的企业，在产品问世前就大量投入商业或个人品牌建设。因为低垂果实已被摘完，你需要逆向操作——在提供实际价值前先建立情感共鸣。

We see all these companies now where somebody's almost influencer profile, it leads all of the actual product market fit. You look at like, Cluly, you look at all these kinds of especially on the AI space. There's so much brand building, whether it's in a commercial or personal context before you even have a product. And I think that a lot of that is important because the really low hanging stuff has been picked. So you want to almost work backwards again, build the emotional resonance before you even have a value offer.

是啊，

Yeah,

这很有趣。你提到Clueq，因为我不确定他们是否真有可行的产品，但他们的品牌和市场做得非常出色。

it's interesting. You mentioned Clueq because I'm not sure they actually have a viable product, but like their their branding and their market is great.

对，我是说，我认识的人里没一个用过那产品，但好像我认识的每个人都听说过它。

Yeah, I don't I mean, I don't know a single person that's used the product, but like I, but everyone I know knows about the product.

嗯。好的。第三点是所有权和风险偏好。所以这是否类似于拥有自主权还是？

Yeah. Okay. And then the third one is ownership and risk appetite. So is it like just having agency or?

对，我觉得这两者其实是相辅相成的。因为人们常谈论所有权，所有权意味着什么？意味着你要对结果负责。而对结果负责只是等式的一半，另一半是你会尽力争取最好的结果。

Yeah, I think, you know, these two things I think are really paired. Because, you know, people people talk about ownership, What does ownership mean? Oh, that means you're responsible for the outcomes. And being responsible for the outcomes is only half of the equation. The other half is you're going to try to get the best outcome possible.

这其中涉及一定风险，因为不承担风险就无法获得最佳结果，对吧？如果选择最安全的结果，你可能不会彻底失败，但也意味着你的上升空间非常有限。所以你需要有适当的风险偏好，同时承担结果，这意味着有时你会遭遇失败。你得跌倒后想办法重新站起来，这确实很烦人。

And there's some risk involved in that because you can't get the best outcome possible without incurring Right? If you go for the safest outcome, you might not completely fall flat, but it also means that your your upside is really capped. So you kinda have to have the right amount of risk appetite and also own the own the result, which means that sometimes you're gonna have failures. Right? You're gonna have to, you know, get knocked down and then try to maneuver your way back out, which is, like, super annoying.

对吧？尤其是在企业环境中，你还得考虑政治资本之类的东西。这就是游戏规则，你必须真正准备好并愿意参与其中。

Right? Especially in a in a corporate context where, you know, you have political capital or whatever it happens to be. Right? Like, that's, you know, that's that's the game, and you have to really be ready and willing to play that game.

是啊，你得管理预期。感觉相关的一个能力或许不算能力——就是你得真心在乎产品，必须不惜一切代价让它成功。

Yeah, you gotta manage expectations. Feel like one skill that's relevant is maybe not a skill. It's like you just have to give a shit about the product, You have to do whatever it takes to make it

没错。你必须在意，并且准备好让人们把最终结果的责任都推给你。比如我们开发一个功能，在发布前，我知道所有功劳和辛苦都在工程团队和设计团队身上。但如果大家都不喜欢，挨骂的肯定是我。

Yeah. You have to care and you have to be ready to just like be for people to look to you for whatever the result ends up being. Right? Like, when we build a feature and before we launch, it's like, I know that all of the credit and the hard work happened on the engineering team and the design team who built all this stuff. But if everyone hates it, I'm the one they're gonna be yelling at.

所以我必须接受这种情况会发生。

So I just have to understand that that's gonna happen.

第四点其实挺有意思的，因为在大公司里，利益相关者管理基本就是产品经理的日常工作，对吧？差不多占了80%的时间。所以你觉得这依然很重要是吗？毕竟大家还是需要相互协作的。

And and how about, like, number four is pretty interesting because, like, stakeholder management is kind of like at at big companies, it's kinda like what a PM does all the time. Right? Like, it's almost like 80%. But so so you think that's still very important, right? Because people have to work with each other still?

没错。必须相互配合，特别是在大机构里，你得争取足够的预算和空间，才能让自己的想法真正落地。要管理资源等等，这又回到风险偏好的问题——你得清楚自己有多少回旋余地。

Yeah. You have to work with each other, you know, and then, like, especially in larger organizations, there's like, you just have to get, you know, get enough budget and get enough runway to, like, let your ideas, you know, actually happen. Right? You have to, like, manage resources and all that stuff. And it kind of goes back to sort of risk appetite, which also means that you have to know how much sort of, like, leeway you have.

比如我能惹毛这些人到什么程度？但只要结果好，最后大家都会笑着碰杯。你得会做这种权衡然后放手去做。

Like, how much can I piss these people off? But if the result is good, then everyone's going to smile and laugh and have a beer at the end. You kind of have to be able to make that calculation and go for it.

我觉得这对小公司也很重要。YVAN Lin你需要揣摩Curry的想法，让团队协作更顺畅。

And I think that's actually kind of important even as a smaller company, right? YVAN Lin, you need to know, you kinda wanna learn what how Curry thinks and, like, you you wanna, you know, make it easy to collaborate with other folks.

当然。我们不可能100%达成共识，有时候就得冒点险，需要你跟着我走一段。如果创始人Kari、Jorrey或Twams特别坚持某件事，那就按他们的来。

Oh, yeah. For sure. Right? We're not gonna agree on on 100% of things, and there's gonna be some stuff where, look, we're we're gonna we're gonna take a bit of a risk here, and you're gonna have to come along with me a little bit. And and sometimes, you know, if if one of the founders, Kari or Jorrey or Twams have a, you know, have a very strong point of view, it's like, okay.

即使我觉得自己的方案成功概率更高，但你的想法也不是完全没戏。我们可以调换执行顺序试试。

I let's let's try your way. Right? Like, you you know, even if I think that my way has a higher chance of success, it's not your your idea is gonna be zero. So, like, we can just do things in the opposite order as well.

所以第四点其实直接关联第五点的情商对吧？包括对客户和内部团队的情商，你是这个意思吗？

So I guess, like, four directly ties to five, which is e, EQ or emotional intelligence. Like, I I guess it's both with customers and internally. Is is that what you're thinking?

对，这些本质上都关乎情商。品牌建设就是在获得实体价值前先获取情感认同，产品调性同理，利益相关者管理更是如此。

Yeah. I mean, like, I think all of these things really tied to emotional intelligence, right? Like the branding stuff, it's like, you're basically getting all of the emotional kind of framing and benefit before any of the actual kind of physical value, right? And product taste is the same thing and stakeholder management, a big part of that.

好吧...这挺难的。我真不知道该怎么提升情商，太难了。

Okay. I don't know. Kind of hard. I don't know how to build emotional intelligence. It's hard to build, man.

我唯一验证有效的方法是设身处地思考：如果我是B2B客户，面对他们的岗位压力和各路利益相关者，什么会驱动我？我的动机和顾虑是什么？通过这种分析慢慢培养直觉。

Is. I think the only successful way that I found to practice it is to really make a concerted effort to say, if I was doing this job that my user is doing I'm in B2B, So it's always like, hey, if I was in their role and their position with their pressures and stakeholders and stuff like that, what would be motivating me? What are my motivations? What are my fears? Do that kind of analysis and then you can start really kind of building that and start intuiting it.

好的。这有点像共情，把自己代入客户或利益相关者的立场，就是那种感觉。

Okay. So it's kind of like empathy and putting yourself into customer order, stakeholder shoes, like that kind thing.

对，对。你可以用技术化的方式处理。就像经济学家那样直接说：这些是我面前的实际激励因素，如果我根据这些行动，我会怎么做？

Yeah. Yeah. And you can be kind of technical about it. You can literally just like do the economist thing and say like, here's the actual incentives that are in front of me. And if I was acting according to these, what would I do?

明白了。有道理。那我们聊聊第二类话题吧，这个可能更有趣些。

Got it. Okay. That makes sense. Okay. Let's talk about the second category, which is probably more fun to talk about.

所以第一个观点算是热门话题对吧？关于战略和优先级。你是说这个领域正在被颠覆？

So the first one's kind of like a hot take, right? Like the strategy and prioritization. You're saying it's being disrupted?

哦，如果你问别人，这就是个老套说法——AI会颠覆所有人的工作，除了我的。对吧？这是每个人的观点。或者稍微好些但仍有动机的说法是：它会颠覆我工作中不重要的部分。

Oh, I think if you ask people, right, like this is the trope, right? It's like, AI is going to disrupt everyone's job except for mine. Right? Like that's everyone's kind of point of view. Or the sort of you know, slightly better, but still kind of motivated, sort of thing they'll say is like, it'll disrupt all the parts of my job that are not important.

对吧？工作中更重要的部分，那些是我无法触及的。但这其实没有物理定律支撑。我们说会持续相关的那些能力都是情感导向的。不管你怎么看AI，它们现在确实不太擅长情感方面。

Right? That the more important parts of the job, like, those are the parts that I can't touch. And, like, there's no there's no physics about why that's true. The stuff that we said that are gonna be kinda continue to be relevant are very like emotion oriented. Like, say what you want about AIs, they're not really quite good at the emotional stuff yet.

但在逻辑推理、系统化思考、整合证据和广泛考量方面，它们非常出色。而我们称之为战略思考和优先级排序的那些事，比如画各种二维矩阵来逻辑推演——这正是AI的强项。这很棒，因为责任还在我们身上。目前它是很好的辅助工具，但不会成为人与人之间的区分标准。

But at reasoning and at, like, thinking through things in a, you know, in a kind of, like, organized fashion and pulling in all the evidence and considering a really broad scope, like, they're very good at that. And, you know, a lot of sort of what we call strategic thinking and prioritization and things like that, it's like we have to draw a bunch of two by twos and really figure out how to logic our way through it. That is something that they are excellent at, which is great. Because we're still responsible for that. Right now, it's a tool that really helps us, but it's not gonna be a thing that necessarily differentiates one person from another.

是啊，我现在制定策略就是先让AI做个深度研究项目，上传一堆文档。然后问你：这个合理吗？那个合理吗？为什么不行？

Yeah, I always make my strategy now by just making an AI project and then asking to do deep research and then uploading a bunch of documents. And I talked to you about the strategy. I'm like, Hey, does this make sense? Does that make sense? And then why not?

没错没错。它是个绝佳的思维伙伴。除非你用的模型有问题，否则它通常都会严谨诚实。它能快速完成你希望随时做到的思维过程。

Yeah, yeah, yeah. It's a great thought partner there. It'll be rigorous. It'll be honest, unless there's a weird model that you're working with right now. But most of the time it'll do the mental process that you wish you could do really fast and all the time.

所以如果人人都能使用它，这就不会成为某个产品经理比别人优秀的关键因素了。

So it's just like, if everyone has access to that, then that's not gonna be the thing that makes 1PM a lot better than another.

是的。那种把自己关在房间里苦思冥想的套路已经行不通了。完全没用。好，下一个是数据分析和综合。

Yeah. The trope that you go off to a room and think hard by yourself, doesn't work anymore. It doesn't work. Okay. So the next one is data analysis and synthesis.

这就是让AI编写SQL查询并找出洞察，对吧？

So this is like getting AI to write SQL queries and find the insights, right?

对，对。我觉得这类事情就像有很多模板化的东西，无论你在哪个行业，相同的逻辑都适用于相同的问题。任何具有这种形态的事物，意味着外面有海量的训练数据。意味着这些操作在多个地方都是高度可复制的。

Yeah. Yeah. And I think with this kind of stuff, it's like there's a lot of boilerplate and the same logic no matter what business you're at applied to the same problems. Anything that has that shape means that there's a ton of training data out there. It means that the actions are pretty reproducible in multiple places.

所以这些东西注定会被颠覆。不同事物受影响的速度不会相同。我们常说AI的能力边界参差不齐。它不会同时擅长所有事情，但只要是逻辑性强、循序渐进、以推理为导向的领域，它迟早都会触及。

So these things are invariably going to become disrupted. The speed at which some of these things are affected is not going to be equal. We talk about AI having a ragged edge. It's not going be equally good at everything simultaneously, but everything that is very logical and step by step and reasoning oriented, it's going to end up touching sooner or later.

没错，它现在还不太擅长简单算术，但写SQL查询的话，只要提供表名之类的信息，它就很在行。

Yes, it's not very good at simple math yet, but for writing SQL queries, if you get the table names and stuff, it's great.

是啊，挺奇怪的。然后还有一点就是...我不知道该怎么说，就像以前必须雇个分析师才行。

Yeah. It's kind of weird. Yeah. And that's what it is And then another one is just like, I don't know. It's just like, I have to have an analyst.

否则根本不可能得到任何结果？而现在这种最基础的差异已经消失了。所以现在你得往更高层次走，比如数据解读这类工作。当然AI目前在这方面还不算顶尖，但它显然正在这条路上前进。

Otherwise, there's no way I can I can even get any kind of result? And like that difference, that sort of like very basic differences is evaporated. Right? So now it's like, okay, you have to kind of move higher level with like interpretation and those kinds of things. And, you know, and the AI is not necessarily the best at that yet, but like, it's, it's clearly on that, on that path.

太棒了老兄，因为我真的不想学SQL，学习过程一点都不有趣。

That's great, man. Cause I, I don't, I don't want to learn SQL. Like it's not a fun thing to learn.

我懂。我常开玩笑说，我从来没学过正则表达式，因为总是想着'明年再学吧'之类的。结果现在发展到我这辈子都不用再学正则了。

Yeah. I know. I, I, my joke is like, I never learned regular expressions because like, I just I just never got around to it. It was always gonna be all I'll I'll learn it next year or something like that. And then, now we're at a point where I never need to learn regress ever again.

没错。

Yeah.

对，好的。下一个话题是市场调研。我经常收到伯克利学生的邮件，他们说什么伯克利咨询俱乐部之类的，说要帮我做行业调研三个月。兄弟，我直接用深度调研工具不香吗？

Yeah. Okay. And the next one is market research. So I so feel like, you know, I get these emails from Berkeley kids. Like, they're like there's like a Berkeley okay.

伯克利的朋友们如果听到这段，别误会。但那些大学咨询俱乐部总说'嘿彼得，我能帮你做这个行业的市场调研'。拜托，这种活我用深度调研工具分分钟搞定。

If you're listening to this from Berkeley, like, don't take it the wrong way. But there's like this Berkeley Consulting Club or like some college consulting club. And they're like, Hey, hey, Peter. I can do market research for you for this industry, you know, like, for like three months. Like, dude, like, let me just do let me just use deep research to do it.

我哪还需要你们？所以我觉得这类服务基本没市场了。

Would I need you? So yeah, so I feel like a lot of this is gone.

你知道，调研这事挺有意思的。你刚才说的那些案例，虽然号称深度调研，其实很浅层对吧？就像初次接触某个领域时，只能摘些低垂的果实那种。

You know, it's like, yeah. Yeah. I mean, research is interesting, right? The cases that you just talked about where you're kind of doing, I know it's a deep research, it's kind of shallow, right? It's kind of like a first pass type of research where you're going in kind of naive and you're learning all the low hanging stuff.

这种初级调研已经完全被工具取代了。那些咨询俱乐部提供的通常就是这种。现在真正有价值的调研，是分析师长期跟踪某些领域，甚至和高管们吃饭聊天，了解他们个人情绪那种。

That part has been completely overtaken. And that's usually what these kinds of, like, consulting clubs and stuff like that, like, are are offering. The research right now, right, that's that's kinda gonna you know, that exists. It's it's like if you if you, you know, are an analyst and you have a couple of things that you follow over a very long period of time and you, like, you know, are having, like, dinners with, like, the executives and and really, you know, understanding their, you know, their their their personal emotions and all that kind of stuff. Okay.

这种高端定制化调研依然有市场。但这是非常特殊、需要高度接触的研究类型。

That that's still alive. Right? Like, that kind of research, great. But that's a very, very special, like kind of like high touch type of research.

对，就像一手和二手调研的区别。真正获取新信息那种。

Yeah, that's like, just like there's a primary and secondary research, right? Like you're actually getting like new information.

没错，没错。

Yeah. Yeah.

然后项目管理这块，我猜Lin也在尝试服务这个领域？你说的项目管理具体指什么？Jira工单那些吗？

And then project management is, I I guess that's what Lin here is also trying to serve, right? What do you mean by project management? You mean like Jira tickets and stuff? What are you thinking? Yeah.

项目管理范围很广，目前还没有被全面攻克。核心是细节把控——你想啊，要是问'你合作过最棒的项目经理有什么优点'，十有八九会有人说'他们特别注重细节'。

Yeah, I mean, project management, it's a really wide space, right? And it's not being sort of evenly kind of attacked, but a lot of it is keeping track of details, right? If you think about like, hey, name the best PM you ever worked with, right? Sorry, best project manager you ever worked with and tell me some good qualities about them. And, like, invariably, someone will say, oh, they're very detail oriented.

他们注意到每一个小细节。他们阅读每一份文件。他们就像把所有细节都装进脑子里，随时掌握最新动态。这听起来不就是电脑吗？就像你说的那样，甚至可以说他们像台机器，对吧？

They notice every little thing. They read every single document. They, like, have all the details in their head all loaded in, you know, and they always know the the current state of affairs. It's like, that sounds like a computer to me, right? Like that, like what you're describing is, is, you know, is, you know, like, you know, you might even say like, oh yeah, they were like a machine, right?

这就像你会用的比喻。要知道电脑以前不擅长这个，因为它们无法解读很多这类信息。比如有人在文件里写'十万火急，大家快抓狂了'——

Like that's, that's like a metaphor you would use. And, you know, computers used to be bad at this because they they couldn't interpret a lot of these, things. Right? If someone wrote, like, in a document, this is an emergency. Everyone, you know, pull your hair out.

电脑根本理解不了严重性。但现在它某种程度上可以了。所以我们现在这点微弱的智能，真正解锁的其实是海量的勤勉力。因为电脑超级超级勤勉。

Like, the computer wouldn't understand the the severity of that. Whereas now, like, it kinda can, you know? So, like, so so the all of the, you know, the the the little bit of intelligence that we have right now, what it's really unlocked, right, is a a whole bunch of, of of diligence. Right? Because the computers are super duper diligent.

对吧？它们极度注重细节，只要插着电源就永远在线。只不过以前完全没有智能。现在有了点智能，就释放了与生俱来的勤勉特质。

Right? They're super detail oriented. They're always online as long as you plug them into the wall. So but they just weren't intelligent at all before. Now that they've they've had a little bit of intelligence, it's unlocked all of that diligence that they they already come with.

但我感觉现在有种对产品细节管理的执念，比如揪出产品里的小毛病。AI当然也能做这个，但关键是你得真正在乎这些。所以这可能属于产品品味的范畴——

But I I feel like there is, like, obsession about detailed product management, like, you know, pointing out little details, little inaccuracies in the product. I guess you can use AI for that too, but you also just need to care enough about this stuff. So maybe that's a product taste or

对对，这就是产品品味问题。AI大概能说'这里视觉回退了'之类，但没法真正判断'这个设计不协调'、'页面垂直节奏有问题'或者'这个按钮很丑'。这些还是主观判断。

Yeah, yeah. That's a product taste thing, right? AI could probably do things like, say like, Oh, there was a visual regression here or whatever it is, but they can't say really, Hey, I think this design is incongruent, The vertical rhythm of this page doesn't make sense, or This button's ugly, or whatever it is. Those are still up to opinion.

明白了。对了你之前发推说'我们终于不再讨论赋能型产品经理了'——说真的哥们，我从来不懂那套理论，完全搞不清什么意思。

Okay. That's great. You tweeted like, this is kind of random, but you tweeted that we finally stopped talking about empowered PMs. Yeah, dude, I was never into that whole thing, man. I don't even know what that means.

是啊，我们最终是要交付产品的对吧？要让客户满意。只要结果好，具体怎么实现的其实没那么重要。

Yeah. It's like, we're trying to ship product at the end, right? Like, we're trying to make an outcome happen, you know, make customers satisfied and everything. It's just like, if that's the result, then how you get there really doesn't really have that big of an effect.

确实不是非黑即白的事。对了，我把研究总结和文档整理分开列了，因为你提到这部分已经被颠覆了？啊对，总结归纳确实说得通。

Yeah, it's not like black, black and white. So, okay. And then I separated summarizing research and documentation because you made a point that this stuff is already disrupted, right? Oh yeah. I guess summarizing, it makes sense.

你只管总结就行。

You just get to summarize.

对，对。我觉得总结和文档记录几乎是同一回事。你有一些输入来源。做研究时，就像是，好吧，这些是我在网上或某个档案中找到的所有资料。

Yeah. Yeah. I think summarizing and documentation are almost the same thing. You have some kind of input source. With research, it's like, Okay, here's all the sources that I found on the internet or in some archive.

文档记录的话，就像你有一个代码库，有一大堆人们的意见、做出的决策之类的东西。你只是把所有这些东西整理到一个地方，形成一份研究报告或文档。然后接下来困难的部分是，当某些东西发生变化时，你必须更新它。比如有人改变了我们前进的方向，你就得更新文档；或者有了新进展，研究过时了，你就得用新发现来修订它。这部分其实做起来也不难。

With documentation, it's like you have a code base, you have a whole bunch of opinions that people have and decisions that were made and things like that. You're just taking all that and you're collating it into one spot, into one research report, into one document. Then the next part, which is the hard part, something changed and now you have to update it. So someone changed their mind about the direction we're going so you gotta update documentation or there's a new development so the research is out of days, so you have to you know, amend it with like the new the new, the, you know, the new findings. And that part is once again not that hard to do.

你需要一点智慧来做这件事，但更需要大量的勤奋。对吧？你需要做的就是盯着所有来源，一旦其中任何一个发生变化，就迅速采取行动。明白吗？

You need a little bit of intelligence to do it, but you need a whole lot of diligence. Right? You need a whole lot of, I'm just gonna stare at all of the sources. And the moment one of them changes, I'm gonna act really fast. Right?

而且，就像我们之前讨论的，为什么Linear不雇一堆产品经理？很多时候人们雇佣产品经理就是想要这种动态——希望有很多双眼睛盯着这些需要思考的东西。一旦有变化，他们就会说，哦，得有人非常勤快，立即更新所有文档。而我们觉得，我不会因为这个原因雇人。

And, like and, you know, when we talked about, like, why doesn't linear hire a bunch of PMs? It's like, like, a lot of times people hire PMs because they want that dynamic. They want a bunch of eyeballs on all this stuff, that you have to think about. And the moments something changes, they're like, oh, someone's gotta be really diligent and just update all of the documentation immediately. We're like, look, I'm not gonna hire someone because of that.

因为我认为计算机会100%地替你完成这件事，比任何人类都做得好。在过渡期，我可以暂时没有这种便利，因为我实际上想迫使我们尽早让计算机来做这件事。

Because I think computers are gonna do that just for you 100% of the time way better than any human could. And I'm okay in the interim just not having that benefit because I wanna actually force us to like, you know, at the earliest possible moment, like to get our computers to do it.

有道理，有道理。确实，即使是现在，我收到一堆利益相关者的反馈时，就直接跟AI说，嘿，这是我的文档，你能根据利益相关者的反馈更新它吗？然后也许把你修改的部分加粗，这样我就能看到你改了哪里，做个简单的检查。

That makes sense. That makes sense. Yeah. I mean, even right now, I get a bunch of stakeholder feedback, you know, like I just talk to AI, I'm like, hey, you know, here's my document, can you update this document based on stakeholder feedback? And then maybe bold your edits so that I can see what you changed so that I can like just do a sanity check.

是啊，你知道，我不可能全靠自己完成这些。

Yeah. You know, like I'm not gonna do it all by myself.

对，没错，正是这样。

Yeah. Right, right, exactly.

好的。然后是新技能培养，我觉得这些都挺不言自明的。我把上下文工程放在这里，因为它不仅仅是提示工程，你得考虑加载哪些内容到AI的上下文中，以及如何有效管理，对吧？是的，以免不堪重负。

Okay. And then new skills to build, I think these are pretty self explanatory. I guess I put context engineering here because it is more than just prompt engineering, that you gotta think about what to load into the AI's context and how to manage it well, right? Yeah. To not get overwhelmed.

然后是评估，肯定有相关的内容，确保AI的回应总体上合理或准确。但像AI工作流设计和AI代理管理，这正是Linear也在直接研究的内容，对吧？

And then Evals, I'm sure there's episodes on that, making sure that AI's responses are overall, make sense or accurate. But like AI workflow design and AI agent management, this is like directly what Linear is also working on, right?

没错，没错。我是说，设计和上下文工程是紧密相关的，对吧？上下文工程就像是理解：好吧，这个模型需要哪些信息才能在我雇佣它完成的工作中表现更好？对吧。

Yeah, yeah. I mean, design and context engineering are very closely related. Right? Like context engineering is sort of understanding like, okay, like what information would this model need to have better performance for the job I'm hiring it for? Right.

然后，工作流设计就是：现在如何确保它能在正确的时间接收正确类型的上下文？所以这两者实际上是相辅相成的。工作流设计的一部分包括：我希望代理如何表现？如果它没有正确的上下文，它怎么知道自己需要去收集更多？然后我该给它什么工具来帮助它获取所需的上下文？

And then, workflow design is okay, now how do I make sure that, it's always able to receive the right type of context at the right time? So those two things really of build on each other. And part of workflow design includes like, well, how do I want the agent to behave? If it doesn't have the right context, how does it know that it needs to go gather more? Then what tools do I give it to go actually help itself gather the context that it needs?

所以我认为上下文工程真正关注的是它需要什么样的上下文？而工作流设计则是：好的，现在它通过什么动态过程持续获取正确的上下文？

So I think context engineering is really about what kind of context does it need? And then the workflow design is like, Okay, now what is the dynamic process by which it keeps getting the right context?

明白了，那么可以使用哪些工具或采取哪些步骤来获取正确的上下文呢？

Okay, so what kind of tools can you use or what kind of steps should it take to get the right context? Yeah,

正是如此。其中一些内容，我们实际上仍需编写具体的提示，对吧？比如：嘿，如果你处于这种情况，那就先到这里看看，诸如此类的。模型仍需要一些指令，但随着时间推移，这方面正变得越来越好。

exactly. And some of that stuff, we still actually have make specific prompts, right? It's like, hey, if you're in this situation, then go here, look here first and all that kind of stuff. There's a bit of instruction that the model still need, but it's getting better and better as time progresses.

好的。说到代理管理，我们都有这样的幻想——或许正在成为现实——比如拥有五个AI代理为我们工作，我们只需管理它们。这是现实还是还有点遥远？

Okay. And agent management is kind of like, I mean, we all have this fantasy or maybe it's becoming reality of like, you know, we have like five AI agents that just doesn't work for us and we just kind of manage them. Yeah. Yeah. Is that reality or is that still kind of far off a little bit?

你知道，我认为这些技术的发展很不均衡。在某些角落，它已经是现实了，对吧？有些情况下，你真的只需说几句话就能派遣一个代理，它会为你处理一切，然后你就完事了。但有时，实际需要说的话可能比你预期的多得多。

It's, you know, I think these things are so unevenly distributed. It's it's reality in some corners sometimes. Right? Where there there are some cases where you can literally just, you know, say a couple of words and dispatch an agent, and it'll just candle it for you, and then and then you're done. I think sometimes it's it's the the number of words you need to say are actually a lot greater than you expect.

而且，很容易指定不足。我认为这就是代理管理的一个方面：需要知道详细到什么程度？对吧？比如，你真的需要微观管理这些代理吗？还是说只要给出大致方向，它就能做出让你满意的表现？

And and, you know, it's easy to underspecify. And I I think that that's where that's one aspect of agent management is to know, like, detailed do have to be? Right? Like, do you need to really micromanage these agents? Are you are you perfectly fine just kind of, like, giving it a general direction, and it's gonna do something that's that's good enough for you to accept the the performance?

一旦进入那种模式，问题就变成：你同时能驾驭多少个这样的代理？考虑到当前的技术状态，需要理解异步操作的程度。但我认为代理管理很大程度上是关于你的指令需要多么详细或简略。

And then once you're in that mode, then it also becomes like, Okay, how many of these things do you juggle at the same time? And understanding something about how asynchronous you have to be given the current state of technology and stuff. But I think a lot of agent management is really about how under or over specified do you actually need to make your instructions.

是的，这个观点很好。所以我的理解是，产品需求文档（PRD）并没有消亡，只是由AI来撰写，然后它有点与提示融合了。比如当我想要用Cloud Code构建某物时，我会要求它写一份关于技术栈和所有里程碑的详细PRD。

Yeah, that's a good point. So my take is that the PRD isn't dead. You're just getting AI to write it, and then it's kind of merging a little bit with the prompt. Why I want a Cloud Code to build something, I asked to make a pretty detailed PRD on the tech stack and everything, right? The milestones.

我只是不自己动手写。

I'm just not writing it by myself.

是的。我认为这非常正确。我常告诉人们的是，你写的每样东西，主要受众其实不再是人类。因为即使是会读你东西的人，他们也是通过AI来阅读的，对吧？

Yeah. I think that that's really true. One of the things I try to tell people is that everything you write, the main audience is not really people anymore. Because even the people that are gonna read your stuff, they're reading it using an AI. Right?

就像我今天在Linear写一个问题时，第一个读到这个问题的就是我们的AI。它会利用这个问题去查找各种相关问题，并试图确定该由谁来处理等等。所以我心里清楚，写这个问题时，我需要给AI足够的信息来完成它的工作——这恰好对人也很帮助。这样别人读到时会觉得‘好，不错’。

So like if I write an issue, you know, like in linear today, if I write an issue, the first thing that reads that issue is our AI. Right? And it uses that issue and it goes in and it finds all sorts of like related issues and tries to figure out who should work on it and stuff like that. Like, so I need to know, I sort of know in the back of my head that like, when I write this issue, I need to give the AI enough information to do its job, which also happens to be really helpful for people. So when someone else reads it, go like, Okay, cool.

它包含了所有正确信息，有所有重现步骤之类的。但以前这是个模糊的概念——如果我不写够细节，别人会抱怨。但现在如果细节不足，所有自动化流程都会失效。

It has all the right information. It has all the repro steps or whatever. But before, it was kind of a squishy thing. You're like, if I don't put enough detail, then someone's going to yell at me and be upset. But now if I don't put enough detail, then all the automations won't work.

所以你看，动机其实完全不同了。

So, you know, like, the motivation is actually quite different.

对，这很合理。所以我总是把文档控制在半页关键点，再加一堆附录。

Yeah. That that makes sense. Yeah. Yeah. That that that that's why I keep all my documents to, like, a half a pager of key points and then maybe a bunch of appendices.

因为如果主文档超过半页，别人就会用AI来总结它。是啊，就是这样。

Cause, you know, if my main doc is more than half a page, someone will just use AI to summarize it. Yeah. You know, it's like, yeah. Okay. Great.

所以这个很实用，希望人们能用这个表格找出需要提升的技能。好了伙计，我们多聊聊Linear吧——这个播客叫《匠心背后》，所以我特别喜欢和你讨论以及与Linear合作。

So this is this is pretty tactical. So hopefully people can use this table to figure out what skills they need to work on. All right, man. So let's talk about linear a little more. So this podcast is actually called Behind the Craft, so this is why I love talking to you and working with linear.

但我觉得人们对匠心有种误解，以为就像苹果那样花一年时间打磨出超级精致的产品。你对匠心与速度之间的平衡怎么看？

But I think people have a misconception about craft. People think that craft is kind of like what Apple does, where, like, you just kind of keep working on something for, a year, and then you have a big bang. Like, oh, this is, super polished, super beautiful stuff. What what what is your point of view on that, like, between craft and speed?

苹果是硬件公司，他们的工作流程源于硬件设计和开发领域。改造硬件生产线极其昂贵，所以在那领域追求高质量就必须确保量产前一切完美。

Yeah, I think it's software Apple's a hardware company, right? So all of their working processes and things like that originate in sort of like a hardware design and development space. Like retooling hardware production lines is enormously expensive. So it just makes a lot of sense. If you want high quality in that universe, you have to really, really make sure that things are super polished before you go to mass production.

软件方面，你可以几乎零成本地部署、撤销部署以及持续集成这类操作。所以，尽管最终目标相同，但实现高质量、高工艺最终状态的具体思路却大相径庭。对我们而言，重点不在于精雕细琢至完美，而是尽可能多地获取反馈循环——因为每次迭代都会让产品变得更好，它能揭示产品与真实世界的契合度，这远比单纯苦思冥想来得有效。

Software, you can deploy and undeploy and continuously integrate all this kind of stuff at basically no cost. So, you know, the the idea for, you know, even though the end goals are the same, right, the ideas around how to, like, get to that high craft, high high quality, you know, final state are are quite different. And then for us, it's it's not about, whittling on something until it's perfect. It's about just getting as many feedback loops as possible because every iteration that you go through makes the thing better. And it tells you about how your product fits into the world way more than you could by just thinking very hard about it.

因此我们认为，实现真正高质量成果的途径就是尽可能快地做出可用版本。经验法则是：在10%的时间预算内，你就应该完成一个能解决问题的雏形。这样你才能开始注入审美判断，或者感知它是否令人满意。

So for us, the way to get to really high quality outcome is to get to a working version as absolutely fast as possible. So, like, the rule of thumb is, like, you know, about 10% of the amount of, like, time budget that you have. Like, by the time 10% has elapsed, you should have something that gets the job done Mhmm. And that you can start, applying taste to. Or you can start feeling like, does this feel good?

怎样才能优化我的工作流程体验？它接近理想状态了吗？是否存在某些瞬间体验极佳但整体不均衡？还是始终感觉糟糕？这些都需要你做出判断，但只有将产品置于真实场景中使用后才能获得答案。

What would make it feel good my workflow? Is it close? Are there moments where it feels great and it's just uneven? Is it just consistently feels bad? There's all of those of judgment calls that you want to be able to make, but you can't do that until you're able to use the product in a real situation.

明白了。那你们如何收集反馈呢？是通过测试用户群组，还是像Antonio这样既有内部人员又有真实客户来获取意见？

Yeah. And then how do you get feedback? You have beta groups and you have people, both both Antonio, guess, and with real customers, to get feedback?

没错。我们采用分级发布流程：受众范围会随时间逐步扩大。最初只有开发团队使用——毕竟我们自己也用这些产品。接着扩展到Linear全体员工，再逐步开放给精选测试组、公开测试组，最后进入预发布和生产环境。

Yeah. Yeah. We we have a we know what we call like a like a doll release process where we have, like, sort of nested audiences that get bigger and bigger as as time progresses. So the first audience is just the developers, right, who are working on it. And we use our own products so they're able to kind of play with it and things.

就像昨天的重要发布，我问团队'正式发布前还需要完善什么？'工程师们表示毫不担心，因为这个版本已经通过了所有测试层级的考验。

And then the next audience is just people who work at linear, so the internal audience. And then from there, it'll escalate to a select beta, a public beta, and then pre release and production. Right? So, like, there's, like, a whole bunch of different shells that this thing goes through Yeah. As as it reaches wider and wider audience.

我们昨天刚完成重大发布时，我问'正式发布前还需要注意什么吗？'工程师们回答'完全不用担心，这个版本已经通过了所有测试阶段的考验'。

Like, we we had a big release yesterday, I was like, okay. Anything, you know, we wanna take care of before we go to GA? And then the engineers were like, I don't know. It's been in, like, all these different beta levels. Like I have absolutely no fear of anything falling over or anything like that because it's been battle tested already.

这做法太理所当然了！真不明白为什么更多公司不采用渐进式发布策略来优化产品。何必冒险搞大爆炸式发布呢？太奇怪了。

Yeah, this is such a no brainer thing, man. I don't know why more companies don't do this. Roll like out to more people progressively, right, and make the product better. It's like, why do you want to risk it with a big launch? It's weird.

确实。具体原因我也不清楚，但这个方法在每家公司实施时都获得了认可。我觉得很多事都取决于执行路径——只要有人认定这事值得做就行。

Yeah, yeah. Don't know yet. I think it's pretty obvious. Everywhere I've gone, I've installed this kind of process, I don't think anyone's really felt bad about it. So I think there's so much of this stuff is path dependent.

关键就在于有人要下定决心推动这件事。

Someone just has to decide that this is something worth doing.

但在最初版本的10%阶段，让我再追问一下。如果我想把产品推向开罗或某些要求极高标准的用户时，如何确保它看起来不至于太糟糕？在这10%里你们重点打磨什么？是核心体验还是其他方面？

But how do you in that 10% of the initial version, let me push a little bit. How do I if I want to ship something to Cairo or somebody, he has like very high standards, right? So how do I make it actually not look like crap? Like, what do you focus on the 10%? Is like the core experience or?

对，就是核心体验。你可能不需要覆盖所有边缘情况，关键是确保黄金路径能按预期运行。

Yeah, the core experience, right? Like you might not cover all the edge cases. You kind of just want to make sure that the golden path does the thing you expect it to do.

明白了。

Got it.

如果有人使用频率高到遇到边缘情况，那反而是好事。这说明他们已经远远超出了黄金路径的使用范围。

And then if someone is using it enough to run into edges, then that's good. That's a good thing, right? Oh, you ran into an edge case? Great. That means you were using it way past the golden path.

确实。尤其是AI产品，边缘情况实在太多了。

Yeah, yeah. Got it. And then, yeah, that is especially for AI products because there's so many edge cases.

没错没错。

Yeah, yeah.

好。关于AI产品还有个问题——我们需要撰写特定邮件来确保产品在某些属性上得分高，这其实也是速度与工艺之间的平衡。

Okay. And I think another thing about Let's talk about AI products real quick. So I feel like with AI products, you gotta write these emails, right? To make sure that it scores high on certain attributes. And that is also kind of like a balance between speed and craft.

因为刚开始编写提示词时，没必要写200个测试用例，因为随着认知深入提示词会大幅调整。这方面也需要权衡，不知道你是否遇到过这种情况。

Because if you're just starting to write a prompt, you probably shouldn't write 200 test cases because your prompt will change dramatically as you learn. So you you kind of to balance that too. Like, I'm I'm not sure if you've seen seen that.

是的。我们最初完全不设评估标准，随着用户群体扩大，会收到类似'我输入这个却得到意外结果'的反馈。

Yeah. I mean, for us, a lot of our eval development like, we start with, no evals. Right? And then as we kind of, you know, start deploying to bigger and bigger groups of people, people will give us feedback about, hey. I said this and I expected this outcome, but it did this other weird thing instead.

然后我们分析问题根源：需要调整哪些提示词？能否根据反馈建立对应评估标准？这是个动态迭代的开发过程。

And then we can analyze, Okay, what went wrong here? What might we change about the prompt to make sure that we cover this kind of case? And then is there an eval that we can build against this feedback that we just got? So it's a very active and iterative development process.

好的，确实，这比一开始就做200个黄金评估合理多了。对，这样合理得多。那么在这些同心圆里会发生什么？然后人们会说，嘿，如果有这个或那个功能会更好。你怎么保持产品简洁呢？尤其是B2B领域，人们总是想要各种各样的功能。

Okay, yeah, that makes a lot more sense than trying to do 200 golden evals in the very beginning. Yeah, that makes a lot more sense. Okay, so what happens in these concentric circles and then the people are like, hey, this could be a lot better with this feature or that feature. How do you keep the product simple, man? Especially for B2B, people want all kinds of things.

你如何进行这些权衡？

How do you make those trade offs?

是的。我认为没有一个固定的方法来做这些权衡。很多产品经理可能陷入的误区是，他们认为存在某种哲学可以自动替他们做所有决定。比如，'哦，我不需要思考了，因为我们有这个政策'。这个政策可能是'客户永远是对的，所以我们会答应所有需求'。

Yeah. I don't think there's one way to make these trade offs. I think where a lot of like, you know, kind of PMs maybe fall into, like, a trap is that they think that there's some philosophy that you can have that just makes every decision for you. That, like, oh, I I don't have to think anymore because, like, we have this this policy. And the policy could be something like, the customer is always right, so we'll always say yes to every request.

或者'不不不，我们只按一种方式做事，保持极度僵化，不理会任何用户请求'。关键在于选择在哪些方面坚持己见，在哪些方面设置合理的默认值但保留灵活性，同时明白有些边界情况必须覆盖。而在某些领域则保持中立，优先提供可配置的体验。在这个领域，常见陷阱是每件事都做成完全可配置的，这其实不可取——因为相当于把产品设计权交给了管理员，最终用户体验会变得很奇怪，毕竟管理员不一定是优秀的产品设计师。

Or oh, no, no, no, we only want to do things one way and we're going to remain really rigid and not do anything people ask us for. So I think a lot of it is about picking and choosing which where you're gonna be really opinionated and specific, where you're gonna be more purposeful with really sane defaults, but also understand that there's some edges that you have to cover. And then where you're gonna be just fairly unopinionated and and make it a very configuration first, kind of experience. And, yeah, and I think in this space in particular, a lot of the trap is everyone ends up being like fully configurable configuration first for every decision, which is not what you want, because then you're just like releasing the product design choices to administrators effectively. And then everyone has a really weird experience because like their admin may not be the best product designer in the world, right?

这对用户来说会变成非常奇怪的处境。我们的思考方式是：在所有企业决策中，最高阶的是选择行业——我是钢铁厂？软件公司？还是餐厅？

So it becomes a very strange situation for users. So for us, when we think about it, of all the decisions that a company makes, the highest order of decision is what industry am I gonna be in? Am a steel mill? Am I a software company? Am I a restaurant?

听着，我们对这个毫无立场。你想从事什么行业，那是你100%的决定。至于高层战略决策，比如是否采用OKR体系，是否推行战略计划等——

Like, look, we have no opinion about that. Like, whatever business you wanna play in, like, that's 100000%, like, your decision. Right? And the high level strategic decisions, are we going to run OKRs? Are we going to run more strategic initiatives or whatever it is?

我们也不会强加观点。但当涉及开发者日常工作中被迫采取的微观操作时，我们会注入更多指导和默认设置。越是深入这个层面，产品就越聚焦。通过这种方式，我们既能扩展到大企业，保留他们需要的灵活性，又不会让一线员工的体验变得怪异。因为一线员工需要可预测的体验来协助日常工作——他们的本职是设计、写代码、对接客户等等。

We're not going to have a strong opinion about that either. But as it gets more and more specific into individual actions that software developers are taking or are sort of forced to take in their day to day, like minute to minute kind of workflow, that's where we're gonna put a lot more guidance and a lot more kind of like defaults into the product, So it gets more focused as you get into that layer. So that's the general way that we are able to expand to larger and larger companies, retain the flexibility that they want, but also don't make experience totally weird for ICs. Right? Because ICs want a predictable experience that helps them get their job done day to day because, like, their job is to make designs, write code, talk to customers, whatever.

他们的工作不是在Linear里不停点击按钮。这就是我们做权衡的方式。

Their job is not to, like, push buttons inside of linear. So that's how we try to make those kinds of trade offs.

那么IC（一线员工）算是理想客户画像吗？因为他们的需求有时与管理员或经理不同，对吧？

So is the IC kind of like the ideal customer profile? Because sometimes their needs are different from a manager or admin, right? It's like, yeah.

对，对我们来说理想客户是整个公司或组织。如果按席位数量统计，绝大多数用户确实是一线员工。所以我们必须确保他们的体验始终稳定、一致且不令人烦躁。

Yeah, yeah. I think ideal for us, a customer is a company, It's an organization. So in terms of users, are certainly our most ubiquitous user. If you just count the number of individual seats that exist inside of linear, the vast majority of them are gonna be ICs. So that's who we have to make sure that the experience is like kind of always stable and consistent and not annoying, right?

所有这些事情，对吧？因为那确实是我们的首要任务，毫不夸张地说，他们是我们最普遍的用户群体。然后随着你在管理层级中逐步上升，我们必须在不影响基层员工体验的前提下，为他们提供灵活性。

All that kind of stuff, right? Like that's the number one priority, literally because they are our most common user. And then as you kind of move up in the sort of layers of management, we have to give them flexibility without negatively impacting the IC experience.

明白了。这个视角非常到位。是的，这很合理。另外我发现的一个技巧是，当有人提出功能需求时，你可以直接问他们：你想解决什么问题？也许存在比他们当前方案更好的解决方式。

Got it. That's a really good way to look at things. Yeah, that makes a lot of sense. And also another tip that I have found is that when someone's asking for a feature, you can just ask them, hey, what problem are you trying to solve? And then maybe there's a better way to solve the problem than what they're asking for.

对对对，这种情况太常见了。我想告诉所有为产品经理或项目经理群体开发产品的人，你们进入这个领域时默认要接受的现实就是——PM们总会带着极其强烈的观点来找你。他们会说'听着，我有自己的一套方法论和做事流程'。

Yeah, yeah, yeah. That happens all the time. Think one of the things for anyone out there who's building for a PM or project manager kind of audience, one of the things that you're going to go into you know, this this sector assuming is that PMs are gonna come at you with, like, really, really strong opinions. Right? They're gonna be like, oh, yeah.

但事实并非如此。他们之所以会形成这些未必正确的观点，是因为需要说服团队跟随他们的思路。当他们向别人阐述时，会显得像是经过深思熟虑的完美方案，实际上他们也是在实践中边做边摸索。

I I there's my my process and my way of doing things. And that's not actually true. Right? Like, they they have to form those wrong opinions because they have to get everyone to kinda go along for the ride so that they can they when they when they when they explain to people, it's gonna sound like they've just, like, really thought it through, and everything's, like, super perfect. And the reality is, like, it's they're figuring it out as they as they go along.

所以如果你能提供一个现成框架让他们倚靠使用来完成工作，他们会非常乐意接受。

So if you come and you provide them with, a framework that they can, like, you know, lean on and and and use to get their job done, like, they'll they'll be very happy about it.

没错。PM们总是固执己见。但你必须挑战他们的观点。

Yeah. PMs always have strong opinions. Yeah. Yeah. But you gotta challenge them.

好的，这部分很棒。现在我们进入收尾环节，希望你能演示下Linear里新开发的AI智能体功能，或者你如何结合其他AI产品使用它。不如展示几个你最得意的功能？

Yeah. Okay, so that's great. So let's kind of wrap up by actually hopefully having you demo some of the new AI agent or AI features in linear or how you use it with other AI products. Yeah. So maybe you can demo some of your favorites.

好的好的，我正好准备了几项功能。现在共享屏幕——首先从Claude开始。我要展示的都是我日常真实使用的工具，绝非概念演示。

Yeah, yeah. I think I have a few that I have ready for you here. So let me share my screen. First, I'll start inside of Claude. And one of the things that What I'm gonna show you is real tools that I use on a basically a daily basis.

这些都是经过生产环境验证的成熟方案，不是那种'未来可能很酷'的幻想，而是实实在在的解决方案。

Right? These are things that happen all the time that And that work at production grade quality today. Right? So this is not like pie in the sky, oh, it would be cool with. This is like real stuff.

第一个要演示的是我们的MCP服务器。通过它，你可以用当前主流AI工具（比如Cursor或这里的Claude）来访问Linear系统。这意味着你无需复杂配置就能完成常规分析或编码流程——只需OAuth授权连接MCP。现在我要快速分析我们正在开发的'版本管理'项目，这个项目关联着大量客户需求和功能请求。

So the first thing I'll show you is our MCP server, basically. So you can use most of your AI tools available today, Cursor or in this case, Claude, to access linear through the MCP, which means that you can do a lot of the normal analysis or coding workflows in these tools without having to do a whole bunch of weird setup. So you just do a bit of OAuth to connect your MCP, and now I'm gonna do a quick analysis. So we have this project that we're kind of developing right now called release management. And there's a bunch of customer requests and feature requests and stuff like that attached to it.

我只想做个分析。人们都在说什么？帮我开个头。对吧？如果我是这个项目的新产品经理，我完全不了解情况。

I just wanna do an analysis. What have people been saying? Get me started. Right? If I'm a new PM on this project, I just I don't know.

比如，我们正在整理初步收集的信息，目前已经掌握了哪些资料？对吧？所以我打算让Claude来处理。然后Claude会开始使用MCP工具获取项目所有细节、内部问题、客户需求，并开始思考分析。这时候你就可以起身去喝杯咖啡了，对吧？

Like, we're doing kind of the initial batch of, what information have we collected already? Right? So I'm just gonna ask Claude to do this. And then, you know, Claude is gonna gonna start using the MCP tools to get all of the project details and the issues inside of it and the customer requests and start thinking about it. This is where you go, you get up and you go have a cup of coffee, right?

这需要几分钟时间，没关系对吧？因为如果让我手动阅读所有这些材料，轻松就要花上几个小时，前提是我还得能找到这些资料。所以它正在自动检索之类的工作。好了，它找到项目了。

It's gonna take a few minutes to do this, is fine, right? Because if if I was gonna manually read all this stuff, it would take hours easily, provided that I could even find the stuff, right? So it's doing all the automated lookup and things like that. Here we go. It found the project.

它查看了部分问题，正在检索其他内容——就像在说'嘿，我找到了主要问题集，现在要查找可能带有不同参数的其他问题'。它正在专门查找已发现的其他问题的具体细节，明白吗？这就是正在进行的流程。

It looked at some issues. It's looking at some other, you know, it's like, Hey, I found the main body of issues. I'm gonna look for other issues that are maybe, they maybe have different parameters. So it's looking specifically for some details for these other issues that it found, right? So that's what's going on here.

现在它要生成报告了：'发现28个相关客户需求'，然后给出我对这个主题下用户诉求的分析。所有这些信息都存储在Linear里，所以需要回溯时，可以要求它提供引用来源之类的。这是第一个切入点，对吧？

And now it's gonna write, Hey, I found 28 customer requests against this. And then here's my analysis of what people have been asking for around this subject. And all this information is like inside of linear, right? So if I ever wanna kind of link back to it, can ask it for references and things like that. And this is the first kind of starting point, right?

这是Claude进行的初步基础分析，是对话的起点。这时候你可以开始互动交流，比如问'这里面哪三个主题领域最突出？'

This is the first basic analysis that Claude is making. This is the start of a conversation, right? So at this point you kind of go back and forth. I might say, you know, what are the top three thematic areas within this? Right?

所以我可以说：'好，我需要关注的三大重点是什么？'比如发布可见性——对，我也这么认为。

So I can sort of Okay. What are the three big rocks that I'm looking for? Right? So it's like release visibility. Yes, I think so.

我思考这个问题很久了，完全同意这个观点。发布管理的核心就是让开发团队之外的人员能清楚掌握版本动态，直接与版本控制平台对接，这样当二进制文件发布时，Linear能自动获知。这涉及工作流构建、变更日志生成等环节，与我所有的用户访谈结果高度吻合。

I've been thinking about this for a while. I agree with this. A big part of release management is visibility for people outside of development team, directly implementing with version control platforms so that when the binary goes out, Linear knows about it. And it's sort of like workflow, kind of building the change log and all that kind of stuff. So like this very much jives with all my user interviews and things like that.

而且对于单纯想知道'我们研究进展如何？从客户那里学到了什么？'的人来说，直接这样操作就行。任何安装了quad的Linear员工都能做到。

But also it's for somebody who just wants to know like, Hey, what's the state of our research? What have we been learning from customers? You could just do this. Anybody who has quad installed, who works at Linear can do this.

太棒了。确实，如果没有MCP工具，我就得把大量内容复制粘贴到quad里提问，特别麻烦。所以这个工具真的解决了...

Yeah, this is awesome. Yeah, even if I had to, without the MCP, I'll have to do a bunch of copy and pasting into quad, which is a pain to ask. So yeah, there's a lot of-

对对对。当我们讨论工作流程时，这就是关键所在。就像，你本可以手动完成这些——把所有数据导出到CSV再粘贴到这里。但我说不，我只执行了两条命令。这才是理想的工作状态。

Yeah, yeah. And when we talk about workflows, that's what this It's like, yeah, you could have done all this. I could dump everything to a CSV and throw it into here and copy pasting them. It's like, no, I issued two commands. That's kind of where you wanna be.

明白了。MCP系统实际上会做出修改吗？还是主要用于读取信息？

Okay. The MCP actually make changes to, or is it mostly reading information?

目前是读取状态。我们正在开发合适的接口端点，

Right now it's it's reading. You know, we're we're developing the right endpoints as well,

但你必须成为某种

but you kinda have to be a

这方面需要格外谨慎，因为你肯定不希望代理程序失控，导致数据被意外覆盖。

little bit careful with those because you don't want you don't want agents going, you know, going completely wild and overwriting about your data.

明白明白。好的，很酷。

Yeah. Yeah. Got it. Okay. Okay, cool.

接下来你想给我演示什么用例？

And what's the next use case that you wanna show me?

下一个用例是我们昨天刚发布的——就在这次录制前。让我找给你看，这个很酷。其实我们内部已经用了好几个月，昨天才正式对外发布。

The next use case is something that we actually just released yesterday as of this taping. So let me find it for you here. It's cool. This is something that, you know, we've been using internally for months, right? We just, you know, did the full public launch yesterday.

它叫产品智能系统，我把窗口调大些。简单说明下：在Linear系统里有种叫'分诊'的概念，当有人提交新缺陷或功能请求时就会进入这个状态。原本我们提供的是非常结构化的处理流程，需要指定专人监控，确保提出正确问题并转交对应负责人。

It's called product intelligence and I'm gonna just make it a little bit bigger. And then, just to give you basic framing, In linear, there's a concept called triage. So like whenever someone reports a new bug or requests a new feature or something like that, it goes into this triage state. And originally, we were provided this very kind of structured way for you to kind of assign somebody to monitor it, to make sure that they're asking for the right questions and directing to the right people. So that's good.

结构化流程虽好，但问题在于过度依赖处理人是否具备足够的组织知识储备。如果他们不清楚，就得四处询问，整个处理过程就变成了调研工作，对吧？

You have a structured process. But the problem with that structured process is that you're kind of reliant on this person to have enough institutional knowledge to know what to do with all of these incoming issues. If they didn't know, would have to ask somebody. It'd be a whole research process basically for them. Right?

他们可以用这些工具来搜索东西。然后，到了某个时刻，我们就在想，好吧，看看我们到底能自动化多少？能不能把这个枯燥的研究过程彻底转变成一个小游戏，让你可以像这样操作，哦，对，好的。

They could use the tools to, like, search for stuff. And, you know, at some point, we're like, okay. Let's how much of this can we actually automate? Can we, like, completely transform this from a tedious kind of research process to a, like, to a little mini game where you can go like, oh, yeah. Okay.

太好了。是的。对。对。对。

Great. Yeah. Yep. Yep. Yep.

然后你批准一些建议。比如这个例子，有人说，嘿，我基本上想允许将模板应用到现有项目上。然后产品智能马上找到了几个匹配项。它说，嘿，我觉得我找到了合适的团队、负责人和项目，这里还有个可能适用的标签。如果你把鼠标悬停在建议上，比如，为什么推荐Maschak负责这个？

And then you approve some kind of suggestions. So for this example here, someone said, hey, I want to allow applying templates to existing projects, basically. And immediately, Prod Intelligence found a couple of things. It's like, hey, think I found the right team, the right assignee, the right project, and then here's a tag that might make sense for it. And if you hover over the suggestions like, hey, why was Maschak suggested for this?

原因很简单，他是项目模板项目的负责人。而且他还负责过几个与此相关的问题。

And it's just like, well, he's the lead for the project templates project. And, and he's also been the assignee for several related issues to this.

哦，不错。

Oh, nice.

所以我觉得他是合适的人选，对吧？即使我不指派Majek，我也知道如果有问题该问谁。嘿。

So I think he's the right person. Right? So even if I don't assign Majek, I know, like, if I have a question, I know exactly who to ask. Right? Hey.

这个——这个建议合理吗？公司里有100号人，我该问谁？很明显，关于这个问题，我得问Machik。它还发现了一个重复项。

This is a is this sensible, whatever it is? Like, you know, there's 100 people in the company. Who do I ask? It's like, well, clearly, I'm asking Machik, like, this particular question. It also found a duplicate.

它说，嘿，我觉得这基本上是一样的。标题不同，但本质上是在请求同一个已经存在的问题。这是解释。

And, and it said like, hey. I think this is basically the same. It's not the same title. But I think it's basically asking for the same thing as this other issue that already exists. And here's the explanation.

看起来它直接针对同一件事。有多个客户需求。而找到的那个问题已经有一堆客户请求了。所以它认为，哦，那个可能是主要的。所以如果你要处理，应该把这个标记为那个的重复项。

It looks like it directly addresses the same thing. It has multiple customer. And the one that found has a bunch of customer requests on it. So it thinks like, Oh, that's probably the main one. So if you're gonna do anything, you should mark this one as a duplicate of that one.

所以它对应该采取什么行动有点自己的看法。我觉得是对的，所以我要点这个勾选标记。它说，标记为重复项。很好。

So it has a little bit of opinion about what action I should take. So I'm like, yeah, I think that's correct. So I'm gonna hit this check mark. It says, mark it as duplicate. Great.

我已经处理好了这件事。现在所有与这个新客户请求相关的信息、这份新报告都会被合并到那个主要问题中，然后我们就能继续推进工作了。

And I've taken care of this. So now all of the information that's attached to this new customer request, this new report gets merged into that main issue and then we can kind of move on with our lives.

哦，太棒了。这样既能添加相关元数据，又能发现其他类似的工单。对对对，否则我就得手动处理，要花很多时间整理待办列表。

Oh, that's great. So it both adds relevant metadata and also finds other tickets that are similar. Yeah. Yeah. Yeah, because otherwise I have to do this mali and they'll take, yeah, like try to prune the backlog.

上面还写着'付费咨询'呢。

It says, pay to ask too.

是啊是啊，我知道。这就是我们喜欢的功能之一——'整理待办列表'算是你们的工作对吧？就像在说：为什么雇这么多产品经理？总得有人来整理待办列表啊。那如果我们把这个效率提高100倍呢？

Yeah, yeah, I know. It's one of the things that we just like to, oh yeah, that's one of your jobs, prune the backlog, right? It's like, why did you hire a whole bunch of PMs? Well, we got to have somebody to prune the backlog. It's like, Okay, what if we made that like 100 times easier?

这太赞了，真的太赞了。老兄，我觉得或许还可以做个公开界面展示正在处理的工单。你知道有些公司会和用户共享产品路线图吗？这对Lehin可能也很有价值。

Yeah, this is awesome. This is awesome. I can dude, I feel like maybe there's also a way to have a public facing thing where you can see the tickets that people are working on. You know how some companies share their public roadmap with their users? That could be interesting for Lehin here too.

没错。最棒的是当你查看我刚合并的内容时，能看到所有关联的客户请求，这些都是其他问题被识别为重复后归集过来的，因为系统能自动识别。

Yeah. Yeah. The nice part is when you go to the thing I just merged it into, right? If I look at this and you can see all of these customer requests that are attached to it, all of these things are a result of other issues being duplicated into this, right? Because the system knows.

所以当新请求进来时，系统会立即识别：'我知道这个，这是某个大问题的重复项'。标记后数据就会自动累积。产品经理看到这个附带大量客户请求的大问题时，就可以启动Cloud说：'帮我分析这些客户请求'，梳理出所有细节需求，整个系统就是这样协同工作的。

So as soon as another request comes in, it goes like, I know what this is. This is a duplicate of this other big, issue over here. And then you just mark it and it kind of accumulates the data. So then when a PM has to come in and they see this big issue with a bunch of customer requests on it, then they fire up Cloud and they go like, help me analyze the customer request and then figure out what's, like, being asked for here and all the little nuances and details, right? So, like, the whole system kinda works together.

那有没有办法把这些内容都汇总起来？比如...

So is there a way to, like, summarize all all other things here? Like,

目前应用里还不行，但这是MCP的用途之一。我可以复制问题ID让Claude来'总结一下...'

Not not not in the app right now, but it's this is one of the this is one of the uses for the MCP. Like, I could, like, copy the the issue ID and ask Claude to be, like, Okay. Summarize for

明白了。这样很合理。因为...确实很酷。

Okay. That that makes sense. Yeah. Because Cool. Yeah.

关于钢琴动物园的阅读越少越好。是的。

The less reading of the Piano Zoo, the better. Yeah.

对，就是这个意思。另一个案例也是类似情况，属于待处理队列里的。这个也是同类问题，对吧？就是一堆建议。它某种程度上指向了这个团队。

Yeah, that's the idea. Then another case, is another one of these, right, things that's in triage. This one is the same sort of thing, right? It's a bunch of suggestions. It kind of suggested this team.

我觉得有意思的是，这个产品团队负责列表排序逻辑。很多时候我们团队很少，你甚至不知道哪个团队该处理这类问题，懂吗？这甚至不是记不清具体负责人的问题，而是根本不知道该把任务派给哪个团队。

And I think what's interesting here is that this product team owns the list sorting behavior. And I think that idea, we have very few teams, a lot times you're like, I don't even know what team does you know, is meant to like, think about this. Right? It's like, it's not even like, forget what person. It's like, I don't even know what team to to to put this on.

对吧？所以这种情况至少可以转交给他们，放进待办队列，后续让他们决定怎么处理。你不需要立即采取完整行动，能推动进度就够了。

Right? So so this is one of these things where like, look, at least I can I I can maybe just hand it over to them, put it in their triage queue, then they can they can figure out what to do with it next? Right? So you you don't have to, like, fully take action on it. You can it's enough for you to, like, move it forward in the progress.

没错。关于这个功能，我看到很多人在Slack上问：等等，这个工单该哪个团队负责？往往要等很久

Yeah. With with all this feature, I've I've seen a lot of people just asking on Slack, hey. Wait. Which team is responsible for this ticket? It takes forever

是啊，耗时太久。等你追问时发现没人回复，一周后才发现问题还卡在那里。现在你不用再操心这些了。

Yeah, takes forever. Do that and then you forget to, know, no one answers and a week later, you're like, oh, wait, no one answered my question. And it's still sitting here. You don't have to deal with any of that anymore.

明白了。最后还有什么要展示的吗？

Okay. All right. And last but not least, you have something else to show us?

有的，分两部分。首先想给你看下面这个——客户提出的功能请求：希望能将周期与日历同步。讨论中我们突然发现：等等，这个功能是不是已经有了？

Yeah. It's actually in two parts. The first bit I want to kind of show you is actually down here, which is like this person asked for a This is just another feature request from a customer. What it asked for is, hey, I want to sync my cycles with calendars. And in the discussion thread, we discovered like, wait, do we already have this?

我当时也疑惑是否已存在。支持团队的同事确认说应该有的，还提供了相关讨论链接和文档。

I'm like, do we already have this? And then this colleague of mine, he's the support team. He's like, yeah, I think we do. Here's where we talk about it. And then maybe he found some documentation or something.

但我想彻底确认。所以我准备咨询工作空间里的一个智能体。注意这不是我们开发的，而是第三方开发者基于我们平台创建的编码智能体。

But I really wanna make sure. So what I'm gonna do is I'm gonna ask one of the agents that's living in our workspace. Right? And this is not an agent developed like us. This is a third party coding agent that somebody else develops that's using our agent platform.

我直接问问这个代理吧。这个代理的名字恰好叫查理。我们会把周期与日历同步吗？对吧？我就这么问问。

And I'm gonna just ask this agent. The agent's name happens to be Charlie. Do we sync cycles to calendars? Right? I'm just gonna ask.

就像对人提问一样。你之前提到在Slack里@某人问'这种情况该怎么处理'对吧？现在查理会看到这条消息，它会回复'我会检查一下'，然后查理就会去查看代码库。

And just like it would a person. So the thing you said about you go to Slack and you at mention somebody like, hey, what do we do here? So Charlie's gonna see this. It's gonna say, hey, I'm gonna review this stuff. Then Charlie's just gonna look in the code base.

因为我们都经历过这种情况。作为工程师，我们会想'系统真的是这样运行的吗？'然后就得去问其他工程师，他们再翻查代码。最后得到答复'对，看起来确实是这样运行的'之类的。

Because we've all been through this. We're like, we're an engineer. We're like, hey, is this how the system actually behaves? And then you have to ask an engineer, then they dig through the code. Yeah, okay, yeah, it looks like that's exactly what it's behaving as expected or whatever.

我们都干过这事。但以后不用再这样了，对吧？不必打扰工程师，占用他们时间中断手头工作，就为了让他们帮你确认代码确实如你所想的那样运行。

We've all done this. You don't have to do that anymore. Right? You do not have to bother an engineer, take time out of their day to drop what they're doing, to go do a little research project for you to just tell you that the code base works the way that you think it does. Right?

你可以直接让代理帮你查。虽然需要几分钟，但想想工程师做这个得花几小时。所以咱们先去喝杯咖啡，回来时就能看到留言告诉你系统是否真是这样运作的。

You can just ask an agent to do it for you. It's gonna take a few minutes, but again, like engineers gonna take a couple hours to do this. Right? So we're again, go go have a cup of coffee. And then, you know, when you when you come back, there's gonna be a comment on here that tells you whether or not this is how this is how the system works.

那我们现在先稍等片刻，等着结果出来。

So we're gonna we're gonna chill here for a second and and wait for that to happen.

太棒了。而且这风险很低对吧？因为它只是读取代码，没有做任何重大修改。

That's great. Yeah. And this is pretty low risk here, right? Because it's just trying to read the code. It's not making any major changes or anything.

没错。趁等待的时候，我再给你演示另一个场景。这是我准备的另一个问题。

Yeah. Yeah, exactly. And then I'll sort of While we're waiting for this, I'll sort of show you a different scenario. Okay. Here's another question I have prepared.

关于某某功能的调整...这次我想问'我们执行这个操作用的系统提示词是什么？'因为系统表现有点奇怪，我想知道实际用的系统提示词。咱们看看会得到什么回答。

Tweaks in blah, blah, blah. Yeah. So in this case, I'm like, hey, what's the system prompt we're using to do this? I'm like, the system became kind of weird, like what's the actual system prompt? So I'm gonna ask, you know, I'm gonna ask that and we'll sort of see what happens.

好了，我们等了一阵子对吧？喝完咖啡回来，查理回复说'是的，周期可以按团队通过ICS协议同步日历'。这证实了我们客服团队根据讨论记录产生的猜测，我只是想再确认下。

Okay. So we, we waited for a while, right? We had a coffee for, for Charlie to do his work and, you know, and we came back and Charlie says, yeah, cycles can sync the calendars, per team, a, on a ICS. So like, this is, you know, confirming what our, you know, what our support team member suspected, right, based on the discussion that you found. I I just wanted to double check.

我就说，我从来没听说过这个功能。我只是想确认这真的存在。然后查理就说，没错，我在代码库里找到了。这这这是个热代码路径。

I'm like, I've never heard of this feature before. I just wanna make sure this is actually happening. And then Charlie's like, yeah. I found it in the code base. It's it's it's a hot code path.

我们没问题。而且，通常你知道，我可能会去找个工程师确认这事，他们就得放下手头工作。他们会思考这个问题，然后我会等他们回复。现在我就要欠他们人情了，对吧？

We're good. And, like, normally, you know, I I might go to a like, an engineer to confirm this, and they're gonna have to drop what they're doing. They're gonna think about it, and and, you I'll I'll wait for them. Now I'm gonna owe them a favor. Right?

通常事情就是这么发展的。但现在我不需要这么做了对吧？只要这个问题存在，上下文都在那里，你能推导出来，代理就能帮你找到答案。

Like, that's that's usually how these things go. I don't have to do that anymore. Right? Like, this is this is a question that as long as, you know, it exists, right? There's the context out there, and you can reason through it, the agents can find an answer for you.

这太棒了。是啊，真不错。

This is awesome. Yeah, that's great.

好的。那么最后要说的是，查理和我们平台上的其他代理都是编码代理对吧？它们真的能帮你写代码。我来演示下，但我要切换到个人工作区，因为我不想意外在生产代码库里写任何东西

Cool. All right. So the last sort of thing is Charlie and other agents that we have on the platform, they're all coding agents, right? So they can actually write code for you. So I'll demonstrate that, but I'm gonna switch workspaces to my personal workspace here because I don't wanna write any code in production code based on accident

在操作过程中，

while doing And,

好了。为了铺垫下场景，我有个个人网站，这个网站是基于静态站点生成器的。我给你们看下它长什么样，这就是我的个人网站。

all right. So to sort of, you know, kind of kind of set the stage here, right? I have a I have a personal website and, that personal website, it's it's it's based off of a static site generator. I'll sort of show you what it looks like, right? This is my personal site.

就是这个，这是开发版本。你们能看到现在是暗色模式对吧？它会跟随你的操作系统设置。我现在想做的是，给用户加个控件，让他们能自由切换到亮色模式。

Here it is. This is like the development version of it. And you can see that it's in dark mode, right? And what it does is because my operating system's in dark mode, it follows your OS. And what I wanna do, let's say, is I wanna add a little control for users to be like, hey, if you wanna just switch it to the light mode version of it, like go ahead and do that.

这没问题。但我不想自己写这个功能——要设计按钮、搞定CSS和本地存储什么的。好吧无所谓，我就像平常那样写个需求。

That's no problem. But I don't, like, I don't wanna write that feature. Like, I gotta design a button and figure out how to do the CSS and local storage and everything. Okay, whatever. So I'm just gonna write the issue like I normally would.

比如：现在是通过媒体查询控制的，但我想让用户能手动切换，并且保存在本地存储里。不要用媒体查询，要变成可控制的。我对样式也有要求：就用简单文字链接当控件，不要按钮或表情符号，不要边框，纯文字就行。

Like, hey, right now let's control our media query. But I wanna give people a way to toggle on and off without And have it saved at local storage and then they just do it there. So instead of being you know, media query, I want you to have control. And I also have some opinions about styling, right? I would like use a simple text link as a control, no buttons or emojis or anything, no borders, you know, just plain text.

对吧？那样还行。不，我是说，我对想要的外观非常挑剔。明白吗？

Right? Like that's Okay. No. I mean, I'm very specific about the look that I want. Right?

因为我网站上没有任何图标。所以，你看，如果我想做这个，我可以自己写代码，或者雇个工程师，或者直接交给平台上某个编程代理来做，对吧？Linear是个开放平台，代理程序可以部署到Linear里。任何为Linear开发支持的编程代理都能做到。

Because I don't have any icons on my website. So, you know, if I wanna do this, I could write the code or I could hire an engineer or I could just assign it to one of the coding agents that lives on the platform, right? So Linear has an open platform. Agents can be deployed into Linear, right? Any kind of coding agent out there that's built support for Linear can do so.

我们在这个工作区安装了一个叫CodeGen的代理，合作公司非常优秀。我只需把这个任务分配给它，接下来你会看到有趣的事情——我负责这张工单，而实际工作由CodeGen完成。

So we have this one installed in this workspace called CodeGen, right? It's a really great company that we work with. And what I can do is I can just assign this issue to this agent, right? So when I do that, you'll see something interesting happens here, right? I have a responsibility for this ticket and CodeGen is doing the work, right?

这是Linear直接集成的功能。在生产环境部署代理时，最关键的是必须有人对结果负责。绝不能出现'机器人干的'就推卸责任的情况，无论是出了bug还是做了优秀功能，都需要明确责任人。

So this is something that we directly express inside of linear. Because one of the things that you really, really wanna make sure when you're deploying agents into your production environment is that some human being has responsibility for the outcome. This is really, really important. You can't just be like, oh, the robot did it, and it shipped a bug, or it dropped the table, or whatever it is. Or if, it did something well, hey, like, is a great feature.

谁开发的？说是机器人做的可不行。必须有人与机器人深度协作开发功能。所以我把任务委托给了CodeGen。

Who built this? It's like, well, the robot builds it. Like, no, no, yeah. But someone like, know, was working with the, the robot in kind of a high bandwidth way to, to build a feature, right? So, you know, so I delegated it to cogen.

现在CodeGen正在工作，这个连接线程可以看到它的思考过程。打开面板就能观察：它正在像普通编程代理那样规划步骤，从CSS开始逐步实施。和其他代理一样需要些时间，你可以去喝杯咖啡，回来应该就能看到拉取请求了。

So cogen's working. You can see this is like the connected kind of thread here. If I wanna understand the sort of chain of thought and what's going on, I can open up this panel and sort of see, okay, here's what cogen's doing, like the sort of normal coding agent kind of thing where it's, you know, thinking about what it should do first, making a plan, starting with the CSS, right, and kind of going from there. You know, like any other coding agent, it's gonna take a little while. So again, go have a coffee and you'll come back to a pull request hopefully.

关键在于你可以像我这样实时监督，也可以高效操作——工程师都有大量待办工单，有些简单任务完全可以全选后批量委托给CodeGen或其他代理。

Okay. The interesting part about this, right, is that you can choose to babysit it like I'm doing right now. Or, you know, if you're really efficient, right, you can you know, you have a you have a backlog. Like, every engineer's got, a whole bunch of tickets assigned to them, and some of them are, you know, kinda low hanging and and can probably be achievable in this way. You can literally just go, like, command A, select all, and then delegate it to cogen or to any other agent that you're using.

这是全新工作模式，你可以同时启动100个这样的任务。我的CodeGen已经返回结果，生成了PR并附上了工作摘要。

This is a new pattern that you can just do. You can just spawn off 100 of these simultaneously if that's what you want to do. Then they're all going to eventually come back. My cogen just has here and says like, hey, here's the PR, right? And then, you know, I've done, you know, here's here's a here's a small sort of, you know, kind of summary of what I did.

对吧？如果想查看，我现在就打开这个PR。

Right? Yep. And if I want to, I can I can go to that PR? I can check out I'm gonna go to the PR right now. Here it is.

我要复制分支代码，检查它的修改。来看看这个实时成果。

I'm gonna copy the branch. I'm gonna check it out and see what it did. Right? So let's take a look. This is some live stuff here.

让我们看看发生了什么。好的，我已经把它拉下来了，看看它做了什么。好极了，它在这里添加了一个小链接，我要点击它，它会切换到亮色模式，对吧？

So let's see what happened. Okay. So I've pulled it down and let's see what it did. Okay, great. It added this little link up here and I'm gonna click it and it's gonna switch it to light mode, right?

它完全按照我说的做了。我说，给我一个角落里的非常小的链接，只用文字，不要做成带图标的按钮，要非常简单。它基本上做到了这些。

So it did exactly what I said. I said, give me a very small link in the corner, make it only text, don't make it a button with an icon, have it be really simple. And it kind of did those things.

是的，太棒了。现在你可以设置PR，然后，是的。

Yeah, that's great. And now you can set the PR and then, yeah.

是的，我可以设置PR，可以合并它，可以做任何事。我还可以手动调整它，对吧？甚至可以回到问题中，对吧？如果我愿意，可以要求Cogent进行修改。所以这是一种非常自然的互动方式。

Yeah, I can set the PR, I can merge it, I can do whatever. I can go and manually tweak it, right? Can even go back into the issue, right? Ask Cogent to make modifications if I want to. So it's a thing, it's a very natural kind of interaction that you can do.

好的。这太棒了。谢谢你展示了所有三个用例。是的，它既能阅读代码又能修改代码，这很棒。你甚至可以让多个代理同时提交多个PR。

Okay. This is awesome. Thanks for showing all three use cases. Yeah, it's great that it can both read a code and also make changes to the code. Even you can get multiple agents to submit multiple PRs at the same time.

你可能应该更仔细地检查它的工作，但你可以做到。是的。让我们快速结束这个问题。Linear最初是一个简单的工单跟踪工具，对吧？未来的方向是什么？

You probably should double check it's work, even more, but you can do that. Yeah. So let's wrap up this quick question, man. Linear started as a simple kind of ticket tracking tool, right? And what is the future here?

未来是要构建这种操作系统，让代理和人类可以互相交流并共同开发产品吗？是这个想法吗？

Is the future here to build this kind of operating system where agents and humans can talk to each other and work on a product together? Is that kind of the idea?

是的，我认为这几乎是近在咫尺的未来。它是一个系统，因为现在，如果你放大看，Linear本质上是一个协调许多独立行动者工作的系统。代理在很多方面也类似。它们不同，对吧？不同之处在于我们讨论过的那些方面，比如你不能让它们对结果负责之类的。

Yeah, I mean, think that's what almost the very near future looks like. It's a system because right now, linear is ultimately, if you zoom out enough, it's a system where you're coordinating work across a lot of different independent acting people. And agents are shaped a lot like that. They're different, right? And they're different in that, in the ways that we talked about where you can't hold them accountable for outcomes or anything like that.

所以你必须继续让人参与其中。但它们也确实在很多方面表现相似，它们做决定，可以独立工作，可以汇报它们做了什么，可以完成所有这些活动。所以在很多方面，它们是Linear系统的新用户，当然也有一些特殊的注意事项。

So you have to continue to involve people in the mix. But they also do act in a lot of the same ways where they make decisions, they can do work kind of out of band. They can report back on what they did. They can do all of these kinds of activities. So in a lot of ways, they're a new kind of user of linear system, again, with sort of like special caveats.

好的。希望如此，我想我会启动一个新的Linear项目，我会引入一批代理，是的，为我设计和开发。生活会更美好。是的。

All right. So hopefully, I think I'll start a new linear project with I'll onboard a bunch of agents Yeah. Cool for me and design for me. Yeah, life will be good. Yeah.

好的，太棒了。那么现在，人们在哪里可以在线找到你并了解更多信息呢？

Yeah. Cool. All right now, well, where can people find you online and learn more?

是的，我觉得在X或Twitter上我最活跃。我的账号是thehnanyu，T H E N A N Y U。私信我，看看我分享的内容，那是最佳的联系方式。

Yeah. I think I'm most visible on X or Twitter. My handle is thehnanyu, T H E N A N Y U. And DM me, kind of take a look at the stuff I'm sharing and then that's the best way to contact me.

非常感谢，兄弟。能再次联系总是很棒。

Thanks so much, man. It's always great to connect again.

是啊，当然，彼得。

Yeah, yeah, of course, Peter.