本集简介
双语字幕
仅展示文本字幕,不包含中文音频;想边听边看,请使用 Bayt 播客 App。
好了,各位听众。
Alright, listeners.
现在正是感谢Anthropic公司的好时机,他们是Claude的开发者。
Now is a great time to thank our friends at Anthropic, the makers of Claude.
这里有一件事,可能Acquired的听众还不知道。
So here's something Acquired listeners may not know.
我们实际上一直在用Claude的代码来搭建Acquired平台的基础设施。
We have actually been using Claude code to build infrastructure for Acquired itself.
这让我们能够打造一些如果没有Claude就根本无法实现的听众体验。
This let us create listener experiences that otherwise would not have existed if we did not have Claude.
没错。
Yes.
比如支撑我们YouTube视频字幕的自动转录系统。
Like the transcription pipeline that powers our YouTube episodes.
那就是用Claude写的代码。
That is Claude code.
对。
Yep.
我们还用它来让Acquired的日常运营效率提升了很多。
And we also use it to run our business here at Acquired much more efficiently.
以前我得手动维护一堆老旧的静态表格,用来追踪下载量,试着从中找规律。
I used to maintain these old static spreadsheets for tracking downloads and trying to draw conclusions.
我们用Claude Code写了一些脚本,把这些表格改成了内部的实时仪表盘,能自动抓取整个节目库的数据和我们每周都看的关键指标,好判断每期节目的表现。
We wrote scripts with Claude Code to turn those into custom internal dashboards that pull in live data across the whole catalog and metrics that we actually look at every week to figure out how episodes are doing.
而且我根本不用靠自己那点早就生疏的编程知识,那都过时很久了。
And I didn't need to rely on any of my old software developer knowledge for this, which is, quite expired at
到现在。
this point.
谁都能做到。
Anyone can do this.
我只要说清楚需要什么,Claude Code就全帮我搞定了
I just described what I needed, and Claude Code handled
是的。
Yep.
这些可不是原型。
And these are not prototypes.
每次我们发布新节目,转录系统都会自动运行。
The transcriber runs on every episode we publish.
仪表盘实时获取数据。
The dashboards pull live data.
这是我们日常运营依赖的核心系统。
This is the production infrastructure that we depend on.
没错。
That's right.
如果你是个开发者,整天被那些想甩掉的琐事拖着,ClaudeCode 直接在终端里运行,读取你的代码库,就能帮你写测试、重构代码或者调试,根本不用你一步一指导。
And if you are a developer and you're spending half your day on tasks that you wish you could just hand off, ClaudeCode runs in your terminal, reads your code base, and can take on things like writing tests, refactoring, or debugging without you hand holding through every step.
这可不是代码自动补全。
And this isn't autocomplete.
使用 ClaudeCode,你可以给它一个完整的任务,比如修复一个漏洞或开发新功能,它就能独立完成。
With Claude code, you give it an entire task like fixing a bug or building a new feature, and it handles it.
所以,对于值得解决的问题,赶紧去 Claude.ai/acquired 开始使用 Claude 吧。
So for problems worth solving, get started with Claude at Claude dot a I slash acquired.
嘿,Siri。
Hey, Siri.
你被苹果收购后表现得怎么样?
Were you a good acquisition by Apple?
这个问题挺有意思。
Interesting question.
谁掌握了真相?
Who got the truth?
是你吗?
Is it you?
是你吗?
Is it you?
是你吗?
Is it you?
现在谁掌握了真相?
Who got the truth now?
欢迎来到《收购》第五集。
Welcome to episode five of Acquired.
这一集讲的是Siri和苹果的收购案。
This episode is about Siri and Apple's acquisition.
我们开始吗?
Are we starting?
是的。
Yeah.
我们开始了。
We're starting.
欢迎各位听众再次回来。
So, welcome back to all of our listeners.
和往常一样,你们可以在 Twitter 上关注 Acquired FM,给我们提任何建议或反馈。
As usual, you can, give us any feedback or anything at Acquired FM on Twitter.
最近几期节目真的很有趣。
It's been a fun last few episodes.
我觉得这是新年之前最后一期了。
Think this is our last one before, before the New Year.
是的。
Yeah.
祝大家节日快乐。
Happy holidays, everybody.
对。
Yeah.
没错。
Yeah.
不过在这之前,我们为大家准备了一期特别的内容。
In the meantime, though, we've got a, special one for you here.
我们来聊一聊Hey Siri。
We're doing Hey Siri.
没错。
We are.
没错。
We are.
没错。
We are.
大卫,你来给我们讲讲Siri的收购历史和相关背景吗?
David, do wanna start us off with the acquisition history and facts?
一如既往。
As always.
Siri的背景非常有意思,它是从SRI国际分离出来的,而SRI本身是个很特别的机构,很多人可能都听说过,总部在门洛帕克。
So Siri, super interesting history, a spinout from SRI International, which itself is a fascinating organization, many people are probably familiar with, based in Menlo Park.
它其实成立于1946年,由斯坦福大学的董事会创办,最初的名字叫斯坦福研究所。
It itself was actually started in 1946 by the trustees of Stanford, and was originally called the Stanford Research Institute.
它的目标是将研究成果商业化,也就是做研发,然后孵化出新的公司,把基础技术推向市场。多年来,它一直得到政府、斯坦福大学以及许多其他机构的资金支持。
And the purpose of it was to commercialize well, to research and then spin out and potentially spin out and commercialize basic technology and has been funded by the government, Stanford itself, and many other sources over the years.
做了很多了不起的工作。
Done a lot of great work.
到了2000年代中期,他们开始参与一个叫CALO的项目,这个项目由DARPA资助,SRI是主要参与方之一。
And in the sort of mid two thousands started working on this project called CALO, which was funded by DARPA and that SRI was a major participant of.
这个项目的目标是开发一款基于人工智能的个人助手,专门服务于政府和军方。
And the goal of that was to develop an artificial intelligence based personal assistant for the government and the military.
SRI觉得这项技术很有潜力,可以独立出来,推向市场,甚至用在智能手机上。
And SRI thought that the potential was there to spin this out and commercialize it and put it on smartphones.
所以2008年,他们把后来成为Siri的项目分拆出来,关于命名的经过其实有个很有趣的故事,是在Quora上讲的。
So in 2008, they spun out what would become Siri and there's a really cool story about how they decided to name it actually on Quora.
联合创始人之一、Siri的工程副总裁亚当·切耶尔提到过他们是怎么想出这个名字的。
Adam Cheyer, one of the cofounders, and the VP of engineering at Siri, talks about how they came up with the name.
当然,这个名字和它出身的SRI有着明显的呼应。
And, obviously, it echoes SRI, that the company spun out of.
但达格·基特劳,当时是Siri的CEO,据说在他女儿出生时,曾考虑过用这个名字给她命名。
But Dag Kittlau, was the CEO of Siri, had apparently once considered it as, the name for his daughter, when his daughter was born.
每次你说‘嘿,Siri’的时候,不妨想想这一点。
And Just just think of that every time you say, hey, Siri.
每次你说‘嘿,Siri’的时候。
Every every time you say, hey, Siri.
在挪威语里,Dag是挪威裔美国人常用的名字。
And in in in Norse, Dag is is Norwegian American.
在挪威语中,它的意思是‘引领你走向胜利的美丽女性’。
In Norse, means beautiful woman who leads you to victory.
而在斯瓦希里语里,它意思是‘秘密’——这正是我最喜欢的部分。正如亚当在Quora上写的,这是向项目早期‘隐身模式’阶段致敬,那时他们还叫stealth-company.com,那个网站至今还在。
And then apparently in Swahili, it means secret, which this is my favorite part, is as as Adam writes on Quora, a tip of the hat to their pre named days of the project when they began as stealth-company.com, which still exists.
如果你访问stealth-company.com,会看到一个很复古的网站,上面写着:‘Stealth,尽早加入。’
If you go to stealth-company.com, there is this really old school looking website, saying, Stealth, get in early.
我们正在打造硅谷下一个伟大的公司。
We are forming Silicon Valley's next great company.
我们的目标是彻底改变消费互联网的面貌。
We aim to fundamentally redesign the face of consumer Internet.
在新闻和动态方面,他们的办公室地址已于2月1日确定。
And then in news and events, they have office location finalized on February 1.
1月18日,他们从顶级风投那里融到了一大笔钱。
And January 18, money raised from top VCs, in parentheses, lots.
这才是成功的关键。
That's that's the key to success.
我的理解就是这样。
That's my understanding.
这些才是最重要的事情。
Those are the those are the biggies.
喂。
Hey.
你有办公室了吗?
You got an office?
你拿到投资了吗?
You got funding?
你要彻底改变消费互联网的面貌。
You're gonna revolutionize consumer Internet.
好的。
Alright.
我们跳到后面。
So fast forward.
它已经独立出来了。
So it's it's spun out.
它成了一个独立的公司。
It's its own independent company.
它转型成了面向消费者的產品。
It becomes a consumer offering.
有个应用可以在商店下载,我今天早些时候随便搜了下截图,记得以前手机上装过,但不记得当年那界面有多差了。
There's an app you can download in the store where if you look at some screenshots of it, I was just Googling around earlier today, I remember having it on my phone, but I don't remember how bad that UI used to be.
哦,差劲透了。
Oh, so bad.
但那时候还太早,现在回头看,很难说那只是基本要求,还是他们真的把功能放在了外观前面——抱歉,我说反了,是功能优先于界面。
But it was early days, it's hard to know at this point looking back if like, that was sort of table stakes or if if they were actually just, like, yeah, way way way form before function or I'm sorry, function before form.
那时候还是iOS 3,iPhone 3G和iPhone 3GS的时代,特别老派了。
I mean, this was back in the days of iOS three, I believe, and the iPhone three g and the iPhone three GS.
有意思的是,我记得在苹果收购Siri之前就用过Siri 2,我最喜欢的功能是可以打字输入,不用非得说话。
Super old school and what was interesting, I remember using Siri two before Apple acquired it and my favorite feature actually was the ability to type into it.
真遗憾,苹果后来把这功能给取消了。
You didn't have to use your voice And I'm so bummed that Apple ditched that over the years.
是的。
Yeah.
没错。
And well, yeah.
其实这种问题在美國的很多地方都能看到。
I mean, it's it manifests in so many other parts of The US.
但他们最主要的做法是,跟Nuance签了一份大规模的授权协议,用Nuance来做所有的语音识别。
But the the the main thing that they basically did was they took out a massive license with Nuance, and used Nuance for all the speech recognition.
Nuance本身其实是从SRI分拆出来的公司。
Nuance itself being an SRI spin out from the past.
我之前还真不知道这一点。
I did not know that.
不过那时候它还是家独立公司,多年来一直把语音识别技术授权给各大厂商,还和Siri要整合的那些独立功能服务商合作。
But still an independent company, that over the years has licensed their speech recognition technology to all the major players, and they combined that with partnering with a bunch of the independent functions that Siri was going to serve.
比如查找餐厅,
So looking up restaurants,
打开地图,预订座位。
Open mapping table, making reservations.
天气查询,还有所有那些你可能会用到的接口功能,
Weather, kind of all the ABIs Yeah, that you would yeah.
然后他们就从这里开始推进了。
And they kind of went from there.
所以它变成了一款应用,理论上用语音来查询这些特定功能会更方便,老实说,背后其实用了大量if-then逻辑和正则表达式,简单解析你说的每一句话。
So it became an app where it was theoretically easier to use your voice to look up all those specific functions, and quite honestly, probably a lot of if then statements, a lot of of regular expressions, like kinda just parsing whatever you were saying.
背后全是临时凑合,用胶带绑着撑着。
Bailing or wearing duct tape behind the scenes.
有意思的是,我记得当时这款应用特别酷,上了不少新闻,我还下载了,偶尔用一用,但实际操作起来,远不如直接打开OpenTable应用来得方便。
What's interesting though is, like, I remember this being super cool as an app, getting a lot of press, downloading it, using it occasionally, but it was actually way harder to do anything on it versus just opening up the OpenTable app yourself.
对。
Yeah.
而且想想看,那时候iPhone还没原生支持语音输入。
And it's and and thinking about that too, I mean, that was before there was proper dictation on the iPhone.
我记得那时候经常走着走着,打开Nuance的Dragon听写应用,对着手机说话,再把内容复制粘贴到别的地方用。
So I remember, you know, walking around and opening the Nuance app, the Dragon dictation app, to talk into my phone and then copying and pasting out of that into other things where I wanted to use it.
所以当时的情况是,现在回头看,其实我也没怎么用Siri。
So at the time, you know, now we're looking back at like, well, I don't really use Siri that much.
我用语音输入的频率远高于用Siri,但这些功能后来都被苹果打包成了一体的东西。
I use dictation a lot more than I use Siri, and that all kinda has been bundled together, in Apple as as sort of one big thing.
是的。
Yep.
于是这家公司独立出来了。
So the company spins out.
他们从摩根通纳和门罗风险投资公司筹集了两轮风投,总共2400万美元。
They raised two rounds of venture capital from Morgenthaler and Menlo Ventures, 24,000,000 in total.
这款应用在App Store上大概运营了一年,可能还不到一年,我记得是这样。
The company the apps in the App Store for just about a year, maybe less I believe.
到了2010年4月,当时它只在iOS上推出。
And then in April 2010, and it was only on iOS.
那时候他们还宣布正在开发安卓版和黑莓版。
They'd announced that they were working on Android versions and BlackBerry versions at the time.
你还记得黑莓吗,本?
Remember BlackBerry, Ben?
已经过去很久了
It's been
很久以前了。
a long time.
很遥远了。
Far away.
到了2010年4月,苹果宣布收购了这家公司,虽然价格从未公开,但传闻大约是两亿美元。
And and then in April 2010, Apple announces that they're acquiring the company for they've never disclosed the price, but rumors are around $200,000,000.
再过一年半,到2011年10月iPhone 4S发布时,Siri被整合为iPhone 4S的专属功能正式推出。
And then fast forward another year and a half after that to the launch of the iPhone four s in October 2011, and Siri relaunches as a baked in feature exclusive to the iPhone four s.
而且它是有史以来最紧张的测试版之一。
And as as one of the most intense, betas of all time.
我的意思是,我觉得它作为测试版持续了一年,还是说它经历了两次正式发布?
I mean, I think for a year or maybe it was a was it a full release or two full releases?
Siri在技术上一直属于测试版。
Siri was technically beta.
当时Siri要是出了什么离谱的问题,大家都会说,哦,这是测试版嘛。
And, you know, any any ridiculous things that happened with Siri, ah, it's a beta.
就是Gmail那个情况。
It's the Gmail thing.
永远的测试版。
Perpetual beta.
有趣的是,苹果采取了一个非常典型的做法——iPhone 4S发布时,他们把Siri应用从App Store下架了。
And and interestingly, in a very Apple like move, when the iPhone four s came out, they removed the Siri app from the App Store.
所以你想用Siri,就必须买一部苹果的iPhone 4S。
So if you wanted to use Siri, you had to have an Apple iPhone four s.
你知道吗?
And you know what?
他们这么做的原因肯定有不少,我敢打赌。
There's a variety of reasons they they did that, I'm sure.
其中有一条,我猜很多人会搞混。
One of which I bet people get confused.
我的意思是,你听说Siri发布了,看到广告宣传,你第一反应是不是会去下载那个大家都在新苹果手机上热议的App?
I mean, I bet you hear Siri comes out, you see it advertised, and where do you go to check out the thing that you're seeing everybody talk about on the new Apple phone?
你会上App Store去下载,这可能会让人特别困惑。
You're going to the App Store, and that could be enormously confusing.
更别说,我觉得那年圣诞节帮着卖出了不少iPhone。
Not to mention, I think it helped sell a lot of iPhones that Christmas.
是的。
Yeah.
对。
Yeah.
我想深入聊聊这一点。
And I wanna get into that.
这就是整个来龙去脉。
So that's the history.
各位听众朋友,
Alright, listeners.
现在正是感谢我们一家特别喜欢的公司——Sentry的好时机。
This is a great time to thank one of our favorite companies, Sentry.
Sentry,就是守卫那个意思,我们一会儿会给Acquired的听众们送上一份专属福利。
That's s e n t r y, like someone standing guard, and we have something special for the Acquired community to share here in a minute.
没错。
Yes.
Sentry 帮开发者排查错误和延迟问题,几乎任何软件故障都能发现并修复,赶在用户发火之前搞定。
Sentry helps developers debug errors and latency issues, pretty much any software problem, and fix them before users get mad.
他们官网上说,超过四百万名开发者都认为他们‘还不错’。
As their homepage puts it, they are considered, quote, unquote, not bad by over 4,000,000 software developers.
今天我们要聊的是 Sentry 如何与 Acquired 宇宙里的另一家公司 Anthropic 合作。
Today, we are talking about the way that Sentry works with another company in the acquired universe, Anthropic.
Anthropic 以前用过一些老式的监控系统,但面对他们那么庞大的规模和复杂性,最终选择了 Sentry,好更快地解决问题。
Anthropic used to have some older infrastructure monitoring in place, but at their massive scale and complexity, they instead adopted Sentry to help them fix issues faster.
对。
Yep.
在人工智能领域,崩溃可是个大问题。
Crashes can be a massive problem in AI.
如果你在运行一个大规模的计算任务,比如训练模型,一旦某个节点出问题,可能会影响成百上千台服务器。
If you're running a huge compute job like training a model and one node fails, it can affect hundreds or thousands of servers.
Sentry 帮助他们及时发现硬件故障,从而在引发连锁反应前将其屏蔽掉。
Sentry helped them detect bad hardware so they could quickly reject it before causing a cascading problem.
Sentry 还让他们能在几小时内而不是几天内定位并解决大型问题,尽快恢复训练任务。
Sentry also enabled them to debug massive issues in hours instead of days so they could get back to their training runs.
如今,Anthropic 依赖 Sentry 实时追踪异常、分配错误并分析研究团队所用主要语言——包括 Python、Rust 和 C++——中的故障情况。
And today, Anthropic relies on Sentry to track exceptions, assign errors, and analyze failures in real time across all the primary languages used by Anthropic's research teams, including Python, Rust, and c plus plus.
据 Anthropic 团队说,Sentry 为开发人员提供了一个统一的平台,所有排查问题所需的信息都能在这里找到。
According to the Anthropic team, Sentry gives our developers one place that will have all the information they need to debug an issue.
说到 AI,Sentry 现在推出了一款名为 SEER 的 AI 调试工具。
And speaking of AI, Sentry now has an AI debugger called SEER.
SEER 是一个 AI 代理,它会结合 Sentry 的错误上下文和你的代码库,不只是猜测,而是直接定位那些棘手问题的根本原因,并为你量身提出可直接合并的修复方案。
SEER is an AI agent that taps into the issue context from Sentry and your code base to not just guess, but root cause gnarly issues and propose merger ready fixes specific to your application.
我们非常期待与 Sentry 继续合作。
We're pumped to be working with Sentry.
他们的客户名单非常亮眼,不仅包括Anthropic,还有Cursor、Vercel、Linear等等。
They have an incredible customer list, including not only Anthropic, but Cursor, Vercel, Linear, and more.
如果你也想像全球超过二十万家组织那样快速修复代码问题——从独立开发者到世界顶级大公司,都可以访问 sentry.io/acquired 了解详情。
And if you wanna fix broken code fast like over 200,000 other organizations that use Sentry from indie hobbyists to some of the biggest companies in the world, you can check out sentry.io/acquired.
最后提一下,我们的朋友Sentry将在今年九月于旧金山举办官方的Acquired线下聚会。
And one last thing, our friends at Sentry are hosting the official Acquired meetup in San Francisco this September.
想了解更多详情并获取活动时间通知,请访问 sentry.io/acquired/meetup。
To learn more and get notified about dates, visit sentry.io/acquired/meetup.
我们真的特别期待。
We're so pumped.
很期待在现场见到大家。
We can't wait to see so many of you there.
好了,我们继续回到节目。
And now back to the show.
苹果当初以大约两亿美元收购了Siri。
So Apple had acquired Siri for for $200,000,000 ish.
听到过各种说法,但这一集我们就以这个收购价为准吧。
Heard heard a variety of figures, but I think we'll we'll kinda lock on that on on what we look at the acquisition price this episode.
这算是斯科特·福斯特尔的一个重点项目。
And that was sort of a pet project of Scott Forstall.
大家还记得吧,他当时算是公司的二号人物,也就是所谓的“第二个史蒂夫·乔布斯”。人们常提到乔尼·艾夫在设计美学上的贡献,以及那种专注于打造精美产品的理念。
If you guys remember, was sort of the second Steve Jobs of the company, and people talk about Johnny and the design, design aesthetic and and kinda that that focus on making beautiful things.
但斯科特更像是那个仅次于史蒂夫的产品负责人。
But Scott was kinda the product person that was, like, second only to Steve.
而且我记得,我觉得我记忆没错吧,但在乔布斯健康状况恶化那段时间,很多人都在讨论福斯特可能是苹果的下一任CEO。
And and I remember, I think I'm remembering, right, but in the time when Jobs' health was declining, a lot of people were talking about Forstall as potentially being the next CEO of Apple.
对。
Yeah.
对。
Yeah.
确实如此。
It's true.
很多人现在提到Scott,都会想到他负责了苹果地图,就像Siri是他一手带大的项目一样,苹果地图也差不多是他的心血。
And I mean, think a lot of people now know Scott as the guy that kinda took on Apple Maps, just just like just like Siri was gonna be his baby, that Apple Maps was kind of his his
他更出名的是iOS里那些拟物化设计。
Not too much in the guy infamous for skeuomorphic design in iOS.
我觉得这种说法是事后才贴上的标签,简单地把他负责的领域说成是他的地盘。
I think that's a that's a after the fact assignment and and really easy way to say he that was his little fiefdom.
历史总是由胜利者书写的,本。
History is written by the victors, Ben.
确实如此。
It's true.
其实我觉得,今年他制作的那部百老汇音乐剧还拿了托尼奖。
Scott's actually I think he produced the the Tony award winning Broadway musical this year.
我刚才说的不对。
So I talk about quite No.
是去年的事吧?
Different It was last year?
这事儿真值得查一查,因为我听说的时候简直不敢相信。
That this is worth looking up actually, because it's it's totally crazy when I heard about it.
让我想想。
Let's see.
斯科特·福斯特尔是百老汇音乐剧《欢乐之家》的制作人之一,这部剧讲述了一位女同性恋漫画家回顾自己童年,发现父亲一直秘密出柜的故事。
So Scott Forstall was a producer of Fun Home, a Broadway musical that follows the story of a lesbian cartoonist looking back on her childhood with a secretly gay father.
他离开苹果之后,媒体一片寂静,好像什么都没做。
And so interesting to see after he leaves Apple, total media silence, not doing anything else.
结果我们才发现,他原来在忙这个,最搞笑的是,我记得——我觉得这纯属巧合——就在WWDC主题演讲期间,他发了条推文,那可能是他多年来的第一条推文。
Then we find out this is what he's been up to, and the the really funny part, I remember, I think it was totally coincidence, but it was during the WWDC keynote that he tweeted that he would I think it was like his first tweet or his first tweet in a very long time.
他写道:‘非常激动今年能制作出这部获得托尼奖的音乐剧,感谢所有参与的了不起的人们’,大概就是这类话。
So excited to have produced the the Tony award winning musical this year, thanks to all the amazing people that worked on it, something like that.
我当时就想,这太离谱了,居然就在主题演讲正进行的时候发出来。
And I remember thinking, this is unbelievable, like, right in the middle of the keynote.
天啊。
Wow.
总之,这次收购其实是斯科特·福斯特尔主导的,这也引发了一系列苹果后续的收购动作。
Anyway, so that the acquisition was kind of done by by Scott Forstall, and that kicked off a a a long chain of things happening in Apple acquisitions.
Siri,回头来看,其实背后投入了大量深入的研究,尤其是在语音和——嗯,姑且叫人工智能系统这类技术的架构上。
Siri, you know, as it as it sort of turns out looking back, there's a lot of really intense research that has gone on since then in architecting these kind of voice and and, you know, I guess for lack of a better word, AI systems.
我们现在看到的很多语音助手,我不确定Siri是不是,但我知道谷歌的和Cortana都是基于神经网络架构的,而当年被收购时,Siri这家公司根本不是这样做的。
And, you know, we see a lot of them, I I'm not sure if if Siri is or not, but I know that that Google's is and Cortana is architected as a neural network, and Siri, at the time of acquisition, the company that was Siri, was definitely not that.
我觉得从现在开始,有必要把Siri看成两个不同的东西。
And I think it's important to start, kinda at this point going forward thinking about Siri as two things.
它不再只是苹果当初收购的那个产品,而是一个负责听写、语音搜索等各项功能的内部团队,以及如今我们在iOS设备上使用的那个叫Siri的产品。
No longer as the product that Apple acquired, but as both an organization at Apple that is responsible for dictation, for voice search, for a variety of those things, and the product that is is shipped on iOS devices that we now today know as Siri.
这个产品其实融合了太多功能于一体。
And that product is is is a ton of things all in one.
苹果之所以成立这个团队,并决心在每一个iOS版本中都推这个产品,如果没有那次收购,根本不可能发生。但Siri今天能有这么大的成就,远不止靠那次收购那么简单。
I mean, I think that, the impetus for Apple to actually create this this group and to focus on shipping this product, in every version of iOS going forward, That would not have happened without this acquisition, but the theoretical success that Siri has today as a product is attributed to so much more than that.
苹果开始非常认真地招揽顶尖人才。
Apple got extremely serious about hiring great people.
他们聘了一位资深总监。
They they hired a senior director.
他就是现在的Siri资深总监亚历克斯·阿塞罗,之前在微软研究院工作。
He's now the senior director of Siri, Alex Acero, who is ex Microsoft Research.
他们做了大量收购,包括Topsy、Novauris Technologies、OttoCat、Cue、Spotsetter、VocalIQ这些公司,最近还收购了。
They made a ton of acquisitions, including but not limited to Topsy, Novauris Technologies, OttoCat, Cue, Spotsetter, VocalIQ pretty recently Pretty recently.
还有Perceptio。
And Perceptio.
这些收购,尤其是Topsy,为Siri如今的搜索技术和方法论奠定了坚实的学术基础。
And these things, particularly Topsy, are really kind of the academic rigor around the the search and the methodologies that Siri uses today.
想想看,Topsy那笔收购就花了两亿美元。
I mean, think Topsy Topsy was another $200,000,000 acquisition.
差不多吧。
That one Ish.
对,差不多。
Ish.
是的
Yeah.
我觉得那个可能只是传闻而已。
I think actually that one might be Reported probably.
是的
Yeah.
是的
Yeah.
但Topsy实际上构成了Siri搜索功能的核心,以及其底层技术的运作方式。
But but Topsy is effectively the the backbone of how the search works within Siri, and how the the, kind of deeply technical part of it works.
而且,很多优秀的人才都是跟着Topsy一起加入的。
And I I think a lot of really good people came came with Topsy.
相比之下,你看Siri最初的收购,那三位创始人早就离开了。
Whereas, you look at the actual Siri acquisition, all three of those founders are gone.
他们后来创办了Viv,我猜是这么念的,那是一家更新、更好、更下一代的人工智能公司。
They started, I think, Viv, I think is how it's pronounced, but effectively newer, better, next generation New AI company.
是的。
Yeah.
你知道,那次收购之后,很多人离开了。
And, you know, they they had a lot of people attrite from from that acquisition.
对。
Yeah.
Topsy的那批人很多留了下来,我跟一些了解内情的朋友聊过,都是顶尖人才。
Topsy, lot of those people stayed with the company, and from from talking to some friends that are kinda in the know, really, really top people.
我和戴夫之前还聊过,觉得可以做个双集专题,专门讲Topsy和Siri这家公司,因为Topsy对今天Siri产品的核心作用真的太关键了。
And, you know, I I Dave and I were talking earlier, thought about kinda doing this this as a dual episode on on Topsy and this and the Siri company, because that's that's sort of how important Topsy is to the product that Siri is today.
对。
Yeah.
我觉得有几点特别有意思。
I think a couple really interesting notes.
就像本提到的,这个收购居然催生了苹果内部一个全新的部门,负责AI和搜索这类平台技术——而这些,以前根本不在苹果的基因里。
I mean, as Ben mentioned, fascinating that this acquisition has led to the creation of this whole organization within Apple that, you know, AI, search, and those types of that type of platform technology was so not in Apple's DNA prior to the Siri acquisition.
是的。
Yeah.
我真的觉得,要不是我们砸了大笔钱收购了这家公司,这个产品根本不可能面世。
I truly believe that we wouldn't have this product shipped if not for, wow, we just unloaded a bunch of cash acquiring this thing.
我们得认真对待这件事。
We gotta be serious about it.
没错。
Yep.
更有趣的是,苹果当初收购的那些Siri原班人马,其实早就走光了。
And what's doubly interesting about that is that the actual people at Siri that Apple acquired, like, they're gone.
真正重要的是,我们收购了这家公司,把它的产品和功能整合进iOS,这才为苹果后来打造平台打下了第一块基石,之后所有优秀的人才都是陆续招进来的。
It was acquiring this company and bringing this product and this feature into iOS that was the first foundation of a platform that Apple started building and, and all the great people that they've acquired and hired since then.
你可以说,而且说得很有道理,谷歌在人工智能、搜索、语音搜索和语音识别这些领域,目前还是领先于苹果。
And you can argue, I think, probably very justifiable justifiably so that Google is still ahead of Apple in a lot of these areas, AI, search, voice search, speech recognition.
但毫无疑问,如果没有这次收购,苹果现在的差距只会大得多。
But certainly, is not nearly as far behind as they would have been had they continued along the path they were before this acquisition.
不行。
No.
我觉得,谈到任何收购时,都可以用‘自建还是收购’这个框架来思考。
And I think that, in kind of thinking about any acquisition, you can use the framework of build or buy.
如果你是公司领导,在决定收购之前,你得想清楚:我们确实需要这个东西,花两亿美元买下来,或者自己开发。
What's, you know, if if you're thinking about it before you make the acquisition, you're a leader at that company, and you've decided that, you know, this is something we need to pursue, and it's gonna cost us $200,000,000 to buy, or we could build.
但我觉得这里的情况不是这样。
And I don't think that that is the case here.
我不觉得苹果公司
I don't think that Apple
他们真能自己搞出来吗?
Could they even have built?
他们能请到机器学习、人工智能、语音识别领域最顶尖的人才吗?
Like, would they be able to hire the really top people that they would have needed to in machine learning, AI, speech recognition Yeah.
要做出这样的成果,光靠自己根本做不到。
To make this happen.
是的。
Yeah.
我觉得这挺有意思的。
I don't it's interesting.
这本来就是个巨大的挑战,现在依然是。
It would have been a tall order, and and still was.
我的意思是,即使他们收购了Siri之后,要把它打造成今天这个产品,也还有很长的路要走。
I mean, I think that even after they they did the Siri acquisition, they they had a long, long road ahead of them to to turn it into the product it is today.
所以让我想的是,就算它换了名字,起步方式也不一样。
And the the question that sort of begs to me is, you know, it would have a different name, and it would it would have started off in a different place.
但如果他们根本没收购Siri,而是只靠其他收购和关键人才引进呢?
But what if they didn't acquire Siri at all, and they just made all these other acquisitions and these other key hires?
毕竟他们已经从Nuance挖走了不少顶尖人才。
Because they've hired a lot of the really great people away from Nuance.
对。
Yep.
他们实际上已经自己从头搭建了整个语音识别系统,以前这些可是从Nuance授权来的。
They've actually architected the whole kind of, speech recognition pipeline that they were licensing from Nuance internally.
他们把从Siri买来的代码全部重写甚至彻底扔掉了。
They've they've rewritten and thrown away all the code that they've acquired from Siri.
现在既没有Siri的团队,也没有Siri的技术,只剩下Siri这个品牌,以及当初那个产品原本的功能。
So there's no Siri people, there's no Siri technology, there's merely a Siri brand and basically what that original product did there.
但今天iOS里Siri这个产品之所以成功,主要功劳并不在当年收购来的那些东西。
But a lot of the success of what Siri, the product in iOS is today, is not attributed to what they actually acquired.
是的。
Yeah.
顺便说一句,挺有意思的是,关于Nuance,现在大部分主流移动平台公司——微软、谷歌、苹果——都已经不再用他们了。
And it's interesting, a little bit of an aside, on Nuance, most of the major mobile companies, mobile platform companies, Microsoft, Google, Apple.
我觉得,目前几乎所有主要公司都已脱离Nuance了。
I believe most or all of them have moved away from Nuance at this point.
是的。
Yeah.
我觉得确实如此。
I think that's true.
Nuance的销售代表们肯定气坏了。
Gotta be some really upset account reps at at Nuance.
是的。
Yeah.
我很喜欢我们刚才的讨论。
So I love the discussion we've been having.
说起来,我们现在都快到2015年底了,还在聊这个话题。
In a way, you know, we're sitting here at the end of twenty fifteen talking about this.
这次收购都过去快六年了。
It's been almost six years since this acquisition happened.
我们还没怎么聊过Siri这个产品或功能本身。
We haven't really talked about Siri, the product slash feature yet.
我觉得,很多人也会同意,到目前为止,它还没达到当初的预期。
And I think, you know, I would argue as would a lot of people, it really hasn't lived up to its promise so far.
我不知道你怎么样,本,但我用Siri就干一件事:设闹钟。
I don't know about you, Ben, but I I use Siri for one thing, setting alarms.
我真想看看苹果内部关于这个的使用数据分布图,我觉得Siri根本没走多远。
It's a it's a I got I would love to see the histogram of this at Apple, because I think I just don't think Siri has gotten that far.
他们其实有两个问题。
Well, they have two problems.
第一,我觉得他们在真正实现所谓‘学习’能力方面,根本没做到位。
One, I don't think they've gotten that far from actually being able to, you know, do things that are like, quote, unquote, learning.
对吧?
Right?
比如我,我就用它做几个固定的操作,就像当初的Siri那样——我会说:‘嘿,Siri,帮我找找附近最好的早餐店’之类的。
Like, I know I have these very specific functions, much like the original Siri, where I go, hey, Siri, look up the best breakfast restaurant around me, or something like that.
第二个问题是,就算他们将来真能解决这个问题,推出一套更复杂、真正深入底层的搜索和理解系统,我又怎么知道它升级了呢?
And the sort of second problem around that is if they have fixed that, or they they do in the future kinda roll out something that's much more technically sophisticated, that's truly like a deep systems approach to search and and understanding and, you know, all all these deeply technical things, how am I supposed to know about it?
因为像我们这种人,至少会看WWDC主题演讲,觉得那挺酷的。
Because I think that, like, there are some of us that, you know, will watch the WWDC keynote, and that's super cool.
但这种没有界面、看不到所有功能的语音交互方式,其实挺危险的。
But it's a really sort of dangerous thing with these voice based interfaces that don't have a layout of all the possible functions in front of your face that you can use.
要是你试了几次它都搞砸了,你就会觉得它根本做不到那些事,以后再也不想用了。
If it falls on its face a couple times for you, you basically assume it can't do things in the future and you never try again.
是的。
Yep.
而且你很容易就掉进那种诡异的谷底,然后直接放弃它。
And you definitely fall into that uncanny valley and then you just reject it.
所以我才说,2015年快结束了,我们还在讨论这个话题,真的很有趣,因为可以说,Siri这个产品本质上已经失败了。
And that's why I said it's super interesting we're sitting here at the end of 2015 talking about this because I think it's noncontroversial to say Siri as a product has been a failure.
就连一些最基本的功能,比如我老婆的名字叫Jenny,J-E-N-N-I-E。
All these, like, even just basic basic stuff like, you know, my wife's name is Jenny, j e n n I e.
这是她的正式名字。
That's her given name.
每次我跟Siri提起我老婆,它都念成J-E-N-N-Y。
Every time I talk to Siri about my wife, it says j e n n y.
她的姓跟我不一样。
And she has a different last name to me.
她没改姓。
She didn't change her name.
Siri 就是搞不明白这一点。
Siri doesn't, like, figure that out.
真是烦人。
It's so annoying.
这些都是最基础的功能。
Super basic stuff.
但我们现在还坐在这儿。
But we're sitting here right now.
你知道吗,今年亚马逊推出了 Echo 和 Alexa。
You know, Amazon came out with the Echo and Alexa this year.
他们在上面投入了大量资源。
They're investing heavily in that.
毫无疑问,亚马逊认为语音和语音界面将成为未来科技的重要组成部分,谷歌和苹果也是这么想的。
Definitely, Amazon believes that voice and voice interfaces are gonna be a huge part of technology going forward as does Google, as does Apple.
在很多方面,我觉得这次收购的完整故事还没写完,因为接下来还有更多事情要发生。
In a lot of ways, I think this the the full story on this acquisition hasn't been written yet because there's gonna be so much more to come.
是的。
Yeah.
而且确实还应该继续下去。
And then there needs to be.
有意思的是,我们这个节目讲的是那些真正成功的科技收购案,而我们今天能做这一期,正是因为苹果必须这么做。
I mean, the interesting thing, you know, we say this show is about technology acquisitions that actually went well, and if you look at kind of the app we can do this this episode of this show because Apple needed to do this.
说白了,把个人助手内置到移动操作系统里,已经是基本要求了。
I mean, it's it's effectively table stakes to have a, personal assistant baked into your mobile operating system.
要是不这么做,那简直太丢脸了。
Like, they they it would be embarrassing
继续这样下去可不行。
for to continue do that.
你刚说那话的时候,我左手扭了一下,不小心激活了苹果手表上的Siri。
As you're saying that, I I had my my left hand twisted back, and, I accidentally activated Siri on my Apple Watch.
你开了‘嘿,Siri’功能吗?
Oh, do you have Hey Siri on?
没有。
No.
是我一直捏着数字表冠造成的。
It was holding the the digital crown.
这怪你。
That's your fault.
但从根本上说,这功能必须要有,尤其是现在用上了手表、电视、智能音箱,你得能对着手机说‘嘿,Siri’,这已经是任何平台都不可或缺的部分了。
But but it's so fundamentally, like, this needs to be a part of especially as you move into watches and televisions and connected speakers and being able to say, hey, Siri, to your phone, a critical part of any platform.
是的。
Yeah.
没错。
Yeah.
确实如此。
It is.
挺有意思的。
It's interesting.
要想出Siri做不到的事情,其实挺难的,因为你的脑子已经被它能做的事给框住了。
In trying to think of examples of things that Siri can't do, it the thought exercise of it is actually difficult because your mind is trapped in the things that you can do.
我刚拿起手机,想着:嘿,Siri,我明天上班前有时间去跑步吗?取决于我第一个会议的时间,还有我想走路去上班,但又不知道现在天气怎么样。
Like, I'm pulling up my phone thinking of like, you know, hey, Siri, am I gonna have time to go for a run tomorrow before work based on when my first meeting is and the fact that I wanna walk to work not knowing what the weather will be right now.
你好。
Hi.
本·莱德海姆。
Ben Ledheim.
‘莱德海姆’这个念法不对,正确读法跟‘光明节’一样。
That's a mispronunciation of Ledheim because it's Hanukkah.
比如
Like
没错。
Right.
还有很长的路要走。
There's a long way to go.
路还很长。
Long way to go.
各位听众朋友们,
Alright, listeners.
现在正好跟大家介绍一下我们特别喜欢的一家公司——WorkOS。
Now is a great time to tell you about one of our favorite companies, WorkOS.
正如很多朋友知道的,现在开发软件的成本已经低到几乎可以忽略不计。
So as many of you know, the cost of building software has collapsed to basically zero.
你甚至一个周末就能做出一个能用的产品,代码大部分都是AI帮着写的。
You can build a real working product in a weekend, and AI writes most of the code.
这当然很棒,直到你打算开始卖它的时候。
Which is great right up until you wanna start selling it.
展开剩余字幕(还有 272 条)
因为企业客户根本不关心你有多快把产品做出来。
Because enterprise buyers don't care how fast you built it.
他们在乎的是数据安全,比如单点登录、SCIM、审计日志、基于角色的权限控制,所有这些都要有。
They care about their data, and they care about security, like SSO, SCIM, audit logs, role based access, the entire list.
这些你都没做,因为它们并没有让产品本身变得更优秀。
None of which you've built because it didn't make your product any better.
它们只是挡在你和成交之间的障碍。
It just sits between you and the deal.
而这个差距,现在显得格外明显。
And that gap has never felt more abrupt.
你几天之内就从一个想法变成了可用的产品,可现在却要面对好几个月的认证系统开发,才能让大企业愿意跟你开个会。
You went from this idea to a working product in days, and now you're staring down months of authentication plumbing before a large enterprise will even take the meeting.
这正是WorkOS存在的全部意义。
This is the whole reason that WorkOS exists.
对。
Yep.
他们把过去阻碍你拿下大单的企业级需求,变成了即插即用的 API。
They turn enterprise requirements that previously gated your big deals into drop in APIs.
你的团队下午就能搞定企业单点登录,不用再耗上一个季度,赶紧回去做那些用户真正想要的功能。
Your team adds enterprise SSO in an afternoon instead of a quarter and gets back to building the thing that people wanted in the first place.
你知道的,比如让你的啤酒更好喝的那个东西。
You know, like, what makes your beer taste better?
你周末搭出来的那个产品,周一就能直接满足企业客户的需求。
The product you built over the weekend really can be ready for the enterprise on Monday.
我们还真在旧金山和 WorkOS 团队一起度过了一个特别有意思的晚上,采访了 Claude Code 的负责人 Boris Cherny。
We, actually got to spend a really fun evening in San Francisco with the WorkOS team doing interviews about this with the head of Claude Code, Boris Cherny.
没错。
Yes.
那晚真的特别棒。
It was, it was an awesome evening.
说到那晚,听众朋友们,如果你想听 Acquired Unplugged 活动中那些完整的对话,可以在节目笔记里点击链接,或者直接访问 acquired.fm/unplugged。
And speaking of that evening, listeners, if you wanna hear those full conversations from that Acquired Unplugged event, you can click the link in the show notes or go to acquired.fm/unplugged.
WorkOS 还为那晚特别定制了一副《Acquired》主题的扑克牌,而且多出了几副。
And WorkOS also made super cool custom acquired themed decks of playing cards for the night, and, there are a few extras.
你可以去 shop.workos.com 买一副,只要输入代码 ACQ deck,只要库存允许,就能免费拿走。
So you can go to shop.workos.com and snag yours as long as you enter the code ACQ deck for free while supplies last.
好的。
Alright.
那我们进入节目的下一个环节,来分析这次收购属于哪种类型。
So let's move into the next segment of the show where we categorize this acquisition.
我现在没看屏幕,不过我觉得我们有这几个分类:
I'm not looking at it right now, but I think we've got
我们有用户、技术、产品、业务线,还有其他。
We've got people, technology, product, business line, and other.
对我的观众来说,他们当然希望听到不同意见,但我真觉得,本,你到底觉得该归哪一类?
For me, you know, our audience is craving disagreement here, but I just don't think well, Ben, what's your category?
如果你没做足功课,那肯定觉得这是技术类。
So I think if you, like, didn't do your research, it's technology.
你觉得你是在收购一个特别学术的东西,确实它部分属于技术,但自从Siri刚起步以来,整个领域已经进步太多了。
And you think you're acquiring this super academic and I guess it is partly technology, and we've just come a long way since since Siri was started.
但归根结底,对苹果来说,这更像一个产品收购,因为他们根本没保留原来那支团队,技术底子也早就不是当初那套了。
But, I mean, what it ultimately became for Apple is a product, because they sure as hell don't have the people, and they they don't have the technology that it was originally built on.
整个行业的发展方向已经变了。
And the world has moved in a direction.
值得肯定的是,苹果一直紧跟这股潮流,投入了大量资源,还请来了这个领域最顶尖的人才,来推动技术真正向前发展。
And Apple is, to their credit, really kept up with this and and and put some serious muscle behind this and hired some of the top people in the field to to build out the the direction of the technology that's actually moving in.
所以我把它归类为产品收购。
So I'm calling it a product acquisition.
没错。
Boom.
我不同意。
I'm disagreeing.
好吧。
Alright.
也许我之前没做足功课。
Maybe I didn't do my research.
我觉得这是一次技术收购。
I think it's a technology acquisition.
就像我们刚才说的,Siri 这个产品其实挺烂的。
I mean, as we were just saying, like, Siri as a product sucks.
是的。
Yeah.
我没说他们收购了个好东西。
I didn't say they acquired a good one.
我的意思是,
I mean
说得有道理。
Fair point.
说得有道理。
Fair point.
是的。
Yeah.
我的意思是,
I mean,
我觉得你看,
I think look.
我们已经聊过这些了,但对我来说,这是一次技术收购,因为我根本不把Siri当成一个产品。
We've already talked about all of this, but but I think for me, it's a technology acquisition because I don't view Siri as a product.
它确实差劲,但我觉得,哪怕往后看,它也不会真正成为一个独立的产品或功能。
Like, a, it sucks, but, I don't think it is even really especially going forward, will be a product or a feature.
我觉得它会彻底融入整个系统。
I think it's just gonna be so fundamentally baked into the platform.
语音、语音交互,还有它的预测性助手功能,这些都会成为一项基础技术,并且会越来越重要。
Voice and voice interface, and and the predictive assistant parts of it, that I think that is just gonna be a a fundamental technology piece is a fundamental technology piece and will be more so over time.
没错,他们收购的并不是完全这个东西,但核心就是这个,之后慢慢发展起来的。
And, yes, that's not exactly what they acquired, but that was the kernel of it that they built over time.
是的。
Yeah.
我的意思是,我毫不怀疑,未来人类与机器交互的方式,语音会变得更加核心。
I mean, I I have no doubt that voice will be more central to the way that human beings compute in the future and interact with machines.
我觉得屏幕不会消失。
I don't think screens are gonna go away.
但我们会越来越少依赖屏幕,更多地用对话的方式跟机器互动。
I think we'll become much less reliant on them and much more conversational with with machines.
但问题是,苹果现在做的这些,真的走在正确的方向上吗?他们能打造出这样的未来吗?
But the question is, is Apple doing this and building what they're doing today actually moving in the right direction so they're gonna be the ones to create that future?
还是说,会有人从根本上重新定义什么是计算机?
Or is it gonna be someone that fundamentally kind of takes a new approach to what is a computer?
你怎么跟它互动?
How do you interact with it?
能不能完全靠语音来操作,而不被过去那种把计算未来硬绑在屏幕交互模式上的思维拖累?
Can you do it all over voice and not be hampered with a legacy of let's try and tie this future of computing into the way that people interact with this screen?
对。
Yep.
说得真好。
That's a great point.
我这算是偷懒了,直接跳到科技趋势上去了。
I guess that's me cheating and preemptively launching into technology trends.
太棒了。
Love it.
我们节目的下一部分是:你认为这反映了当前科技领域的哪些趋势?
The next part of our show is, what trends does this illustrate for you in technology right now?
这确实是一个。
That's absolutely one.
每次有人跟我聊他们正在做的项目,只要是关于自然交互、无屏计算,或者想到《她》这部电影,最近几个月我至少已经跟三个创业者聊过,他们来自微软、苹果这些公司。
I mean, every time I have someone talk to me about a thing that they're working on that's more about a natural interface or screenless computing or thinking about the movie Her, I've had probably three conversations in the last few months with with different entrepreneurs working on this, people at Microsoft Apple.
现在有太多AI公司了,是的。
So many AI companies right now Yeah.
他们做的所有事,就是搞一个AI研究平台,目标就是被脸书、谷歌或者苹果收购。
Doing everything from just, like, we are an AI research company platform that is building to be acquired by Facebook or Google or Apple.
我觉得这个主题特别棒,也确实反映了科技界的现实。
I think I love that theme and is so true in technology.
你不能用老一套的思维去应对未来。
You can't bring old world thinking into the future.
这就像是亨利·福特说的,与其造更快的马,不如直接发明汽车。
It's it's it's like it's like the Henry Ford analogy of a faster horse.
是的。
Yeah.
如果十年后苹果真的成了这个领域的领头羊,人人都在跟语音设备互动,那我们今天这话就算说错了,但我总觉得不太可能。
And, know, we'll eat these words if in ten years Apple's the the leader in this category and everyone's interacting with their their voice device, but I don't know.
我现在很难想象Siri能演变成我心目中计算技术的未来形态。
It's hard for me to see Siri today evolving into what I think is clearly the future of computing.
对。
Yeah.
但我也不觉得屏幕会消失。
But I also I also, I don't think screens are going away.
我觉得人机视觉交互不仅不会停止,反而会随着计算技术的发展越来越普遍。
I think visual interaction with computers is not gonna stop and is only gonna continue to rise with computing.
但你也不是总想这么做。
But you don't always wanna do that.
有时候你更想要语音交互,比如在厨房里——这正是Echo的主要使用场景。
Sometimes you want a voice based interaction, like in the kitchen, which is the primary use case for the Echo.
这让我想起一个类似的趋势。
It reminds me of a similar trend.
想想看,自从平板电脑推出以来,销量出现了大幅增长。
So thinking about the fact that you've got this massive spike in tablet sales since they came out.
对很多人来说,平板电脑虽然不如手机那么成功,但恰恰是他们真正需要的那台电脑。
For so many people, you know, the the tablet is not as successful as the the phone, but for so many people, the tablet is the computer that they actually needed.
自从个人电脑问世以来,人们一直购买并用它来做各种事情,但当时我们能造出来的设备,往往被强行用于它并不擅长的多种场景。
And ever since personal computers came out, people have been buying them, people have using them for a variety of purposes, but it was really the device that we could build at the time that was over applied to a variety of use cases that were outside of the the specific thing that they were actually good at.
对
Mhmm.
说得太好了。
And perfect story.
去年感恩节在我奶奶家,帮她搬东西。
At my grandma's house, over Thanksgiving, helping her move some stuff around.
她得把电脑 unplug,因为有人要在这房间铺新地毯,得把电脑挪到另一个房间。
She needs to unplug her computer because someone's gonna be doing they're gonna be putting new carpet in in that room, and they're moving it to the other room.
我就说,哦,那我插上电,先在这边房间把电脑设好。
And I'm like, oh, well, I'll plug it back in, and we'll set it up in this other room for now.
她却说,不用麻烦了。
And she goes, you don't need to.
我一个月都没碰过它了。
I haven't been on it in a month.
我当时就说:什么?
I was like, what?
因为她平时用电脑干很多事,不是那种很少用的人。
Because she, I mean, she she really does a lot on her computer, it's it's not like she she's
她用的是iPad。
On her iPad.
她对电脑其实很在行。
Or well well, she's very proficient with her computer.
她现在完全靠iPad和iPhone了。
She's been fully reliant on her iPad and her iPhone.
我之前根本不知道这一点。
I didn't realize this.
我跟她一直发邮件,其实我不确定她是不是在iPad上做表格,但她是完全可以的。
Mean, I've been emailing with her, and I actually don't think she does her her spreadsheets on the iPad, but I think But she could.
她完全可以。
She could.
她完全做得到,只是我还在跟她说清楚这个新的操作方式。
She totally could, and it's it's just me kind of explaining the new new input mechanism to her.
但对很多人来说,我想说的是,过去电脑承担了太多任务,而且每个任务都很明确。
But for so many people, the point I'm getting at is is that the computer was doing a lot of jobs, and it had a very specific job.
后来平板电脑出现后,人们发现它在处理很多消费类任务——比如轻量级邮件、这类事情上,其实比电脑更合适。
And then when the tablet came out, it revealed that the tablet is actually a better thing to do a lot of those jobs around consumption, around lightweight email, things like that, than the computer was.
所以现在大家确实用平板来做这些事,因为它本来就是为这个目的设计的。
So then people are actually using tablets, a better thing for that purpose, for its intended purpose.
我觉得现在的屏幕和当年的个人电脑一样,已经过度饱和了。
I think screens right now are oversaturated the same way that PCs were oversaturated.
现在有太多屏幕在做根本不需要它们做的任务。
We've got screens doing a lot of jobs screens don't need to be doing.
当然,屏幕在需要大量信息展示的场景下,还是会派上用场的。
And sure screens will be doing things where there's a heavy display of information.
是的。
Yeah.
说得很到位。
Great great point.
需要同时查看多个信息、不是单一任务的场景,还有那些需要分层上下文、不断叠加新信息的模式化界面。
Heavy display of information, things where you need to reference multiple things in parallel and not just have a single track, Things where you have modal interfaces, where you need to understand context with new context on top.
但确实有很多事情根本没必要放在屏幕上,我们却偏偏用屏幕来做。
But there sure is a lot that does not need to be on a screen that we use screens for.
对。
Yep.
我还想补充一个观点,这正好说明了做消费类产品,尤其是做出色的消费类产品,真的非常难。
Another theme that I'd throw in, that I think this illustrates is that consumer generally and specifically building great consumer products is really hard.
我觉得Siri就是个典型例子,很多公司和产品都会掉进同一个陷阱:我们有项超棒的技术。
I mean, I feel like Siri is a classic example that a lot of companies and products have a trap they fall into, which is I've got this awesome technology.
我们能实现很多了不起的功能。
We can do these amazing things.
看看这些炫酷的演示效果。
Look at these really cool shiny demos.
可一旦进入真实世界,面对各种边缘情况,要持续稳定地提供实际价值,真正解决用户的问题,还不让用户觉得麻烦,那就难多了。
And then you get it out in the real world with, like, edge cases and delivering actual value on a consistent basis, solving people's real problems, doing that without friction.
要在演示中兑现这些承诺,真的非常难。
And it's really, really hard to live up to those promises in the demo.
想想大多数消费类产品,那些真正成功的,往往都特别简单。
And when you think about most consumer products, many or most consumer products that end up being really successful, they're really simple.
它们的价值一目了然。
And it's very clear what their value is.
比如Airbnb,我最近可没少用它。
Like Airbnb, you know, which I've been on an Airbnb kick lately.
说起来惭愧。
Deserve myself.
我和我妻子过去半年刚注册成为房东,收入都快Cover一半房贷了。
My wife and I just signed up to be hosts in the last six months, and it pays half our mortgage.
这太棒了。
And that's amazing.
作为房东,Airbnb给我带来的就是这样的价值。
And that's like as a host, that's what Airbnb does for me.
我们还能遇到很多特别棒的房客,而且这个假期我们也要去好几个地方探亲。
And we get to meet these really cool people who are coming through, and we're traveling for the holidays to a whole bunch of places to see family.
这个月我们打算住三家不同的Airbnb。
And we're staying in, I believe, three different Airbnbs this month.
所以你的意思是,你完全理解这个产品的设计初衷。
So what you're getting at is you you very much understand the intended purpose of that product.
当然。
Absolutely.
它太简单了。
It's so simple.
设计得特别优雅,完美解决了我的问题。
It's so elegant, and it perfectly solves my problem.
Siri?我都不知道它到底该干嘛。
Siri, I don't even know what it's supposed to do.
看着挺酷的。
It looks cool.
我看了主题演讲,心想,太棒了。
I watched the keynotes, and I'm like, that's awesome.
然后我就再也没用过。
And then I never use it.
对。
Yeah.
我是说,说到另一个例子,我坐在那儿试来试去。我有新款Apple TV,前几天晚上躺在沙发上看YouTube找视频,结果发现那输入体验简直糟透了。
I mean, to take it to another I'm sitting there trying I'm on so I have the new Apple TV, and I was looking for some videos on YouTube, laying there on the couch the other night, and it's ugh, it is the worst to type into.
就像只有一行字符在屏幕上滑动,还得去点选单个选项,感觉又回到了用PS2的时代,我当时还在用关键词搜索。
Just like one row of characters swiping around, trying to tap individual it's like being on PS two again, and I was using the keyword.
总之,我就想,哦,等等。
Anyway, so I'm like, oh, wait.
Siri,这本来应该是一个对话式的用户界面才对。
Siri this is supposed to be like a really conversational UI.
我还是用Siri吧。
I'll use Siri.
我就说:‘嘿,Siri,在YouTube上找一下阿齐兹·安萨里的脱口秀。’
And I'm like, hey, Siri, look for the Aziz Ansari stand up on YouTube.
一开始我说‘阿齐兹·安萨里的脱口秀’,结果出来的是他演的电影,就是那种院线上映的片子。
Or at first, said the Aziz Ansari stand up, and it brought up like movies with Aziz Ansari, and they were like, you know, theatrical releases.
我当时就想,哦,我得说得更具体点。
And I was like, oh, I should probably specify.
于是我又说:‘在YouTube上搜阿齐兹·安萨里的脱口秀。’
And I'm like, look up the Aziz Ansari stand up on YouTube.
结果还是一模一样的搜索内容。
Exact same search results.
我当时就震惊了,这玩意儿根本就没连上YouTube啊。
And I was like, oh my god, it's not even plugged into YouTube.
每次我灵光一现,想着‘这次用Siri试试’,结果都不靠谱。
And like, every time I have this brilliant idea where I'm like, I should use Siri for that, It doesn't really work.
对啊。
Yep.
所以我觉得我们这些科技迷还挺怪的。
So And I think we're weirdos, like, we're tech people.
我就是那种走在街上下雨了,不玩手机的人,那我干嘛呢?
I am the kind of person that walks down the street and it's raining, so I'm not typing on my phone, and and what do I do?
我会按住Siri,跟它说:Siri,发个消息, blah blah blah blah blah。
I, like, hold down Siri, I'm like, Siri, text back blah blah blah blah blah blah blah.
是的。
Yeah.
因为你根本不介意走在街上跟空气说话被人看成怪人。
Because you're not worried about the social stigma of like, talking to nobody walking down the street.
十足的怪人。
Total weirdo.
我知道自己这样没问题,但也清楚别人根本不会这么做。
And I, you know, I'm fine with that, but I fully recognize that like, other people aren't doing that.
我猜,如果苹果能统计出每天或每小时真正使用Siri的人数,结果恐怕会让人很失望。
And I had I have to imagine if you can see the numbers at Apple of the people that are actually, like, using Siri on an hourly or daily basis, it's it's probably very disappointing.
是的
Yep.
我很好奇,如果你在听的时候不认同我们的观点,欢迎告诉我们你平时怎么用Siri,都用来做什么。
I'm curious, as you're listening to this, if you disagree with us, you should totally let us know that you use Siri and what you use it for.
对
Yeah.
我真的好想知道。
God, I'd like to know.
我也特别想知道。
I would love to know.
可以在Gilbert、DJ RosenT和Acquired FM上找到我们。
At Gilbert at DJ RosenT at Acquired FM, find us.
对
Yeah.
我还想说另一件事
The other thing I wanna
在继续之前,我想补充一点,做消费类公司、消费类产品真的特别难。
I wanna add on that before before we move on, you know, this is also, you know, building consumer companies, consumer products is super hard.
这也就是为什么,作为风投,我们在看消费类公司时,特别关注用户分群和留存数据。
This is one of the reason why as VCs, really focus when we're looking at consumer companies on cohort and retention data.
因为光看获客数据是不够的。
Because just like straight up acquisition data, user acquisition data isn't enough.
关键问题是,用户会不会一个月又一个月、一天又一天地回来?
Like, the question is, are people coming back, you know, month after month after month, day after day after day?
他们是不是持续觉得你的产品有价值?
Are they continuing to see value from your product?
这真的非常难。
And it's really hard.
对。
Yeah.
我记得2015年2月,Siri的一位创始人接受TechCrunch采访时提到,每月有超过两亿人在用,每天有超过一亿人活跃使用。
I wonder so in February 2015, one of the founders of Siri, was was remarking in in a an interview to TechCrunch that more than 200,000,000 people use it monthly, and more than 100,000,000 people use it every day.
每五个能使用 Siri 的人里,就有一个是每日活跃用户,这有点让人惊讶。
So it's one in five people with access to Siri actually are a daily active user, which is a little shocking.
我不禁好奇,这个数据的覆盖范围到底有多广。
Like, I wonder how wide the purview of that is.
而且我也想说,可能每个人听到都会笑,但你坐在会议里,一按按钮,Siri 就跳出来,那场面真的挺尴尬的。
And I also wonder I mean, every single person out there is gonna laugh, but, like, you're you're sitting in a meeting, you hold the button, it goes into Siri mode, you're super embarrassed.
我都不知道一天里不小心触发 Siri 多少次了。
I can't tell you how many times a day I accidentally
是啊。
Yeah.
不小心唤醒 Siri。
Open Siri.
好在他们把那个提示音关了,不然我在会议里真成笑话了。
I mean, I'm just glad they turned off the little doo doo so that I don't look like a total idiot in meetings.
Siri 最常见的使用场景,估计就是这种在会议上不小心触发的尴尬瞬间了。
That must have been the number one use case for Siri was the accidental embarrassing meeting doo doo.
对。
Yep.
好吧。
Alright.
我们今天就到这里吧,本。
We wrap this one up, Ben.
是的。
Yeah.
我觉得是时候给它打个分了。
So I think it's time to give it a grade.
要不是他们做了这个,苹果可能根本不会往这个方向走,说不定还会跟着谷歌和微软后面学。
You know, I think that if they didn't do this, Apple might not have moved in this direction and might have tried to follow after Google and Microsoft did it.
他们骨子里根本不会这么干,除非先大肆收购,把其他公司的顶尖人才都挖过来。
I don't think it's really in their DNA to start this without going on an acquisition spree to start acquiring, a lot of talent, hiring people from other companies.
现在这东西算是基本配置了,但苹果的iPhone销量,根本不会因为有Siri就多卖一台。
These days, it's sort of table stakes to have it, but Apple definitely does not sell a single, like, one single more iPhone because it has Siri versus it doesn't.
我猜他们因为内置了Siri,卖出了更多iPhone 4S。
I bet they sold a lot more iPhone four s's because it had Siri.
当然。
Sure.
就一次。
Once.
是的。
Yeah.
就一次。
Once.
我觉得他们现在对Siri做的任何改进,都没能让多卖一台iPhone。
I don't think anything they're doing to Siri now sells a single additional iPhone.
同意。
Agree.
比如,当他们把Siri扩展到其他市场的时候。
Or, like, when they, you know, add Siri to another market.
我觉得在中国可能会出现一个有趣的现象,因为那里打字特别麻烦,而用语音输入就方便多了,说不定使用率会慢慢上升。
I think that there's an interesting thing, like, maybe we'll see in China, maybe we'll see it pick up a little bit since in China it's so much more difficult to type, and it's so much easier to
对。
Yeah.
输入文字。
Enter characters.
是的。
Yeah.
那边用起来肯定舒服多了。
It's gotta be much nicer over there.
如果你是中国用户,或者在中國用过iPhone,也欢迎分享一下你的体验。
And if actually you have any experience as a Chinese person, or someone who has used the iPhone in China, would love to get your feedback on that too.
明白。
Yeah.
但到目前为止,很难说语音助手带来了直接的回报,除了他们长远来看本来就得做这件事。至于收购Siri公司,我倒不确定是不是最好的方式。
But, you know, I think to this point, it's hard to point to ROI other than it was something they sort of needed to do long term, and I don't know that actually acquiring the company Siri was the best way to do it.
我给这个打个C吧。
I'm gonna give this one a c.
没错。
Yep.
我觉得这次收购特别典型,就是这几年苹果的风格。
So I think, this acquisition was just like so classically Apple or at least Apple of the last few years.
花不到五亿美元收购一家科技公司,然后把它融入自己的平台,成为整个产品体系的一部分。
Spend south of $500,000,000 to acquire a technology company, that you then use to build into part of your platform, and your whole product platform.
我想到了Authentic和指纹识别。
I'm thinking about, you know, Authentic and the fingerprint.
这收购真棒。
Great acquisition.
这收购真棒。
Great acquisition.
我们可以在未来的节目中聊聊这个。
We should do that in a future episode.
这家公司,哦,名字我一时想不起来了,就是他们收购的那家半导体公司。
The company oh, I'm blanking on the name that they bought that was the semiconductor company.
PA Semi。
PA Semi.
对。
Yeah.
PA,PA Semi。
PA PA Semi.
这笔投资回报可太可观了。
And God has that paid dividends.
天哪。
Oh, man.
他们现在应该是全球产出效率最高的ARM芯片制造商了吧。
They're they're I mean, they're are they the highest yield I think they're the highest yield producing arm chip manufacturer in the world now.
更别提现在了
And Not to mention, like, now
苹果自己设计芯片。
Apple designs their own chips.
没错。
Yeah.
他们能自主开发系统级芯片,还能把芯片和操作系统深度整合,这才实现了像触控ID这样的功能——专门有个独立的安全模块,和主处理器隔开。嗯,我们迟早也会做PSMI。
And the fact that they can do their own system on a chip, and the fact that they tie it directly into their operating system, they're able to do things like Touch ID where they have that secure enclave that sits separately than the processor and I mean, we'll do PSMI too at some point.
好多厉害的地方。
So much good stuff.
是的。
Yeah.
但如果把这些功能排个先后,很遗憾,Siri 排得挺靠后的。
But and when you stack rank those, like, unfortunately, Siri falls pretty far down the list.
我觉得给它打个 B 吧,毕竟语音和人工智能对苹果未来太重要了。
So I think I'd give it a I'm gonna give it a b because I think it is, I'm giving it a b because future importance of voice and AI is going to be huge for Apple.
要是他们没在那时候起步,现在谷歌、脸书,甚至微软早就把他们甩开好几条街,根本追不上了。
And if they hadn't done this, if they hadn't started when they did, there's just like Google and Facebook and and maybe even Microsoft would be so far beyond them at this point they could never catch up.
不管他们花多少钱,听起来都像是……
No matter how much they would spend Sounds
没错。
like that.
或者他们收购了谁。
Or who they would buy.
是的。
Yeah.
不过前几天我看到一个数据,说地图应用的使用量和体验都在提升。
Although I just saw a stat the other day that, like, maps is getting better and getting more usage.
对,确实用的人越来越多了。
Well, yeah, it's getting more usage.
因为它本来就预装在系统里。
It ships with the OS.
没错。
Well, right.
我有一阵子没用过了。
I haven't tried it in a while.
我也没遇到过什么糟糕的体验。
I didn't had a bad experience.
不过我老听人说它在变好。
I I keep hearing it's getting better though.
可是呢,
So But
我们还是不用它。
we still don't use it.
Siri 也是同样的情况。
So same with Siri.
好吧。
Alright.
对我来说,选B。
B for me.
你选B,我选C。
B for you, c for me.
我觉得就这些了。
I think that's that's all we
我们就这些了。
That's all we got.
大家节日快乐。
Happy holidays, everybody.
我们2016年再见。
We will see you in 2016.
Siri,你有什么话想对听众说吗?
Siri, do you have anything to say to the listeners?
抱歉,本。
I'm sorry, Ben.
恐怕我无法回答这个问题。
I'm afraid I can't answer that.
谁掌握了真相?
Who got the truth?
好了,各位听众。
Alright, listeners.
现在是个绝佳的机会,来聊聊我们特别喜欢的一家公司——Statsig。
Now is a great time to talk about one of our favorite companies, Statsig.
没错。
Yes.
我们长期的合作伙伴。
Longtime acquired partner.
像OpenAI、Notion、Atlassian、Figma、Rippling、Brex这些公司的顶尖产品团队,无论是在迭代核心功能,还是大规模推出AI驱动的体验,都离不开Statsig,这背后自然有原因。
There is a reason why the best product teams at companies like OpenAI and Notion, Atlassian, Figma, Rippling, Brex, and more rely on Statsig, whether they are iterating on their core product features or shipping AI powered experiences at scale.
对。
Yep.
在当今AI领域飞速发展的环境下,快速迭代已经成了基本要求。
In the crazy speed of today's AI world, shipping fast is just table stakes now.
构建和持续部署你的应用其实非常简单。
It's basically trivial to build and deploy your app constantly.
真正的优势在于,你能多快地发现哪些改动真正为用户带来了价值,又多快地用这些数据来指导下一步的开发。
The real advantage is how quickly you learn what changes actually created value for customers and how fast you can use that signal to guide what you ship next.
无论是功能微调、价格变动、性能优化,还是像模型更新或提示词调整这样的AI改动,他们都不靠直觉做决定。
Whether it's a feature tweak, a pricing change, a performance improvement, or an AI update like a model change or prompt adjustment, they're not relying on instinct.
他们真正测量的是哪些改动推动了用户参与度、留存率,最终提升了收入。
They're measuring what actually moved engagement, retention, and ultimately revenue.
随着越来越多团队开始用AI开发,这种快速学习的循环变得愈发关键。
And as more teams build with AI, that learning loop becomes even more important.
使用大语言模型会让产品体验变得不确定。
Building with LLMs introduces nondeterminism into your product experience.
同样的输入不一定产生同样的结果,实际使用中,行为可能会悄然发生变化。
The same input doesn't always produce the same output, and behavior can shift in subtle ways in real world use.
离线评估能帮你看到一部分情况,但只有产品上线、面对真实用户后,你才能真正看清这些改动如何影响了用户行为。
So doing offline evals will give you part of the picture, but you can really only understand the impact once your product is live with real users, and then you can measure how their behavior actually changes.
这和以前没有AI的时候完全不一样,那时候你很清楚软件会怎么运行
It's very different than the way that you would ship features in a pre AI world where you knew exactly what the software was gonna do
在上线后。
in production.
是的。
Yeah.
没错。
Exactly.
这时候Statsig就派上用场了。
So this is where Statsig comes in.
它把实验、功能开关和产品分析整合到一个统一的系统里,让团队能安全发布、严谨测试,并直接把改动和用户实际行为联系起来。
It brings experimentation, feature flags, and product analytics into one unified system so teams can ship safely, test rigorously, and directly link what they changed to how users actually behaved.
结果就是反馈循环更紧密,经验不断积累,你不仅发布得更多,还发布得更好。
The result is a tighter feedback loop and learning that compounds over time, so you don't just ship more, you ship better.
所以,如果你想把学习能力变成你的竞争优势——无论你是打造全新的AI体验,还是只是优化现有的核心产品——请访问 statsig.com/acquired 开始使用。
So if you wanna make learning your competitive advantage, whether you're building new AI experiences or just evolving your existing core product, go to statsig.com/acquired to get started.
关于 Bayt 播客
Bayt 提供中文+原文双语音频和字幕,帮助你打破语言障碍,轻松听懂全球优质播客。