Everyone’s shipping more. Does any of it matter? | Claire Vo
频道: Lenny's Podcast
视频: https://www.youtube.com/watch?v=VM5kuvWgwDY
原文语言: en
统计: 共 17 轮 · Claire 17
[0:06] Claire
Good morning everyone. How's it going? Everybody excited to be here?
大家早上好！状态怎么样？来到这儿都挺兴奋吧？
[0:12] Claire
Okay, for those of you that were in the Somebody texted me, they called it the How I AI sauna last night, the party we had. I told you all you need to hype me up this morning because I was out way past my bedtime, which is 8:15. I've got a little baby. Um but I am here prepared to do the thing that I do. So last year, last year I came here and I said product management is dead. I said that because as a product leader I like to make these big grand statements and then be proved completely wrong in the market. Just keeps me really honest. So product management is obviously not dead. There are so many amazing product leaders, product managers, product executives here in this room. But who's with me that between the last time we had Lenny's summit and now product management is completely different. Okay, got some hands. It's really different. And so last year I came for PMS that didn't work. Let's do it again. This year I'm coming for the road map. The Yeah, you're you're welcome. OKR's next and then my job is over.
好，昨晚在那个……的朋友们——有人给我发消息，说大家管我们昨晚那场派对叫「How I AI 桑拿房」。我跟你们说过，今天早上你们得给我打打气，因为我昨晚在外面玩到远远超过了我的睡觉时间——我的睡觉时间是 8:15。我家里有个小宝宝。嗯，不过我已经准备好了，来干我最拿手的那件事。去年，去年我站在这里说：产品管理已死。我这么说，是因为身为产品负责人，我就爱放这种宏大的豪言壮语，然后被市场证明错得一塌糊涂。这样能让我一直保持清醒。所以，产品管理显然没死。这个房间里就有那么多了不起的产品负责人、产品经理、产品高管。但有谁同意我这个说法：从上一届 Lenny 峰会到现在，产品管理已经完全变了样？好，有人举手了。真的大不一样了。所以去年我拿 PM 开刀，没成功。那就再来一次。今年我要拿路线图开刀。这——对，不用谢，不用谢。下一个就轮到 OKRs，然后我的使命就完成了。
[1:23] Claire
[laughter]
（笑）
[1:23] Claire
So the road map has been the defining artifact of our industry, of our careers. It's supposed to tell us and our teams where we're going, what's next, and what matters. And I have been in product for over two decades. And I believe, fingers crossed, a lot of us in our room in this room are about to write our last road map. Before I go into why I think that, I want to tell you all I have a confession. I am shipping more than ever. I have access to the most intelligent coding agents in the world. I have the best developer tools anybody can ask for. I have super smart models. I've got customer context via API, via CLI, via MCP. all the letters I have. I write so many skills. And y'all, I counted yesterday. I have 40 Grock bots. I can build almost anything. And here is my true true confession. I am out of good ideas, y'all. I'm like straight up out of good ideas of things to build. I'm out. I can build. I just don't think any of it's a good idea. So, and and I'm not saying I don't have plausible requests. It's not that I'm saying I don't have things I could build. It's just that execution has outrun my ability to discover meaningful, meaty, my favorite word, commercializable products in the market. And this is a very different situation that I have been in than before. You know, it used to be that engineering capacity was the scarce resource. That was like I mean that's kind of why we exist honestly is like engineering capacity used to be the really scarce resource. We had way more ideas than people and we had way more demand than our ability to deliver against that demand. And the product manager, the products leaders job was to prioritize all those ideas, sequence them, and then spend like all our time in meetings and in Slack and in spreadsheets saying no. Saying no. PMs should say no more. we should have cut lines. It was like very aggressive like here's my cut line for the quarter. This
路线图一直是我们这个行业、我们职业生涯的标志性产物。它本该告诉我们和我们的团队：要往哪儿走、下一步做什么、什么才重要。我做产品已经二十多年了。我相信——但愿如此——这个房间里我们当中的很多人，都快要写下最后一份路线图了。在讲我为什么这么想之前，我得先向大家坦白一件事。我现在交付得比以往任何时候都多。我能用上全世界最聪明的 coding agent。我手里有谁都梦寐以求的顶级开发者工具。我有超级聪明的模型。我能通过 API、CLI、MCP 拿到客户上下文——所有这些字母缩写，我全都有。我写了超多 skill。各位，我昨天数了一下，我有 40 个 Grok Bots。我几乎什么都能做出来。而我真正、真正要坦白的是：各位，我已经没有好点子了。我是彻彻底底想不出有什么值得做的好点子了。一个都没了。我能做，只是觉得哪个都不是好主意。我不是说我手上没有像样的需求，也不是说我没有能做的东西。只是执行力已经跑到了前面，超过了我在市场上发现好产品的能力——那种有意义的、有分量的，还有我最爱的那个词：可商业化的产品。这和我以前的处境完全不一样。要知道，过去稀缺的资源是工程产能。说实话，这差不多就是我们之所以存在的原因：工程产能曾经是真正稀缺的资源。我们的点子远远多于人手，需求也远远超出我们的交付能力。而产品经理、产品负责人的工作，就是给这些点子排优先级、排先后顺序，然后几乎把所有时间都花在会议里、Slack 里、电子表格里说「不」。说「不」。PM 就该多说「不」。我们就该有砍线（cut line）。那时候真的非常强硬，比如：这是我这个季度的砍线。这
[4:05] Claire
is my priority list. This is my stack rank. And while the PM's job was to say no, it felt like engineering's job was to say like no, not like that. Not with all those features, not with perfect architecture. We can only do this much. And then it seemed like design's job was say to like no, wait for us, like please. And so everybody's like just constrained on this like precious engineering capacity. And so things got narrow narrow narrowed down till you only shipped a little bit of your road map. And now I feel like I have more [clears throat] execution capacity than like true conviction about what to build. So my bottleneck has moved from like building what what can I build to believing like what do I believe is actually worth building. And look, I am like the biggest token maxer on the planet. I'm like PR's up 3x. Like do it all. Ship, ship, ship. Agents everywhere. I don't need a PM. I don't need any specs. But like I think this gap between our building capacity and like true in our heart of hearts conviction like does any of this code matter is a much bigger problem than any of us are willing to admit. I think we're all getting a lot of pressure from the board, from the timeline, from each other, from our teams, from our bosses to move faster, to get leverage with AI, to like really lean in and show we're AI native. We can build in this new way.
是我的优先级清单。这是我的强制排名。PM 的活儿是说「不」，而工程那边的活儿感觉像是说：「不，别那样搞。别塞那么多功能，别追求完美架构。我们只能做这么多。」然后设计那边的活儿好像是说：「不行，等等我们，拜托了。」所以每个人都被这份宝贵的工程产能卡着。于是事情被一砍再砍、越收越窄，最后你只交付了路线图上的一小部分。而现在，我觉得我的执行产能反而超过了我对「该做什么」的真正信念。所以我的瓶颈已经从「构建」——我能做出什么——转移到了「相信」——我相信什么东西真正值得做。听着，我大概是地球上头号 token 狂（拼命烧 token 的那种人）。我就是那种：PRs 涨了 3 倍！全都做！发布、发布、发布！agent 遍地都是！我不需要 PM，也不需要任何 spec。但我认为，我们的构建能力，和我们内心最深处真正的信念——这些代码到底有没有意义——两者之间的这道鸿沟，是个比我们任何人愿意承认的都要大得多的问题。我觉得我们都承受着巨大的压力——来自董事会、来自社交媒体时间线、来自彼此、来自团队、来自老板——要我们跑得更快、借 AI 撬动杠杆、真正全情投入，证明自己是 AI 原生的。证明我们能用这种新方式来构建。
[5:38] Claire
Claire said product management is dead. She showed all these scary circles. She said like ship to the moon. Do it, do it, do it. But like I do think none of us are I know this is a question I asked one who's who's shipping more than ever in the last year at their companies whose revenue is going up proportionate with the amount of PRs like that's the fundamental problem we're facing here and so I'm going to tell a story and like bring this home for you this like I have no good ideas and I don't know if they matter. Um so I had this idea for this thing called the product graph for chat PRD. It's like a It's a real product for product managers idea and it's not special because everybody's done it. But it's like we're going to suck in everything a company knows about customers, about what you're working on, blah blah blah blah blah. We're going to parse it. We're going to like slap, you know, opus on it and then it's all going to be available through agents and it's going to be awesome. It's going to make you be able to make these decisions better and build better products and of course write better PRDs. And so I built it. I built this like insights engine. I built this like very fancy semantic product graph. It like generates this auto wiki. Agents can consume it. It was like almost a oneshot. Not quite, but it was a pretty close oneshot. It worked. It looks pretty good. It was like feature forfeature matching competitors in the market. And I kind of just like every time I would build it, I would be like, "This just belongs in the trash. It belongs in the trash. All this work, all this amazing product belongs in the trash." And why? Because to me it was competitive, it wasn't differentiated.
Claire 说产品管理已死。她放了一堆吓人的圆圈图。她说什么发布量要冲上月球，干就完了，干干干。但我确实觉得，我们谁都没有——我知道这是我问过的问题：过去一年里，谁的公司交付得比以往任何时候都多？又有谁的公司营收是跟着 PR 数量同比例增长的？这才是我们眼下面对的根本问题。所以我要讲个故事，让大家切身体会一下这种「我没有好点子，也不知道做的东西有没有意义」的感觉。嗯，我当时有个点子，要给 ChatPRD 做一个叫「产品图谱」（product graph）的东西。这是一个再典型不过的「给产品经理做的产品」式点子，一点也不特别，因为人人都做过。大概就是：我们要把一家公司所知道的关于客户、关于你手头在做什么、等等等等的一切，全都吸进来。我们来解析它，再往上套一层 Opus，然后这一切都能通过 agent 调用，简直完美。它能让你做出更好的决策、做出更好的产品，当然，还能写出更好的 PRD。于是我就把它做出来了。我做了一个洞察引擎。我做了一个特别高级的语义产品图谱。它还能自动生成一套 wiki。agent 可以直接取用。几乎是一次成型（oneshot）。也不完全是，但已经很接近一次成型了。它能跑，看起来也相当不错。跟市面上的竞品功能逐项对齐。可每次做它的时候，我心里都在想：「这东西就该扔进垃圾桶。该扔进垃圾桶。所有这些心血、这么棒的产品，都该扔进垃圾桶。」为什么？因为在我看来，它只是有竞争力，却没有差异化。
[7:15] Claire
And so, you know, even though I had built this thing that like obviously had a market, people would probably want it, customers said it was interesting, like in my core, I felt like it wasn't worth my customers attention. I like the ROI was really shaky to me. I wasn't convinced on the interface. I was like, should this it it still had a web interface. I was like, should it be agent first? Is this the right thing? I wanted to build something surprising that competitors couldn't even imagine. And my big problem was that these tokens and this like road map that I had executed my bet, but it didn't really test or validate or strengthen my conviction. Meanwhile, I'm sitting on all this kind of like code to nowhere. I kept shipping more than ever. So, customer fixes were like automated. They would come in, they go to like one of my bots, and then they get fixed. Goals were like obliterating tech debt. I mean, I re rearchite rearchitected chat purity 70 times because I could. Um, we stopped tracking issues. We're like, we don't need issue tracking anymore. We're just going to ship PRs. This AI factory like manifested and started running itself. It did the thing that we all want it to do. It's magical. And here's the secret.
所以，虽然我做出来的这个东西显然有市场，大家大概也会想要，客户也说挺有意思，但在我内心深处，我觉得它不值得我的客户花注意力。在我看来，它的 ROI 非常站不住脚。我对它的界面形态也没把握。我在想，这东西该不该——它当时还是个网页界面。我就想，它是不是应该 agent 优先？这真的是对的方向吗？我想做的是让人惊喜、竞争对手连想都想不到的东西。我最大的问题在于：这些 token、我手上这份路线图，确实把我的赌注执行出来了，却没有真正检验、验证或者加强我的信念。与此同时，我手里攥着一大堆不知通向何处的代码。我的交付量还在不断创新高。客户问题的修复基本自动化了：问题一进来，就流转到我的某个 bot 那里，然后就被修好了。目标变成了把技术债彻底清零。我是说，我把 ChatPRD 的架构重——重构了 70 次，就因为我做得到。嗯，我们不再跟踪 issue 了。我们觉得，已经不需要 issue 跟踪了，直接发 PR 就行。这座 AI 工厂就这么成形了，而且开始自己运转。它做到了我们所有人都希望它做到的事。简直神奇。但这里有个秘密。
[8:35] Claire
I'm not sure any of it mattered. Like I'm not yet convinced that any of that mattered. And that really got me thinking. It got me thinking about what is product management? What is the purpose of building software? How do we make decisions? How do we prioritize? How do we allocate all these tokens? And I just kept going back to this idea of the road map. road maps made sense under engineering scarcity. They like laid out your strategy. They told you what features were you going to build. We had this thing called effort which was like more than me like voice noting you know Devon saying can you please build this and we thought a lot about sequencing and milestones and MVPs and scarcity helped us theoretically let's be honest it helped us theoretically filter out our bad ideas because the bottom of the list just never shipped. It never could. There were not enough engineers. There was not enough conviction across the organization. and said this like below the cut that's what we called it below the cut line never shipped and now we can ship all our bad ideas like congratulations to us cheap execution I think still needs judgment and I think we're seeing this moment where the road map plus like limitless or feeling like limitless execution capacity is creating three traps that I want to warn you about the first is the backlog track. So this is if you have a backlog, AI will build it.
我不确定这一切有任何意义。我到现在都还没法说服自己，这些东西有任何意义。这真的让我开始思考。它让我开始思考：产品管理到底是什么？做软件的目的是什么？我们怎么做决策？怎么排优先级？这么多 token 该怎么分配？我的思绪一次又一次回到路线图这个概念上。在工程产能稀缺的年代，路线图是说得通的。它铺陈出你的战略，告诉你要做哪些功能。我们那时有个东西叫「工作量」（effort），它可不只是我给 Devon 发条语音、说句「你能帮我做一下这个吗」那么简单。我们还花了大量心思琢磨先后顺序、里程碑和 MVP。而稀缺性在理论上——说实话，只是理论上——帮我们过滤掉了烂点子，因为清单底部的东西压根就没上线过。它也不可能上线。工程师不够，整个组织上下也没有足够的信念。然后我们会说，这个在砍线以下——我们当时就是这么叫的，「砍线以下」的东西从来没上线过。而现在，我们可以把所有烂点子都发布出去了，真是可喜可贺。我认为，廉价的执行依然需要判断力。我觉得我们正处在这样一个时刻：路线图，加上无限的——或者说感觉上无限的——执行产能，正在制造三个陷阱，我想提醒大家当心。第一个是 backlog 陷阱（backlog trap）。意思是，只要你有 backlog，AI 就会把它做出来。
[10:15] Claire
They will build every single thing. It will build every single thing on your list. And so I hear a lot about like, you know, I we're burned through our backlog. It's because all you have to do is slashgo build the backlog and it gets done. But the problem is clearing out requests or ideas does not mean meaningful progress against your business or even meaningful progress against a customer problem. The second thing is the parity trap. This is something I really felt in that story about the product graph. All your competitors are copying each other and they all are talking to the same customers and they're all sucking in the same insights and they're all arriving at the same obvious conclusion and building the same obvious product with the same obvious like run the taste skill. Run impeccable. Run whatever. Make it look good. Get rid of the m dashes. Get rid of the borders. Like we're all doing Can you tell I took max? Yeah. So we're all doing the exact same thing. And so this like parody trap where we used to have this like really interesting dynamic with competitors now it feels like it's evening out and again like this question about what Moes are and what differentiators are is a big one in my mind. And then the final kind of trap I see is the churn trap which is we ship something, we see noise and pick up or not because we can always ship something else, we abandon it and then we never really learn or compound on top of what we're shipping. And so all of these things from the inside feel very productive. Like my backlog's going down, my competitive features are going up. like we're shipping more than ever, but what I think this does is it just like accelerates our path to mid. And that's the thing I want to warn you about. And so this brings me to something that I'm really observing, which is this concept of road map zero.
它们会把每一样都做出来。它会把你清单上的每一样东西都做出来。所以我经常听到有人说：我们把 backlog 全刷完了。那是因为你只要敲一句 /goal「把 backlog 做完」，它就搞定了。但问题是，清空需求或点子，并不意味着你的业务有了实质进展，甚至也不意味着在客户的问题上有了实质进展。第二个是同质化陷阱（parity trap）。这是我在刚才产品图谱那个故事里切身感受到的。你的竞争对手们都在互相抄，都在跟同一批客户聊，都在吸收同样的洞察，都得出同样显而易见的结论，做出同样显而易见的产品，用的也是同样显而易见的——比如：跑一下 taste skill。跑一下 Impeccable。随便跑点什么。弄得好看点。把破折号（em dash）删掉。把边框去掉。我们全都在做——看得出来我是个 token 狂吧？没错。所以我们全都在做一模一样的事。于是就有了这个同质化陷阱：过去我们和竞争对手之间有一种非常有意思的动态，现在感觉正在被拉平。再说一次，护城河是什么、差异化因素是什么，这个问题在我心里分量很重。我看到的最后一种陷阱是弃置陷阱（churn trap）：我们发布了一个东西，看到一点动静、有点起色——或者没有——反正我们随时都能再发布别的，于是就把它丢了，结果我们从来没有真正从中学到东西，也没能在自己发布的东西之上持续积累复利。所以，从内部看，这些事情都让人觉得非常高产。比如我的 backlog 在减少，对标竞品的功能在增加，我们交付得比以往任何时候都多，但我认为，这一切其实只是在加速我们变平庸。这就是我想提醒大家当心的。这就引出了我真正在观察的一件事，也就是「路线图归零（roadmap zero）」这个概念。
[12:15] Claire
I don't mean like we've gone through the road map, although some people truly have road map zero, nothing on the road map anymore. What I mean is that every visible feature becomes both plausible and buildable. You don't have like an empty backlog. That's not what I'm talking about. It's just like everything on your backlog you can kind of do. And when everything on your backlog you can kind of do, what is the point of the backlog? I think buildability and effort stop being like this meaningful proxy for what matters and prioritization again kind of like why PMs exist as we've practiced it I think stops being strategy right like we've been doing these like rice prioritizations but when like many of those letters mean nothing anymore especially effort like why are we still pretending like prioritization is the right way to think about our products and this is why I think you know, road maps are over. They're dead. It's because I think they're really dangerous right now. I think an AI factory plus an old road map will get you to those three traps at machine speed. They will get you there fast because this factory really can't distinguish a truly consequential bet from an idea. And AI will make the consequences of this weak judgment show up at your front door faster. And so your bad ideas will become your problems quicker than ever. And so I don't think the road map is the right thing. I think the right thing is to ask yourself what do I believe strongly enough to go out and try and prove because the remaining constraint is not code. It's not building. It's not features. It is truth. It is like on the ground with our customers truth. AI cannot take an untested assumption into fact. No matter how many adversarial reviews you do, and I know you're doing them, you need real customers. You need real data. You need repeated tests. And then you need a very unique point of view and a very high unique quality bar.
我不是说我们已经把路线图上的事都做完了——虽然有些人是真的路线图归零了，路线图上已经什么都不剩。我的意思是：每一个看得见的功能，都变得既说得通、又做得出来。不是说你的 backlog 空了，我说的不是这个。而是 backlog 上的每一项，你差不多都能做。而当 backlog 上的每一项你都差不多能做的时候，backlog 还有什么意义？我认为，「做不做得出来」和「要花多少力气」已经不再是衡量「什么重要」的有效代理指标了；而优先级排序——还是那句话，这差不多就是 PM 存在的理由——按我们一直以来的做法，我觉得它已经不再是战略了，对吧？我们一直在做 RICE（优先级打分法）排序，可当其中好几个字母都已经没有意义了，尤其是 effort（工作量），我们为什么还要假装优先级排序是思考产品的正确方式？这就是为什么我认为，路线图已经完了，死了。因为我觉得它们现在真的很危险。我认为，AI 工厂加上一份老式路线图，会以机器的速度把你送进那三个陷阱。而且会很快，因为这个工厂根本分不清一个真正举足轻重的赌注和一个单纯的点子。AI 还会让这种薄弱判断的后果更快找上门来。所以你的烂点子会比以往任何时候都更快地变成你的麻烦。所以我不认为路线图是对的东西。我认为对的做法是问自己：我对什么东西的相信强烈到足以让我走出去、试着证明它？因为剩下的约束不是代码，不是构建，不是功能，而是真相——是在一线、在我们客户那里得来的真相。AI 没法把一个未经检验的假设变成事实。不管你做多少轮对抗式评审——我知道你们在做——你都需要真实的客户，需要真实的数据，需要反复测试。然后，你还需要一个非常独特的观点，以及一条非常高、非常独特的质量标准。
[14:21] Claire
And what AI allows us to do is make faster contact with reality, which is great. We all want to make faster contact with reality. But that means you have a higher obligation to encounter reality in your minds. You have to both get into the market and accept what it is telling you. And this is going to feel really retro to people like, you know, me and they're a little older because we've been saying outcomes over outputs all the time, but yet this like central artifact, the road map still lists features and dates. And again, engineering scarcity once made that workable. But like I think road map zero makes those features and dates very dangerous for the reasons that I outlined. This feature list I say becomes ammunition for a very powerful slob cannon. And so what I think we need to do is move upstream from what we need to build to what we actually need to prove. And so here's how I would think about what we do instead of a road map. First we build our convictions. So what is the direction that we want to go and what do we believe the future looks like? Not in 3 months, 6 months, 9 months, in in a year, in two years. Where do we think all of this is going? I do think you need to determine what evidence would show you that your conviction is true. And you need to define that upfront. What would I need to see to prove that I'm doing the right thing? And what would I need to see to stop? You do need a factory. So I do love the factory. I'm saying that's dangerous, but I love something dangerous. You need your factory. You need the ability to build quickly to intersect reality very very fast. So I'm not saying get rid of it. I'm saying put it at the right place in the process.
AI 让我们能更快地接触现实，这很棒。我们都想更快地接触现实。但这也意味着，你更有义务在心里真正直面现实。你既要走进市场，也要接受市场告诉你的东西。对我这种年纪稍大一点的人来说，这听起来会很复古，因为我们一直把「重成果，轻产出（outcomes over outputs）」挂在嘴边，可偏偏这个核心产物——路线图——列的依然是功能和日期。再说一遍，工程资源的稀缺曾经让这套做法行得通。但我认为，路线图归零让这些功能和日期变得非常危险，原因我刚才已经讲过了。我说，这份功能清单会变成一门威力巨大的「slop 大炮」（slop cannon，批量喷 AI 垃圾的机器）的弹药。所以我认为，我们需要往上游走：从「我们需要做什么」，转向「我们真正需要证明什么」。那不做路线图的话，我们该做什么？我是这么想的。第一，建立我们的信念。我们想往哪个方向走？我们相信未来是什么样子？不是 3 个月、6 个月、9 个月之后，而是一年、两年之后。我们认为这一切会走向哪里？我确实认为，你需要确定什么样的证据能表明你的信念是对的，而且要提前定义好：我需要看到什么，才能证明我在做对的事？我又需要看到什么，就该停下来？你确实需要一个工厂。我是真的爱这个工厂。我说它危险，但我就是爱危险的东西。你需要你的工厂，你需要快速构建的能力，好非常非常快地撞上现实。所以我不是说把它扔掉，我是说把它放到流程里正确的位置上。
[16:05] Claire
And then you need to allocate. You need to go through that cycle. And when you get into reality, you need to allocate your investment, your conviction, your effort, your tokens in the right direction. And what's interesting I think most about this new world we're about to move into is I think you need to have durable convictions but disposable features. And I was talking to Mara who is going to come on the stage later and she's like but Claire enterprises love feature road maps and I'm going to talk to some folks at the end of the day who have changed a lot of features around on us a lot of products around us and we as consumers at least some of us have accepted that level of churn. And so I think we're going to get into this place where customers, where teams are able to shift to, do I bet on these convictions? Do I bet on this team? Do I bet on this space? Do I bet on this vision? And I understand features need to come and go. They might be unique to me. They might be unique to a market. And so what you want to see is clear progress against your convictions, but zero ego about your solutions. And what this would look like, I think there's two ways this could go. You could be you have to be stubborn here with your durable convictions. But there's good stubborn and bad stubborn.
然后你需要分配。你需要走完这个循环。当你进入现实之后，你需要把你的投入、你的信念、你的精力、你的 token 分配到正确的方向上。在我们即将迈入的这个新世界里，我觉得最有意思的一点是：你需要持久的信念，可抛弃的功能。我之前跟 Mara 聊过——她待会儿也会上台——她说：「可是 Claire，企业客户就是爱功能路线图啊。」今天收尾的时候，我还会跟几位嘉宾聊，他们在我们眼皮底下把一大堆功能、一大堆产品改来改去，而我们作为消费者——至少我们当中的一些人——已经接受了这种程度的频繁变动。所以我觉得我们会走到这样一个地方：客户——或者说团队——能够转而这样思考：我要不要押注这些信念？要不要押注这个团队？要不要押注这个领域？要不要押注这个愿景？我也理解，功能就是会来来去去。它们可能是专门给我做的，也可能是专门给某个市场做的。所以你想看到的，是在信念上有清晰的进展，同时对具体方案零执念。那这会是什么样子？我觉得可能有两种走向。你可以——你必须在你持久的信念上保持固执。但固执分好的固执和坏的固执。
[17:18] Claire
Good stubborn is you like stay with the problem. You revise the solution. You know, you hold true what's going to happen even if the flavor of the week is trending. Bad stubborn is because you have tokens. You just move the goalpost. You're like, "Ah, we're close. We'll just ship again and ship again and ship again." And so I want you to think about what are my durable convictions? Am I being good stubborn? Am I being bad stubborn? And then how do you actually build into your system the ability to absorb disposable features and keeping rigor around quality and learning? And this is really scary, especially when it hits customers. But I think we're moving to a place where not everything we ship is a promise. And I I don't has anybody feeling come on. Like has anybody shipped some thank you right in the middle? Like you ship things and you're like this is a hypo this is like truly an experiment and I might get it wrong or the market might move underneath me or the technology might change, you know, tremendously. And so I do think we have to think about this idea of like not every ship is a process, a promise. And we used to do these road maps and I would like go hand on heart to go to market and I'd be like I swear on my children this feature will ship with these specific things on this specific date. You can definitely sell it in the contract like you know and make these promises. And now I think we have to really think how we communicate internally, how we communicate with customers. And so I think there's going to be everything from a probe. This is like kind of a bet. We're going to explore a little bit. Um to a more durable kind of experiment against your conviction. I think this is something we really believe in. We're going to stick with it till we get it right. And then there are promises. Promises are I have shipped something. Customers can rely on it. They can build their customers on
好的固执是：你守着问题不放，修改的是方案。哪怕眼下流行的是另一阵风，你依然坚信将来会发生的事。坏的固执是：因为你有 token，你就不停地挪球门。你会说：「啊，快成了。我们再发一版，再发一版，再发一版。」所以我希望你们想一想：我有哪些持久的信念？我是好的固执，还是坏的固执？然后，你要怎样在自己的体系里真正建立起一种能力，既能消化可抛弃的功能，又能在质量和学习上保持严谨？这真的很吓人，尤其是当它落到客户身上的时候。但我认为我们正在走向这样一个地方：不是我们上线的每样东西都是承诺。我……我不……有没有人有这种感觉？来嘛。有没有人上线过一些——谢谢，中间那位——就是你上线了一些东西，心里想着：这是个假……这真的就是一个实验，我可能会搞错，市场可能在我脚下变了，技术也可能发生翻天覆地的变化。所以我确实认为，我们得想想这样一个理念：不是每一次上线都是一个……一个承诺。以前我们做路线图，我会手按胸口去跟市场销售团队说：「我拿我的孩子发誓，这个功能一定会在这个具体日期、带着这些具体内容上线。你们绝对可以把它写进合同里去卖，放心去做这些承诺。」而现在我觉得，我们得认真想想对内怎么沟通、对客户怎么沟通。所以我觉得会有各种层级：从探针——这算是一种下注，我们稍微去探索一下——嗯，到围绕你的信念、更持久一些的实验——我觉得这是我们真心相信的东西，我们会一直坚持，直到把它做对。然后是承诺。承诺就是：我上线了一个东西，客户可以依赖它，他们可以把自己的客户也搭建在
[19:05] Claire
it. We really believe and we'll keep compounding here. And I think you need to be honest about what your commitment is on any feature. I think that's almost the most um helpful lens on the new road map is like how strong is our conviction here and how durable is this promise versus like how hard is it to build and what is our estimated impact and while we do this I you know again you need to remember code is abundant customer trust is not. So going back to that product graph, that feature that I built over and over and over again until I felt conviction, it was feature flagged off because I just knew that as soon as I put this in front of a customer, they were going to build on it. They were going to use it. And if I didn't have conviction that it was right, I was really going to burn that customer trust. And that's something I thought about really deeply as I was rolling out this feature. So I do still think you need a road map, but I think that roadmap needs to be more about ambition. I think it needs to be bigger. I don't want to see onesie twoosies on your road map. I want to see the future you believe should exist. I want to see what evidence shows that we're making progress. What would prove us wrong? And then what earns more time tokens and customer attention. And I want to go back to ambition because I think the last 12 to 18 months as you know it was like you know the year of our claude and we were shipping everything and PMs were writing PRs and it was prototypes everywhere. I really think we're in a velocity game. We're in a inflect PRs straight up velocity game. If I can get prototypes to customers faster they can give me feedback faster. If I can hand things to customers or to engineering faster they can do PRs faster. my agents will take care of tech debt. My agents will take care of features. I think we're really in the last 12 to 18 months building up our muscle for true
这上面。我们真心相信它，也会在这里持续复利。我认为，对任何一个功能，你都要诚实面对自己的承诺程度到底有多深。我觉得这几乎是看待新路线图最有用的一个视角：我们在这里的信念有多强、这个承诺有多持久，而不是它有多难做、我们预估的影响有多大。在这么做的同时，你还是要记住：代码是充裕的，客户信任可不是。所以回到那个产品图谱——那个我一遍又一遍、反反复复重做、直到有了信念的功能——它一直被功能开关关着，因为我很清楚，一旦把它摆到客户面前，他们就会在上面搭东西，就会用起来。如果我对它的正确性没有信念，我就真的会把客户的信任烧掉。这是我在推出这个功能时想得非常深的一件事。所以我依然认为你需要一份路线图，但这份路线图要更多地关乎野心。它得更大。我不想在你的路线图上看到零敲碎打的小东西。我想看到你相信应该存在的那个未来。我想看到什么样的证据表明我们在进步，什么能证明我们错了，以及什么配得上更多的时间、token 和客户注意力。我想再回到野心这个话题，因为过去 12 到 18 个月，大家都知道，简直就是「吾主 Claude 纪年」，我们什么都在上线，PM 在写 PR，到处都是原型。我真觉得我们处在一场速度游戏里。一场 PR 数拐点式飙升、赤裸裸拼速度的游戏。如果我能更快把原型交到客户手上，他们就能更快给我反馈。如果我能更快把东西交给客户或工程团队，他们就能更快地提 PR。我的 agent 会搞定技术债，我的 agent 会搞定功能。我觉得过去这 12 到 18 个月，我们其实一直在为真正的
[20:55] Claire
inflected velocity. I don't think that's the game next year. I don't think that's the game we should be playing for next year. I think next year is the ambition game. I think you should think like what huge swings can I make? What experiments can I run in two weeks, three weeks that would have taken us a year last time? and how do I build sort of this roadmap of very very very large investments in very very ambitious builds because I think that is more important than raw feature velocity and I don't think the road map as it is serves that goal you know I was talking to somebody and again going back to OKRs unfortunately not dead yet and she was like what should our OKRs be relative to our AI transformation like PRs revenue per headcount like what it should be like how many [snorts] huge experiments are you running a month like I don't care what the experiments are I don't know what they should be like how many big big swings are you taking a month with the presumption that most of them won't work out and I think that is just a very new way to think about how you build your road map and a very new way to think about how you do your product again it's big on the ambition and very fuzzy on the specifics So here is my ask slashgift to you all.
拐点式速度练肌肉。我不觉得那是明年的游戏。我不觉得那是我们明年该玩的游戏。我觉得明年是野心的游戏。你应该去想：我能搞哪些超大胆的大动作？有哪些实验，我能在两三周内跑完，而放在以前得花我们一年？我要怎么搭出这样一份路线图：上面是非常非常非常大的投入，投向非常非常有野心的构建？因为我认为这比单纯的功能交付速度更重要，而现有这种路线图并不服务于这个目标。我之前跟一个人聊——又说回 OKRs，很遗憾它们还没死——她问：围绕我们的 AI 转型，OKRs 应该定成什么？PR 数？人均营收？我觉得应该看的是：你每个月在跑多少个巨型实验。我不在乎那些实验具体是什么，我也不知道它们该是什么，关键是你每个月搞了多少次真正的大动作，而且预设其中大部分都不会成功。我觉得这是思考怎么做路线图的一种全新方式，也是思考怎么做产品的一种全新方式——还是那句话：野心要大，细节可以很模糊。所以，这是我给大家的一个请求，或者说一份礼物。
[22:21] Claire
Build your last road mapap. And I do not mean build your last plan. Of course not. I just mean like no more lists of ideas and features in spreadsheets where you guess impact and where you like put a date on them and then lock arms and say we'll never we'll never change anything because this is how we work. I just think that day is over. Instead, what I want you to do is truly raise your conviction, raise your ambition, come up with big ideas, define what success looks like in like a year or two. You're going to have to make some big guesses because I can't honestly tell what's going to happen in a week or two. I really think you want to get to a place where you have a factory that can surprise you. And so, you really want to be able to discard software and then hold your factory to a high bar. And look, you're going to write a lot of code and then you're going to trash it. Good. Your customers do not need trash products. Your AI is going to have better ideas than your team. Good. We need good ideas. And you need to hold that AI to a very, very high bar. And you're not going to be able to promise your team, your customers, what next year looks like. Good. Because it's probably going to be better than you can imagine right now. So, this is my ask to you. write your last road map I promise you is extremely fun on this other side.
做出你的最后一份路线图。我不是说做你的最后一份计划，当然不是。我的意思只是：别再在电子表格里列那些想法和功能清单了——在里面拍脑袋估影响，给每一项标上日期，然后大家手挽着手说：我们绝不、绝不改任何东西，因为我们就是这么干活的。我只是觉得，那样的日子已经结束了。相反，我希望你们真正提升自己的信念，提升自己的野心，想出大点子，定义一两年后的成功是什么样子。你们得做一些大胆的猜测，因为说实话，我连一两周后会发生什么都说不准。我真心觉得，你们要走到这样一个地步：拥有一个能给你惊喜的工厂。所以，你要真的能够丢弃软件，同时对你的工厂保持高标准。听着，你会写大量代码，然后把它们扔进垃圾桶。很好。你的客户不需要垃圾产品。你的 AI 会比你的团队有更好的点子。很好。我们需要好点子。而你需要对那个 AI 提出非常、非常高的标准。你也没法向你的团队、你的客户承诺明年会是什么样子。很好。因为它很可能会比你现在能想象的还要好。所以，这就是我对你们的请求：写下你的最后一份路线图。我向你们保证，到了另一边，真的超级好玩。
[23:47] Claire
Enjoy the rest of your summit. I will see you later this afternoon and please come say hi. [applause]
祝大家享受接下来的峰会。今天下午晚些时候见，欢迎来跟我打个招呼。（掌声）