虚假 PMF 的幻觉与 Serval 的爆发式增长
我们的产品曾有一批狂热的粉丝，他们非常喜欢我们的产品。这让我产生了一种错觉，以为我们已经找到了产品市场匹配（PMF），或者马上就能迎来爆发式的大规模增长。判断 PMF 不应该只凭一次或者三次愉快的客户沟通。你必须进行足够多的交流，直到你真正觉得自己深刻理解了这个问题。对于自己是否找到了产品市场匹配，你应该保持极其严苛的判断。而且，必须出于对他们的真诚兴趣、出于想要改善他们生活的愿望，才能推动这种关系向前发展，并为你提供打造更好产品所需的洞察。
我是 Jake，Serval 的联合创始人兼 CEO。Serval 是一个面向 IT 团队的 AI 原生平台。我们帮助自动化处理服务台请求、员工入职、离职、即时访问权限申请——基本上就是 IT 必须处理的大量长尾事项。我们最近刚宣布了由红杉资本领投的 7500 万美元 B 轮融资，估值达到 10 亿美元。从公司创立到拿到 10 亿美元估值的投资意向书（term sheet），一共历时 18 个月。
英文原文
We had rabid fans of our product that loved what we were doing. And that gave me this delusion that we had found product market fit or we were just around the corner from really unlocking massive growth. It shouldn't be one good customer conversation, three good customer conversations. You have to have enough where you really feel like you understand the problem deeply. You should be really ruthless around your judgment on whether you found product market fit. And it has to be in a genuine interest in them and in making their lives better that drives that relationship forward and gives you the insights you need to build a better product. I'm Jake. I'm the co-founder and CEO of Serville. Servil is an AI native platform for IT teams. We help automate help desk requests, onboarding, offboarding, just in time access requests. Basically, the long tale of things that it has to do. And we just recently announced our $75 million series B led by Sequoia at a billion dollar valuation. From founding to a term sheet valuing us at a billion dollars was 18 months.从脑科学到早期创业与 Neuroplus
我不觉得自己算得上是个天才，不过我在高中时 SAT 考得确实挺不错。我觉得这对我产生的影响主要是给了我很大的信心。那种信心确实很有帮助，让我知道自己能把事情弄明白、能学会新东西，而且我能有竞争力。我没有什么需要去证明的，我只是想去学习。
我当时在杜克大学的一个神经科学实验室工作。但小的时候，我觉得自己就是单纯非常好奇。我非常喜欢学习，能对各种话题迅速产生兴趣，一头扎进去学很多东西；然后就觉得无聊了，转向下一个话题，并没有把任何一件事情从头到尾坚持下去，因为我总是在不同的兴趣之间跳来跳去。
我一直都想创办一家公司。我母亲是一位创业者，我也一直畅想自己去创业。我十几岁的时候就尝试过做生意，我觉得那终究是我的目标。甚至在大学期间，我也加入了许多创业社团，最后还搬进了一个创业团体。因此，虽然我当时学的是神经科学，但我对创业和成立公司非常感兴趣。
我从大学退学创办了我的第一家公司，那是一家通过脑部扫描来测试广告效果的公司。当时就想着，先休学几个月，要是做不起来就回学校继续读书。但那第一个休学的学期变成了第二个学期，又变成了第三个学期，最终很明显我不可能再回去了。事情进展得还算顺利，在这个过程中发生了一件事：我们的一位客户说：“嘿，我能把那个头戴设备带回家吗？我能把它戴在我儿子头上，看看他做作业时什么时候专注、什么时候走神吗？他患有 ADHD，我只是想帮他获得反馈。”
那个故事深深地触动了我。我们决定做一个尝试性的项目，想着也许我们可以为患有 ADHD 的儿童开发一款电子游戏。最终我们对公司进行了转型，为患有 ADHD 的儿童打造一款消费级产品，研发了新的硬件和新的头戴设备，那家公司后来成为了 Neuroplus。
英文原文
I don't know that I would describe myself as a genius, but yeah, I did get a a good SAT score in high school. I think the way I would think it it affected me is it just gave me a lot of confidence. I think that that confidence is really helpful just knowing that I can figure things out. I can learn things. I can I can be competitive. Uh I I don't have anything to prove. I'm just trying to to learn. I was working in a neuroscience lab at Duke. But as a kid, I think I was just really curious. I really like to learn. I really got interested in topics very quickly. dove in, learned a lot, and then got bored and moved on to the next topic and not really sticking with anything all the way through because I was just kind of hopping between interests. And so I always wanted to start a company. My mom was an entrepreneur and I always envisioned myself I tried to start a business when I was a teenager and I think that was ultimately the goal. And even when I was in college, I joined a bunch of entrepreneurship groups. I I ended up moving into an entrepreneurship group. So even though I was studying neuroscience, I I was very interested in in startups and starting companies. I dropped out of college to start my first company at a company testing advertising with brain scans. At the time, just like I'll take a few months off and then I'll just come back to school if this doesn't take off. And then that first semester off became a second semester off became a third semester off and then eventually became clear that I was never going back. And things went well enough. And something happened during that process where one of our customers said, "Hey, can I take that headset home? Can I put that on my son and see when he's paying attention to his homework and when he's not? He has ADHD and I just want to help him give get feedback." And that story really stuck with me and we decided to build kind of a pet project of hey, maybe we could build a video game for kids with ADHD. And we ended up pivoting the company to build out a consumer product for kids with ADHD, building new hardware, a new headset, and that became Neuroplus.早期用户的误导与小众市场的假象
我经营了那家公司大约七年。我们曾拥有粉丝，拥有狂热的产品粉丝，他们热爱我们所做的事情。这让我产生了一种错觉，以为我们找到了产品市场匹配，或者马上就要迎来爆发式增长。
所以当时入选福布斯 30 Under 30 时，我并不觉得公司在分崩离析。但事后回顾，显然我们根本谈不上有什么产品市场匹配。或者说，也许我们确实有某种 PMF，但那对应的市场规模太小了，以至于几乎无关紧要、毫无分量。我们总觉得只要再迈出一小步就能释放出巨大的增长。在那段时间里，数据在往上走，感觉很不错，感觉事情行得通。我们刚完成了一次成功的 Kickstarter 众筹，很多人在买我们的产品。我总觉得只要再做成一件事，就能开启下一个阶段的增长。我们或许能以某种形式维持很长时间，但它绝不会成为我所想象和憧憬的那种公司。
在关闭 Neuroplus 之后，我大概花了一两年的时间才真正能够回过头来看清，我们其实从未真正接近过 PMF。
判断 PMF 不应该只凭一次或者三次愉快的客户沟通。你必须进行足够多的交流，直到你真正觉得自己深刻理解了这个问题，并且理解了你的解决方案将如何真正为市场上足够大的一部分群体解决这个问题。我认为对于初次创业者来说，在判断自己是否找到产品市场匹配时，应该极其严苛。
显然，你总是要从早期采用者（early adopters）切入，然后逐步扩展到更大的大众市场。但我认为你必须对自己诚实：你究竟是从早期采用者起步且未来确实能够打入其余市场，还是说你的整个市场就仅仅只有这么一小群狂热的早期采用者？要弄清楚这一点可能很难。
我认为我们早期发现的一些信号是：我们的客户身上有一些非常独特和与众不同的特质，看起来和市场的其余部分完全不同。我认为这就是一个信号，说明你所处的市场可能并没有你想象的那么大——如果你的早期客户都有某种看起来非常非常奇怪的共同点，而你又看不到从这批客户自然过渡、跃迁到其余市场的平滑路径。
英文原文
ran that company for about seven years. We had fans, we had rabid fans of our product that loved what we were doing. And that gave me this delusion that we had found product market fit or we were just around the corner from really unlocking massive growth. But so at the time when I got Forbes 730, I didn't think of the company as falling apart. In retrospect, obviously we were not anywhere near product market fit. So maybe we did have some product market fit, but it was with a market that was so small that it it was pretty irrelevant and inconsequential. We're just a little bit away from unlocking some massive growth. But during that time, it's it's ticking up. It feels good. It feels like it's working. We just done done a successful Kickstarter campaign. A lot of people are buying our product. There's always one more thing that I felt like was going to unlock the next stage of growth. And we could probably keep it going in some form for a very long time, but it was not going to be the company that I imagined and I envisioned. So, probably took a year or two after winding down Neuroplus for me to really be able to look back and realize that we were never that close. It shouldn't be one good customer conversation, three good customer conversations. You have to have enough where you really feel like you understand the problem deeply and you understand how your solution is really going to solve it for a big enough part of the market. I think for first-time founders, you should be really ruthless around your judgment on whether you found product market fit. And so, obviously, you're always starting with early adopters and you're always having to like grow into the larger mass market. But I think you have to be honest with yourself about, you know, are you starting with early adopters and you actually are going to be able to get to the rest of the market or is your entire market just a kind of a crazy early adopter set and and that could be a hard thing to figure out. I think some of the signals that we had early on was that there there were things that were very unique and different about our customers that didn't look like the rest of the market. And I think that that is one signal that maybe you're not playing in as big of a market as you think is if your your early customers all have something in common that seems very very strange and you don't see a a natural gradient towards the rest of the market from jumping from those customers to the rest.在 Verkada 目睹真正的市场拉动
这让人感觉事情进展顺利，尤其是如果你满怀乐观的话——所有创业者都会这样，特别是早期的、职业生涯刚起步的创业者。面对极佳的客户反馈，你很难保持怀疑和审慎去说“我觉得这行不通，我们应该换个方向”。你必须理解你的解决方案到底要如何为市场上足够大的群体去解决问题。
后来，打造软硬件平台的经历让我加入了 Verkada，之后才离开去创办 Serval。
我觉得对我影响最深的一点就是：一旦你拥有真正强劲的市场拉动（market pull）和产品市场匹配，其余的事情自然就会迎刃而解。我很幸运有一位朋友，他是我读大学时的同学，也是 Neuroplus 的早期顾问。他作为早期员工加入了 Verkada 负责市场营销，起初他给我妻子打电话说：“嘿，我们很缺人，这里太疯狂了。”我还记得他的原话是：“你知道在初创公司里所有事情都非常艰难、没一件能成吧？”后来我飞到现场，走进办公室，感受到了那种能量，意识到这就是我必须待的地方。如果我未来想成为一名成功的创业者，我必须去亲身体会这是一种怎样的感觉、看起来是什么样。
只有在事后回顾加入 Verkada 并亲眼见识到真正的产品市场匹配时，你才会发现两者的差异有多大。当时我参加了第一次客户电话会议，一名销售正在演示产品。我觉得那名销售做得并不好，出了很多差错，在演示中磕磕绊绊，还搞错了好几处地方。客户也有点困惑，我心里直犯嘀咕：“天呐，这下砸了。”结果在通话结束时，客户却说：“好的，听起来不错。能给我发个报价单吗？我想我们下个月大概会买 30 台左右的摄像头。”
英文原文
It feels like it's working especially if you have a lot of optimism which all founders do especially early founders early career founders. It's hard to look at great customer feedback and have the skepticism and the cynicism to say I don't think this is working. We should do something else. And you understand how your solution is really going to solve it for a big enough part of the market. And then that experience of building hardware and software platforms is what led me to Vicata before leaving to start serving. I think the thing that sticks with me the most is this idea of you have really strong market pull and product market fit. The rest kind of falls into place. I was lucky to have a friend an early adviser to Neuroplus that I had gone to college with. He had joined Verata as an early employee running marketing and he initially called my wife and said uh hey we need people uh this is crazy. And I remember his words were you know how everything at a startup is really hard and nothing works. and eventually flew out on site, walked into the office, felt the energy and realized this is where I have to be. I have to see what this feels like, what this looks like if I'm going to be a successful entrepreneur in the future. It's only in retrospect when I joined Vicata and I saw real product market fit that you see how different it is. So, when I was on first call with a customer, a sales rep was demoing the product and I didn't think the sales rep did a very good job and they messed up a lot. They kind of fumbled their way through the demo. They got a couple things wrong. And the customer was a little bit confused and I just thought, "Oh man, this is not going well." And then at the end of the call, the customer says, "Okay, sounds good. Could you send me a quote? I think we'll probably buy 30 cameras or so in the next month."Serval 的平台构想与漫长的验证期
对我们来说，最大的问号从来不是这个方向对不对。我认为这是非常直观且显而易见的：只要你做得出来、只要这项技术存在，这就是唯一正确的路径。如果你把这一部分做对了，在业务的其他各个方面你并不需要做到完美无瑕。
所以从一开始，对我们来说最重要的就是聚焦于产品体验。专注于确保你所打造的东西是客户想要的、真正让客户喜爱的，其余的事情自然就会顺理成章。如果你没有产品市场匹配，你就无法修补组织的其余部分。
我的打比方是：这就像书本末尾附有答案一样。虽然过程极其艰难，你必须不断攻克问题，但你知道你试图解决的所有问题都有一个确切的答案。这就是我所应对的事情。如果你能解决它，那将极具意义。Serval 就是这样诞生的。
在创办 Serval 整整一年之后，我才真正相信我们接近了产品市场匹配，因为我对每一次沟通都抱有极大的怀疑。
我们是一个面向员工支持的 AI 平台。当员工去提出各种诉求时，比如需要某个应用的访问权限、需要重置密码、请假或者仅仅是想获取 Wi-Fi 密码，Serval 会自动给出这些问题的解答。
我们曾觉得自己的方向是对的，但每次把谈话转向“要达到什么条件他们才会购买”时，就会出现长长的一串功能清单，都是平台目前做不到但他们又非常需要的。
这里正是极其棘手的地方：因为就像我在 Neuroplus 的经历一样，你很容易陷入一种幻觉，总觉得自己离产品市场匹配“只差一个功能”。你会产生错觉，觉得“我只要一直做、一直做、一直做，总有一天能找到它”。然而事实几乎绝非如此。
但在我们的案例中，我们当时有一种理论：我们想打造的产品必须是一个平台。它必须将完整的 ITSM、AI 原生工作流构建器、权限访问管理以及服务台自动化整合在一起；只有这样一个完整的平台，才能在市场上取得成功，而在把平台完全搭建出来之前，我们其实并没有真正可供测试的产品。
因此，在那整整一年的时间里，我们一边埋头开发，一边几乎没有得到任何市场动力，支撑我们信念的正是这一点：产品还没有做完，我们必须把它先做出来，到那时我们才会见分晓。
英文原文
The biggest question mark for us was not whether or not this was the right approach. I think it's intuitive and obvious that if you could do it, if the technology existed, that would just be the right approach. And you don't have to be perfect at every other aspect of the business if you get that part right. And so it's always been important for us from the very beginning is focus on the product experience. Focus on making sure you're building something customers want and really customers love and then the other things kind of fall in line. So if you don't have product market fit, you can't fix the rest of the organization. It means that there's an answer to the problem in the back of the book is kind of how I liken it is so hard and you have to work through problems but you know that there's an answer to all the problems that you're trying to solve. That's what I deal with. And if you could solve that, that would be meaningful. And that was really how Serbal was born. It was a year into servo before I really believed that we were anywhere close to product market fit because I was so skeptical of every conversation I had. We are an AI platform for employee support. So when employees go and ask questions like they need access to an application, they need a password reset, they need to request time off or just get the Wi-Fi information, Serville automates the answers to those questions. We felt like we were on the right direction, but every time we kind of turned the conversation into, you know, what it would take for them to buy, there's just such a long list of things that the platform did not do that were really required of it. And here's where it's really tricky because, similar to my experiences at Neurop Plus, you can delude yourself into thinking you're just one feature away from product market fit all the time. And you can delude yourself into thinking like, I just got to keep building, keep building, keep building, and then one day I'll find it. And that's almost never the case. Now in our case we we had a theory that the product that we wanted to build had to be a platform. It had to combine the full ITSM this AI native workflow builder access management help desk automation that it was actually the full platform that was what was going to be successful in the market and we wouldn't really have a product to test until we built the platform. And so that's how we kind of kept the conviction during that one-year period where we're just building and we're not getting any kind of market traction is that well the the product is not built yet and we have to just build it and then we'll know转折点：对话基调的质变与客户发现
最终，我们签下了第一单，接着签了第二单、第三单、第四单。这就形成了一种级联效应：所有这些客户都在提供反馈，产品越做越好，我们的迭代速度越来越快，产品进而变得更加出色，我们签下了更多订单。每一次对话不再以“哦，这挺有意思的，有新进展随时联系”收场，而是变成了“这个要多少钱？我们怎么试用？你们的 POC 流程是怎样的？”
于是在 2025 年 4 月或 5 月左右的大约几周时间里，对话的基调几乎是一夜之间发生了转变——从“这东西很酷，有进展保持联系”变成了“这个多少钱？我们要怎么开始？”
因为在早期的对话中，我们会捕捉到一些客户对我们想做的东西感兴趣的有价值信息，但我们并没有被这些细节牵着鼻子走。你必须对客户的反馈保持非常中立的态度。
而在我职业生涯的早期，面对每一次客户沟通，你就像一个摆锤。遇到一次顺利的通话，你就欣喜若狂，觉得他们太兴奋了、大家太喜欢这个点子了；遇到一次糟糕的通话，你就垂头丧气。但在这次创业中进行客户发现时，我们接听了大量的电话，绝不让任何单一的对话对我们产生过度的影响。
我尽量每天花五到六个小时泡在客户电话上，因为这不仅是一个 IT 服务台自动化平台，也是一个面向整个组织的广泛企业级自动化平台。
很多人把这些客户沟通当作特定时点上的访谈，当作一次性的独立事件：我要做一次客户访谈，我要做一次客户沟通。而我更倾向于将其视为一段随着时间推移建立和维系的关系。我想深入融入到客户中去。所以我不会去想什么“客户访谈”或“客户沟通”，我只是无时无刻不在跟客户交谈。我挂在他们的 Slack 频道里，每天都和他们对话。
这种方式建立起对客户的直觉，远比那些“嘿，我列了一堆问题想请你回答”的时点式访谈要强大得多。客户访谈固然有价值，但我认为更有价值的是真正去理解客户，理解他们的问题、痛点、视角以及他们是如何使用你的产品的。这种理解是无法速成的，通过正式的访谈也无法建立。你只有真正融入他们的日常工作中，随着时间的推移与他们建立关系，才能做到这一点。必须出于对他们的真诚兴趣、出于想要改善他们生活的愿望，才能推动这种关系向前发展，并为你提供打造更好产品所需的洞察。
英文原文
eventually we close our first deal and then we close our second and we close our third and our fourth and then that just kind of creates this cascade and all these customers that are providing feedback and the product's getting better and we're iterating faster and then the product gets better and we close more deals and um every conversation ends up becoming not oh this is really interesting keep me posted but how much does it cost how do we try this out what's your PC process and so the the tenor of the conversation changed almost overnight over a few week period in April or May of 2025 where it went from this is really neat keep us posted to how much does it cost how do we get started because in our early conversations we would get nuggets that were interesting of customer interest in what we were trying to build and we just didn't let that kind of like push or pull us. So, you become very neutral to the customer feedback. Whereas, I think in the early part of my career, every customer conversation, you know, you just you're a pendulum. And so, you have a good call and you're you're you're on the moon. They're so excited. Like, people love this idea. You have a bad call, you're just in the dumps. And in this company, when we were doing customer discovery, we just kind of took a lot of calls. We did not let any single conversation influence us too much. So I try to be on customer calls five or six hours every single day because it's not just an IT help desk automation platform. It is it is a broad enterprise automation platform for the organization. So I think uh a lot of folks approach these customer conversations as these point in time interviews, these kind of discreet moments where I'm going to do a customer interview, I'm going to do a customer conversation and I approach it much more as a relationship that is built and fostered over time. I want to be embedded with customers. So, I don't think about customer interviews or customer conversations. Like, I'm just talking to customers all the time. I'm in their Slack channels. I talk to them every single day. That builds just an intuition for our customers that is much much more powerful than these kind of point in time, hey, I've got a list of questions I want you to answer. You know, customer interviews are valuable, but I think what's more valuable is actually understanding the customer, understanding their problems, their pains, their perspective, how they use your product. And you can't build that understanding quickly. You can't build it through formal interviews. You can only build it by truly being embedded in their day-to-day and and building relationships with them over time. It has to be an a genuine interest in them and in making their lives better that drives that relationship forward and gives you the insights you need to build a better product.应对 AI 模型的演进与产品边界
我认为如今在打造一款 AI 产品时，最需要时刻思考且最具挑战性的事情之一，就是模型未来会发生什么，以及这些能力随着时间推移将如何演进。
说实话，在最早期的时候它并没有起作用。当时的工作流构建器并不能让你随意描述任何自动化流程就能毫无人工干预地自动构建出来。它需要不断迭代，需要提示工程，需要大量的幕后操纵来让这一切运转并变得无缝衔接。
我认为我们所下的赌注在于：依托我们在改进产品方面的努力以及底层所做的很多巧妙设计，再结合底层基础模型本身的升级，我们将能够打造出一款真正实现这种自动化愿景的产品。
因此，我非常看好未来的软件公司在业务范围上会变得宽广得多；但我也认为这样的公司数量会更多，因为大家确实希望从这些平台中获得更多的能力。所以我不认为必然会出现行业整合。但我确实认为，我们对所有这些工具的期望都会大幅提高。
你究竟是在针对模型当下的工作表现来进行开发，还是面向它们未来稍好一点的状态开发，亦或是面向未来模型性能大幅提升、运作方式截然不同的时代开发？要判断哪一个选择是正确的，真的非常困难。
一方面，你确实希望能稍稍超前一步，就像我们当时做的那样：深知模型目前的表现还不足以支撑我们想做的事，但它们很快就会达到要求。但从同样的角度来说，你也可能陷入自我麻痹，误以为“哦，虽然现在行不通，但只要模型能力提升了，我们就会瞬间拥有一家伟大的公司”。我认为那是很糟糕的想法。
但我觉得，去拥抱那些当前“几乎可行”的挑战是明智的——只要带着一点小小的信念飞跃，你就能坚信它们在不久的将来定会变为现实。