Experimentation × AI: Honest Takes with Elena Verna

Hello, everyone.

大家好。

Welcome to the keynote session of Test and Learn 2026.

欢迎来到 Test and Learn 2026 的主题演讲环节。

I'm Rupali Jain, the chief product officer here at Optimizely.

我是 Rupali Jain,Optimizely 的首席产品官。

I'm super excited to be hosting Elena Verna for some candid takes.

今天能主持 Elena Verna 的对谈、听她讲些毫不客套的观点,我特别兴奋。

On how AI is changing experimentation, and what we need to do to stay ahead.

聊聊 AI 正在如何改变实验,以及我们要怎么做才能保持领先。

If you were with us last year, you will remember that Elena's session was one of the most impactful of the event.

如果你去年也在场,一定还记得 Elena 那场分享是全场最有冲击力的环节之一。

So of course, we had to invite her back.

所以我们当然得再把她请回来。

Elena's one of the most respected voices in growth.

Elena 是增长领域最受尊敬的声音之一。

She's led growth at SurveyMonkey, Miro, and created some of the most used Reforge courses on experimentation, monetization, product led growth.

她曾在 SurveyMonkey 和 Miro 负责增长,还在 Reforge 上创作了关于实验、商业化、产品驱动增长的几门最热门课程。

And is currently the head of growth at Lovable, one of the fastest growing AI companies out there, and also one of the most lovable.

她现在是 Lovable 的增长负责人——那是当下增长最快的 AI 公司之一,也是最"可爱"的公司之一。

She's also an advisor and investor across some of the most exciting companies in tech.

她同时还是科技圈多家最令人期待的公司的顾问和投资人。

What I know about Elena is she does not do safe takes.

我对 Elena 的了解是:她从不讲四平八稳的观点。

Let's see what learning she has in store for us today.

来看看她今天给我们准备了什么新知。

Elena, welcome back.

Elena,欢迎回来。

It is fantastic to have you here.

你能来真是太棒了。

Hi, everybody.

大家好。

So nice to be back.

很高兴又回来了。

Very excited for today's session.

非常期待今天的环节。

That's amazing.

太好了。

Before we jump into the meat, and believe me there's a lot of that, I figured I'd start with a quick icebreaker.

在进入正餐之前——相信我,今天干货管够——我想先来个简短的破冰。

And you know we talk about learning.

你知道,我们今天聊的主题就是"学习"。

So, maybe outside of work, Elena, what is it that you're personally excited about, that you're learning?

那么 Elena,工作之外,你个人最近在学什么、对什么特别来劲?

I always love to have additional hobbies outside of work.

我一直喜欢在工作之外多搞几个爱好。

As so, for a little bit, I was very into sourdough in the true millennial fashion.

所以有一阵子,我特别迷做酸面包,标准的千禧一代做派。

But I have moved on from that, uh, to start making lattes at home.

但我已经从那个坑里出来了,呃,开始在家做拿铁。

So, I have a fancy machine that I just bought for my birthday.

我刚给自己买了台很高级的机器,当生日礼物。

And uh oat milk lattes is my experimentation right now.

呃,燕麦奶拿铁就是我现在的实验项目。

I'm almost at the coffee shop level.

我差不多已经做到咖啡店水准了。

Oh wow, I might have to come visit you.

哇,那我可能得登门拜访一下了。

I do like my coffee drowned in chocolate, for what it's worth.

顺带一提,我喜欢咖啡被巧克力淹没的那种喝法。

Oh! So switching topics, and let's do a hard transition to AI, right?

哦!那换个话题,我们硬切到 AI,好吗?

It's changing how we need to approach growth, it is changing how we work.

它在改变我们做增长的方式,也在改变我们的工作方式。

Execution cost is going down, and every channel that I see is flooded with content.

执行成本一路走低,我看到的每一个渠道都被内容淹没了。

A lot of traditional growth tactics as well don't really work that well anymore today.

很多传统增长战术如今也不太管用了。

Super curious to hear your take on what exactly is working and what are you seeing out there.

特别想听听你的看法:现在到底什么在起作用?你都看到了些什么?

Yes, uh, I actually made the statement that anywhere from 60-80% of growth tactics are no longer working.

是的,呃,我确实说过:60-80% 的增长战术已经不再有效。

What do I mean by no longer working?

"不再有效"是什么意思?

uh, Maybe that was a little bit too provocative, but what I really mean is that how we used to grow our product previously is not really how we win any more in this day and age.

呃,也许那个说法有点太挑衅了,但我真正想说的是:我们过去做产品增长的那一套,在这个时代已经不是赢的方式了。

And there's a couple of reasons for that.

这背后有几个原因。

Number one, there is a very low feature differentiation that we can rely on as the reason to win in the market.

第一,如今能拿来当市场制胜理由的功能差异化已经非常少。

That's really impacts marketing quite a bit, because with the speed of velocity of development really accelerating, all thanks to AI.

这对市场营销的冲击相当大,因为开发的速度和节奏都在飞快提升——这全拜 AI 所赐。

Because let's face it, in AI native companies eighty to ninety percent of the code is written by AI.

因为说实话,在 AI native 公司里,80% 到 90% 的代码是 AI 写的。

Now that fact alone really increases velocity of how quickly you can add any given feature to your product.

光这一点,就大大加快了你往产品里加任何一个功能的速度。

So, feature differentiation is much shorter lived nowadays, and you cannot rely on it as the reason that customer should be picking your product.

所以功能差异化如今的存活期短得多,你不能再指望它成为客户选你产品的理由。

Four years, and which we're used to, uh, versus now, it might be only a month or two or three that you can actually do that.

以前我们习惯的是四年,呃,而现在,你可能只有一个月、两三个月还能吃到这个红利。

But more importantly, is that all of our growth channels, such as search, social, uh, the way that we used to do them, are becoming more of a baseline tactics that you have to do.

但更重要的是,我们所有的增长渠道——比如搜索、社交,呃,按我们过去的做法——正在变成一种你必须做的基线动作。

And you have to still do them, there's still a lot of traffic and growth can come from them, but they're not gonna be the reasons that you're gonna win.

这些你还是得做,它们仍然能带来不少流量和增长,但它们不会是你赢的原因。

And that is important to understand, because if you actually want to pull ahead of the pack, you cannot go with your corporate social account and drive a ton of engagement.

理解这一点很重要,因为如果你真想甩开大部队,靠公司官方社交账号是刷不出多少参与度的。

You cannot create a lead ma- ma- magnet out there, or a pdf to download, and expect that there is gonna be massive growth coming from it.

你也不能做个引流……引流磁铁(lead magnet),或者放个 PDF 让人下载,然后指望它带来巨大增长。

You have to reinvent yourself.

你必须重塑自己。

You have to be a lot more vulnerable in the market.

你必须在市场上更敢于袒露自己。

You have to build in public.

你必须 build in public。

You have to deploy your employee-led socials.

你必须铺开由员工个人主导的社交账号。

You have to lead with trust.

你必须以信任为先。

I really think that growth nowadays is a trust problem, more so than it is a tactical execution problem on this very specific channel.

我真心认为,如今的增长更多是一个信任问题,而不是某个具体渠道上的战术执行问题。

It's super super interesting, right?

这太有意思了,对吧?

Like there's a couple of key terms that I picked up here, and reinvent yourself is one of the big ones.

我从里面抓到了几个关键词,"重塑自己"是最重的一个。

It's sort of like you talk about the pace, and if you don't reinvent yourself, it's almost like you're dead anyway in the water, right?

就像你说的那个节奏——如果你不重塑自己,几乎就等于原地等死,对吧?

So, and the leading with trust such an important point on growth.

所以,还有"以信任为先",这对增长来说是特别重要的一点。

But let me pick up on the thread that you started with, the point about velocity and the point about difference in the market right?

但让我接着你开头那条线聊:关于速度、关于市场差异化的那一点,对吧?

Because yes, we can come up with the capability today.

因为没错,我们今天就能做出某个新能力。

Moats have shrunk, and everything on differentiation is getting really really tight.

护城河缩水了,差异化的空间被挤得非常非常窄。

So it's the creativity and the velocity that matters.

所以真正重要的是创造力和速度。

And sort of like, you know, when you think about AI native companies, and you've talked about this concept of being on the product market fit treadmill, when you're moving this fast, and it's amazing, and don't get me wrong, I don't wanna slow things down.

然后就是,你知道,想想 AI native 公司——你也讲过"PMF 跑步机"这个概念——当你跑得这么快的时候,这很了不起,别误会,我并不想让节奏慢下来。

How do you think about the balance between running experiments, wh- versus just taking a bet and going for it?

你怎么权衡"跑实验"和……和"直接下注、放手去干"之间的平衡?

Because that's, it's a problem that is now surfacing, and it's a real challenge that we see.

因为这是一个正在浮出水面的问题,是我们看到的一个真实挑战。

Yeah, so let me explain what I mean by product market fit treadmill first.

好,那我先解释一下我说的"PMF 跑步机"是什么意思。

um, When say 10 years ago, when you'd come up with a product market fit for your product, which is there is a problem, you have a solution, and customers really like the solution, they're enjoying it.

嗯,比如说 10 年前,当你为产品找到了 PMF——也就是存在一个问题,你有一个解法,而客户真心喜欢这个解法,用得很开心。

they're retaining with it, they're continuously solving their problem.

他们持续留存,不断用它解决自己的问题。

You could ensure that your company could survive on that product market fit, and kind of almost go into the growth blitz stage, I would say, on that product market fit.

你可以确信公司能靠这个 PMF 活下去,甚至可以说,几乎能直接吃着这个 PMF 进入 growth blitz 阶段。

Four years, uh, because uh there's gonna be enough product differentiation for you, there's enough uh channel coverage that you can have,

整整四年,呃,因为呃你会有足够的产品差异化,有足够多呃可以覆盖的渠道,

and you can have a really good scaling journey until you looking for your second horizon.

你可以走过一段非常顺的规模化旅程,直到你开始寻找第二增长曲线。

which usually companies would take 5-10 years to have to enter to a second horizon of innovation of the product market fit.

通常公司要过 5-10 年,才需要进入 PMF 创新的第二条曲线。

Well, uh, unfortunately those timelines have been massively compressed.

可是,呃,不幸的是,这些时间线已经被大幅压缩了。

At Lovable, we're only a year and a half years old.

Lovable 成立才一年半。

Yet, I would say that we have to reinvent our product market fit every month to two months.

但我会说,我们每一到两个月就得把 PMF 重造一遍。

And even though we already pass 400 million in ARR, which we shared in February this year.

而且即便我们已经跨过了 $400M ARR——这是今年 2 月我们公开的数字。

I don't think we have product market fit in a bag, and it's happening because of a couple of things.

我也不认为 PMF 已经稳稳揣进了口袋,而这背后有几个原因。

Number one, customer perceptions are changing like we've never seen before.

第一,客户的认知正在以前所未见的速度变化。

Customer expectation of what product is capable of producing.

客户对产品能做出什么的预期。

Is changing every thirty days, I would say, and they expect more and more something that used to take years to accomplish.

我会说每 30 天就变一轮,他们越来越期待那些过去要花好几年才能做成的东西。

um, Because customers were so happy with the existing solution, and the spe- speed of innovation.

嗯,因为客户已经被现有解法、被这种创……创新速度惯坏了胃口。

Number two, technology is changing.

第二,技术在变。

What AI is capable of doing, i- is increasing all the time.

AI 能做到的事,正……正在不断增加。

So your product needs to keep up to capture that advantage, as opposed to just riding on what AI was able to do, let's say six months ago.

所以你的产品必须跟上去、抓住新的技术优势,而不是躺在 AI 半年前的能力上吃老本。

We look at the six months ago models, and we go, oh my gosh, they were so terrible.

我们回头看六个月前的模型,会说:天哪,它们怎么这么糟糕。

So you constantly have to recapture that new technological advantage.

所以你必须不断地重新夺取新的技术优势。

And then most importantly too, your distribution constantly changes, because you don't have these long stages of growth blitz anymore.

而且最重要的是,你的分发也在不断变化,因为你不再拥有那种长长的 growth blitz 阶段了。

You have to constantly readjust to your new product market fit.

你必须不断重新校准到你新的 PMF 上。

And adjust all of your growth strategy.

并且把你全部的增长策略都跟着调整一遍。

So growth work is a lot more stop and go, stop and go, stop and go.

所以增长工作变得更像走走停停、走走停停、走走停停。

uh, Because uh you need to constantly keep up with the product, and product has never evolved at this velocity before.

呃,因为呃你得不断跟上产品的脚步,而产品从来没有以这样的速度进化过。

It's incredible to see the product innovation velocity, and it's both exhilarating and terrifying at the same time when I see it right?

这种产品创新速度令人难以置信,我看着它的时候,既兴奋又害怕,对吧?

So, yes.

所以,是的。

When and when you think about experimentation in this world, Elena, like how do you think about the balance of sort of like You know, let's just go for it, versus let's go run an experiment, validate, and then go for it.

当、当你思考这个世界里的实验时,Elena,你怎么权衡那种,你知道,"直接上"和"先跑个实验、验证完再上"之间的平衡?

Because I think there's certain places where experimentation. still has its place.

因为我觉得在某些地方,实验……依然有它的位置。

And there are some things that I'm sure you just tell all of us, just take a bet and go.

而有些事,我相信你会直接告诉我们所有人:下注,然后冲就完了。

How do you, how do you balance that?

你怎么、你怎么平衡这个?

Yeah, so I would make two points on this.

好,关于这个我想讲两点。

Number one, my entire career I used to focus on optimizations of existing core product market fit as the main avenue of how I provided value to customers and to the business.

第一,我整个职业生涯都在围绕已有的核心 PMF 做优化,把它当作我为客户和业务创造价值的主要途径。

And that was to focus on those onboarding optimizations, activation, conversion rate, retention, uh acquisition.

具体就是盯着 onboarding 优化、激活、转化率、留存,呃,还有获客。

But it was trying to squeeze that like that last 3~10 percent of uh, the value.

但那其实是在挤,就像挤价值里,呃,最后那 3-10%。

Obviously sometimes you get bigger wins, and it'll be like, oh my gosh, I got like 20 percent or 30 percent lift.

当然,有时候你会拿到更大的胜利,那种感觉就是:我的天,我拿到了 20% 或 30% 的提升。

But most of the time, just a smaller tweaks across the entire user journey were enough to deliver a ton of value back to the business.

但大多数时候,在整条用户旅程上做些小调整,就足以给业务带回大量价值了。

Well, the problem right now is that those type types of tiny optimizations take quite a bit of time and effort in order to actually tweak it and find the right um user journey.

但现在的问题是,这类微小优化要真正调到位、找到对的,嗯,用户旅程,需要花相当多的时间和精力。

The issue with what's happening right now is number one, you don't have that time.

而眼下正在发生的事情,麻烦第一在于:你没有那个时间。

You have maybe a month, maybe three months to optimize existing product.

你可能只有一个月、最多三个月,去优化现有产品。

So you don't have even enough traffic, and even we at Lovable, I mean we have so much traffic, we have thousands of users creating accounts every single day.

所以你甚至连足够的流量都没有——哪怕是我们 Lovable,我是说我们流量已经这么大了,每天有几千个用户注册账号。

But I don't have traffic to run a bunch of different optimizations.

但我依然没有流量去跑一堆不同的优化实验。

So, I have to pick uh very carefully of what is it that I am actually gonna optimize.

所以我必须,呃,非常谨慎地挑:我到底要优化什么。

And if I have to pick and prioritize, I'm gonna go for bigger bet.

而如果必须挑、必须排优先级,我就会去下更大的注。

Because uh smaller things, smaller tweaks just don't matter, because by the time I go to them being a really good value, uh, the product changes already, so it doesn't really make sense.

因为,呃,小东西、小调整根本无关紧要——等我把它们打磨到真正有价值的时候,呃,产品早就变了,所以没什么意义。

So I would say bigger bets are much uh more important and valuable in optimization, as opposed to doing smaller copy tweaks or button color tweaks, like all of that has completely fallen off.

所以我会说,在优化里,更大的赌注,呃,重要得多、有价值得多,而不是去改改文案、调调按钮颜色——那一套已经彻底失效了。

And then number two, I would say is that I experiment so much more on monetization than I've ever have in my entire career.

然后第二点,我想说,我现在在商业化上做的实验,比我整个职业生涯任何时候都多得多。

I've always owned monetization models, and there's always been pricing, page testing, or some sort of retention, renewal, um messaging, or workflow optimizations.

我一直都负责商业化模型,以前也总有定价、页面测试,或者某种留存、续费的,嗯,文案或流程优化。

But right now, especially in AI companies over the AI features, the cost of AI is so expensive, and monetization friction comes so early in users' journey, because of those cost dynamics, if you wanna create predictable, sustainable company, obviously.

但现在,尤其在 AI 公司、在 AI 功能上,AI 的成本太贵了,而正因为这种成本结构,商业化摩擦在用户旅程里出现得特别早——当然,前提是你想做一家可预测、可持续的公司。

That you have to immediately start tweaking your monetization, and that is a big difference, because A, a lot of optimization doesn't touch monetization, cause it's like a scary topic still for a lot of people.

那你就必须立刻开始调你的商业化,这是个很大的不同——因为第一,很多优化根本不碰商业化,对很多人来说它至今还是个吓人的话题。

So, they keep those models static.

所以他们就让那些模型一直静止不动。

But in this day and age, actually most of the monetization work ends up being experimentation focus, because you constantly need to evolve those customer expectations, those technological changes, and just like your product, um functionality as a whole.

但在当下这个时代,其实大部分商业化工作最后都是以实验为核心的,因为你得不断跟着客户预期、技术变化,以及,嗯,你产品的整体功能一起演进。

and maybe I'll just add one bonus one, is that there's so much more experimentation that now happens on agentic instruction level, as opposed to actual UI surfaces.

或许我再附送一点:现在有多得多的实验,发生在 agentic 指令层,而不是实际的 UI 界面上。

Because all of our UIs are starting to get collapsed into a prompt box.

因为我们所有的 UI 都开始坍缩成一个提示框。

uh, Regardless of where you are in that journey, it is getting there.

呃,不管你在这条路上走到哪一步,它都正在朝那儿去。

It's not gonna be just feature pages that we need to drive awareness and discoverability and usability of.

未来不再只是我们要为功能页面去拉认知、拉可发现性、拉可用性。

It's all about ask an agent what is it gonna give you as an output.

而是全看你问 agent 一句,它会给你什么样的输出。

So a lot of experimentation starts to become, how do you have that agent be most helpful to user?

所以很多实验开始变成:你怎么让那个 agent 对用户最有帮助?

So it's coding instructions and context for that agent, as opposed to doing UI optimizations um in the first place.

也就是给那个 agent 写指令、写上下文,而不是,嗯,一上来就做 UI 优化。

It's so true.

太对了。

I mean, you know, monetization and pricing models haven't changed at the pace they are changing right now ever in my career at least, and it's a long one.

我是说,你知道,至少在我的职业生涯里,商业化和定价模型从来没有像现在变得这么快——而我的职业生涯可不短。

We won't age me right now, but hahaha, that's what it is.

我们现在就不暴露我的年龄了,哈哈哈,反正就是这么回事。

But it's sort of like really interesting, and I think it ties back to the point you made earlier about product market fit, the life cycle.

但这真的挺有意思,我觉得这又绕回到你之前讲的 PMF 那个点,那个生命周期。

Even at 400 million, that's taken you a year and a half.

哪怕做到 4 亿,你们也只花了一年半。

This would take like 5, 6, 7, 10 years in the past,

这在过去得花 5 年、6 年、7 年、10 年,

and And my position is and they're collapsing right now, the cycles, like one month at Lovable, I feel like is a normal one year at other company, and we have to do it all together.

而、而我的看法是,这些周期现在正在坍缩——比如在 Lovable 的一个月,我感觉就等于其他公司正常的一年,而且我们得把所有事同时干完。

Yeah.

对。

And it's such a rich surface. like, because as user behaviors change, what they expect, what the bar is, what the competitive landscapes doing to you.

而且这是一块特别肥沃的阵地。就是,因为用户行为在变,他们的预期在变,及格线在变,竞争格局也在冲击你。

It's such a rich surface for testing.

这是一块做测试的肥沃阵地。

I think one of the big learnings I've had in my career is It's okay to have a pricing model, it's okay to change it.

我职业生涯里最大的心得之一是:有一个定价模型没问题,把它改掉也没问题。

It's like as long as you're giving your customers the right behaviors and expectations for whenever they signed up, and so there's certain things around Monetization that holds true, but it doesn't mean that we can't test on pricing models to your point about not being precious about it.

就是说,只要你给客户的行为和预期,对得起他们注册那一刻的约定——商业化里确实有些原则始终成立,但这不代表我们不能在定价模型上做测试,正如你说的,别把它当宝贝供着。

I think that we've always been too conservative with our pricing models, uh, and we had privilege to be Because uh they were pretty static user base, maybe some usage based uh platform fees, whatever it was.

我觉得我们在定价模型上一直太保守了,呃,而且我们以前有保守的资本——因为,呃,那时候用户基础相当稳定,顶多加点按用量计费的,呃,平台费之类的东西。

But LLMs really changes margin equation, really changes cost equation for companies.

但 LLM 真的改变了利润率的算式,真的改变了公司的成本算式。

So in order to Rebuild sustainable businesses, we need to really figure out how is it that we actually monetize AI.

所以要重建可持续的生意,我们得真正想清楚:到底该怎么给 AI 做商业化。

I actually don't think anybody has figured it out yet.

我其实觉得,还没有任何人想明白这件事。

uh, I look at Lovable's monetization model, um, it works right now, uh, but do I think that that's what it's gonna be a year from now?

呃,我看 Lovable 的商业化模型,嗯,它现在是管用的,呃,但要问我一年后它还会是这个样子吗?

Absolutely not.

绝对不会。

I hope that we find a way to move away from uh being more covering the cost of LLM and more towards output base,

我希望我们能找到一条路,从,呃,主要为覆盖 LLM 成本的模式,转向更基于产出(output-based)的模式,

and I truly believe the companies That are set up to experiment and evolve their monetization models towards what the ultimate output base model is gonna look like, are gonna be the winners in a space.

而且我真心相信,那些有能力不断实验、让自己的商业化模型朝着最终那个基于产出的形态演进的公司,会成为这个领域的赢家。

And then. the second piece uh that I'd also say on monetization, it's extremely counterintuitive to what we've always done.

然后,关于商业化我还想说第二点,呃,它跟我们一贯的做法极其反直觉。

But when you say, or when I say that We need to test on monetization.

但当你说,或者当我说"我们要在商业化上做测试"的时候。

I actually don't say optimize conversion rate.

我说的其实不是优化转化率。

I don't actually mean optimize um revenue or average revenue per user.

我指的也不是优化,嗯,收入或 ARPU。

What I actually mean is to ungate engagement.

我真正的意思是:解锁参与度。

And put your monetization in a way that it's not blocking adoption from customers.

把你的商业化摆在一个不会挡住客户采用的位置。

Because we are really trying to change habits of our customers.

因为我们其实是在试图改变客户的习惯。

We're trying to introduce completely new workflows.

我们在引入全新的工作流。

And so many companies I see make a mistake of taking their AI feature and just automatically slotting it into a highest paid you uh package that they have on their pricing page.

我看到太多公司犯同一个错:拿到自己的 AI 功能,就自动把它塞进定价页上最贵的,呃,那档付费套餐里。

Because it's so high cost, and finance is trembling in the background, because um they want to maintain those Cushy 80 90 percent SaaS margins,

因为成本太高了,财务在后面瑟瑟发抖,因为,嗯,他们想守住那种舒舒服服的 80%、90% 的 SaaS 毛利率,

and that are going away unfortunately, or maybe fortunately, because um that created a lot of slack in the system that uh is not. Not there anymore, so we have to work a little bit harder to fulfill our customers' expectations.

而这种毛利率正在消失——说不幸也不幸,说幸运也幸运,因为,嗯,它以前给整个系统留了太多松弛空间,而那些空间,呃,已经、已经不在了,所以我们得更卖力一点,才能满足客户的预期。

But you have to actually give a lot of your product away.

但你真的必须把产品的很大一部分白送出去。

And how do you give it away that creates good roi for you, return on investment.

而怎么个送法,才能给你带来好的 ROI,投资回报。

How do you give it in the most strategic way, at the right place, at the right time, to a right customer?

怎么用最有策略的方式,在对的地方、对的时间,送给对的客户?

Is where it's all about to drive that engagement.

这才是拉动参与度的全部关键。

So, when I say testing monetization model for us at Lovable, it actually means we're optimizing towards our north star metric of daily active ops.

所以,当我说我们 Lovable 在测试商业化模型时,实际上意味着我们在朝着我们的 north star metric(北极星指标)——daily active ops——去优化。

And how can we grow that?

以及我们怎么把它做大?

And monetization, and revenue is just an outcome that comes out of it, but it's all about unblocking usage.

而商业化、收入,只是随之而来的结果——一切核心都是把使用的阻碍解开。

Yeah, I think there's a couple of Very very strong nuggets in there right.

对,我觉得这里面有好几个非常非常硬的干货,对吧。

One is it's about adoption and usage, and about the history.

第一个是:关键在于采用和使用,还有那段历史。

And this is one that holds true, is that if you get adoption, you get the money at the end.

而这一条是始终成立的:只要你拿下采用,钱最终会来的。

Search started that way.

搜索就是这么起家的。

SaaS companies actually before they went to the 90 percent uh margin models also started that way, and somewhere along the way we got distracted.

SaaS 公司其实在走到 90%,呃,毛利率模型之前,也是这么起家的,只是走着走着我们分心了。

And I think the second point you made that I really really wanna underscore for our audience, because I'm seeing this as a product leader, is it was easy to think about engagement as a proxy for Outcome metrics and north star metrics, and so that diligence hasn't always been there.

你讲的第二点,我特别特别想给我们的观众划重点,因为作为产品负责人我正亲眼看到这一点:大家很容易把参与度当成结果指标和北极星指标的替身,所以那份严谨并不总是在的。

And your point about organization needs to. help you achieve your north star metric is absolutely true, and you also need to know what your north star metric is.

你说组织需要……帮你达成北极星指标,这绝对没错,而且你还得先知道自己的北极星指标到底是什么。

And so that's one thing that I would encourage all of us to really really hone in on, cause it's so critical in today's day and age, and it is non optional, because you also made the point about agents are becoming the consumers, right?

所以这是我鼓励我们所有人真正、真正去死磕的一件事,因为在今天这个时代它太关键了,而且没得选——因为你也提到了,agent 正在成为消费者,对吧?

It's not just users.

不再只是用户。

And so, good might look like there's no users actively touching your product.

所以,"好"的样子,可能是根本没有用户在主动碰你的产品。

Or humans, no humans are actively touching your product.

或者说人类——没有人类在主动碰你的产品。

No humans, humans versus agents.

没有人类,是人类 vs agent。

Yes, I should make that distinction.

对,我应该把这个区分讲清楚。

Yeah, so super interesting.

嗯,特别有意思。

um, Let me switch tracks on you, Elena, a little bit, and let's talk about the type of organization it takes, and you know, the how we work so Maybe compare a little bit of your experience pre Lovable, and what does it take to truly become an AI native organization?

嗯,Elena,让我稍微给你换个赛道,我们来聊聊这需要什么样的组织,还有,你知道,我们的工作方式——或许可以对比一下你来 Lovable 之前的经历:要真正成为一个 AI native 的组织,需要什么?

What. has to change?

什么……必须改变?

Like we're all carrying baggage.

就是说,我们身上都背着包袱。

What baggage do we have to jettison?

哪些包袱是我们必须扔掉的?

I think every person right now has to define for themselves, so what does it mean to be AI native for them?

我觉得现在每个人都得自己去定义:对自己来说,AI native 到底意味着什么?

There's multiple ways that you can engage with being AI native.

实践 AI native 有好几种方式。

You can just help it with writing.

你可以先让它帮你写东西。

uh, That's like a very basic one, and everybody should be doing.

呃,这算是最基础的一种,每个人都应该在做。

Me as li-, uh, English is my second language, so I really heavily lean on the AI to help me clear up my writing, to make the point, uh, to fix my grammar, so I can communicate.

我作为——呃,英语是我的第二语言,所以我非常依赖 AI 帮我理顺文字、把观点讲清楚,呃,修正语法,让我能好好沟通。

As communication is a really big portion of our success uh in the professional world.

毕竟在职场上,沟通是我们成功里呃很大的一块。

Number two, it can be your thinking partner.

第二,它可以当你的思考伙伴。

So use AI to challenge your ideas, to pressure test them, to give you other options.

用 AI 来挑战你的想法、给它们做压力测试、给你别的选项。

I love using AI for that.

我特别喜欢这么用 AI。

It doesn't come up a final answer for me, but it does give me an opportunity to rethink.

它不会替我得出最终答案,但它确实给了我一个重新思考的机会。

How I am uh putting the pieces together, it gives me additional inputs to consider.

思考我呃是怎么把这些拼图拼起来的,它会给我更多可以纳入考虑的输入。

It can do a deep research for me so then I can make a better decision.

它还能帮我做深度研究,让我做出更好的决策。

Number three, I would really recommend you using it for recording all of your meetings and giving you feedback on your uh, Presentation style, on your communication style.

第三,我真心推荐你用它录下你所有的会议,让它就你的呃表达风格、沟通风格给你反馈。

I use Granola for example, and I constantly ask it, hey, if you were my manager, what feedback would you give me?

比如我用 Granola,我会不停地问它:嘿,如果你是我的 manager,你会给我什么反馈?

It is such an incredible self-improvement tool that you can do, that is very direct, and has a ton of evidence to back up its decisions, which I'm like, wow, I the best manager that I may have ever had.

这是一个特别不可思议的自我提升工具,非常直接,而且它的判断背后有一大堆证据撑着,我当时就觉得,哇,这可能是我这辈子遇到过最好的 manager。

um, in terms of continuous feedback.

嗯,就持续反馈这一点来说。

But then most importantly, learn how to use AI to start building and shipping.

但最重要的是,学会用 AI 去动手构建、去 ship 东西。

This is where truly AI nativeness starts to come in, where it's not just helping your thought process or helping you show up better in your company.

真正的 AI native 从这里才开始——它不只是辅助你的思考过程,或者帮你在公司里表现得更好。

It is where you actually can start using AI to change the product to impact the direction of your business.

而是你真的可以开始用 AI 去改产品,去影响你业务的走向。

So, I use AI to ship code to production.

所以,我用 AI 把代码 ship 到生产环境。

I'm a non-technical person, and yet I commit uh to production code.

我是个非技术背景的人,但我照样呃往生产环境提交代码。

And that's absolutely incredible.

这简直不可思议。

There is a skill that I've unlocked that just still feels surreal to me, even though I've been doing it for six months plus.

我解锁了一项技能,到现在都还觉得很超现实,虽然我已经这么干了六个多月。

Or I use Lovable to build a lot of apps or internal tools for the company or prototypes even uh to help push are in the next direction.

或者我用 Lovable 给公司做一堆应用、内部工具,甚至原型,呃来帮着把事情往下一个方向推。

So, really get in there and get uh get your hands dirty.

所以,真的要扎进去,呃,亲自下场动手干。

But one last thing, all of this will only work if you can unlock trust in your organization.

但最后一点:所有这些,只有在你能解锁组织内部的信任时才行得通。

Because one of the benefits of AI nativeness is how much a person is able to accomplish on their own without cross functional collaboration.

因为 AI native 的好处之一,就是一个人不靠跨职能协作也能独立完成非常多的事。

But the only way that that person will be able to go in the right direction and do the right things, if they have all of the right context in organization for your strategy, for your product direction. for things that you hoping to accomplish.

但这个人要想方向正确、做对的事,唯一的前提是:他在组织里掌握全部正确的上下文——你的战略、你的产品方向、你想达成的目标。

And they can then help you execute and achieve.

然后他们才能帮你执行、帮你达成。

I actually myself decided to go into IC role, even after I've led really large teams.

我自己其实就决定转做 IC,哪怕我之前带过非常大的团队。

Because I want to even myself learn of A, how to be more AI native, and I didn't feel like I could do it while managing large team.

因为我想亲自学两件事:第一,怎么变得更 AI native,而我觉得一边管大团队一边做不到这件事。

uh, Or managing honestly in general, because I wanted to learn how to build uh with AI.

呃,或者说老实讲,只要还在做管理就做不到,因为我想学的是怎么用 AI 呃去构建东西。

And then number two, how is it that uh organization can propagate that context all the way down and drive that trust to enable more in AI native people to accomplish their maximum potential.

第二,组织到底呃要怎么把上下文一路传递到最底层,建立起那种信任,让更多 AI native 的人发挥出最大潜能。

It's kind of funny, uh, when you were talking about the ability to do a PR to check in code again, your eyes lit up, and you looked so happy.

挺有意思的,呃,刚才你讲到自己能提 PR、能提交代码的时候,你眼睛都亮了,看起来特别开心。

So, it was so incredible, right?

所以说,这太不可思议了,对吧?

Like, creating things just gives us so much pleasure, and so I loved seeing that. On you.

就是,创造东西本身就能给我们带来那么多快乐,所以我特别喜欢在你身上看到这一点。

Well, it's been always such a gated experience that I needed an engineering team to do, and uh we still need engineers.

嗯,这件事以前一直是道门槛,必须靠工程团队才能做,呃,当然我们现在还是需要工程师。

I'm not saying we don't need engineers, we actually need more engineers.

我不是说不需要工程师了,我们其实需要更多工程师。

To as a non-technical person to be able to actually build something that really creates my vision, and I can see it live, uh, as a living breathing software out there.

而是说,作为一个非技术背景的人,居然能真的做出实现我愿景的东西,还能看着它上线,呃,变成一个活生生跑在外面的软件。

Or even change in production, like I vibe coded for example, some of the pricing page changes, even even like optimizations across the product,

甚至直接改生产环境——比如我 vibe code 过一些定价页的改动,甚至甚至产品各处的一些优化,

I always had to have engineering teams to do that.

这些以前我都得找工程团队来做。

Now engineering team, if I come to them with this task goes. What are you doing? Go do it yourself.

现在的工程团队,如果我拿着这种任务去找他们,他们会说:你干嘛呢?自己去做。

But like, let me know when PR is ready, and I can review it so, uh, so you can merge it.

不过,PR 好了叫我一声,我来 review,呃,然后你就可以合了。

But I'm not doing it, I have bigger more like more complex problems to go solve.

但我不做,我有更大、更复杂的问题要去解决。

Go do your thing yourself, but To them, it's an unlock because they can focus on deeper more interesting work.

你的事自己去搞——但对他们来说,这是个解锁,因为他们可以专注在更深、更有意思的工作上。

But to me, it's also an unlock because it's can enable me to do something that I've never been able to do by myself.

而对我来说,这同样是个解锁,因为它让我能做以前我一个人根本做不到的事。

So we both feel More empowered to do things that we were not previously, either were capable of doing or had time to do.

所以我们双方都觉得更有力量了,能去做那些以前要么没能力做、要么没时间做的事。

So, yeah, I don't know.

所以,是啊,我也说不好。

It's every time I'm put, I go into GitHub to merge a PR, I just I I like I I scream a little bit on the inside.

就是每次我,我进 GitHub 去合并一个 PR,我就我我就像我我都会在心里小小地尖叫一下。

Ha ha ha. Happy screams, it makes you happy.

哈哈哈。开心的尖叫,这让你很快乐。

Yeah, happy screams.

对,开心的尖叫。

I think it's one of the thread that I want to draw on from the thing that you talked about, right?

我想从你刚才讲的内容里抽出一条线继续聊,对吧?

You talked about trust, and you talked about agency.

你讲到了信任,也讲到了 agency。

You didn't quite use the word, but that's the one that I'm extracting from the individual.

你没有直接用这个词,但这是我从"个体"这个话题里提炼出来的。

People can do so much more.

每个人能做的事多太多了。

You don't have to worry about the cross functional boundaries.

你不用再操心跨职能的边界。

You're able to Achieve the goal that you set out for yourself.

你能达成自己给自己定下的目标。

Talk about that a little bit in the context of experimentation and how our audience should be thinking differently about their role and Not just a role, the empowerment and the unlock on agency, right?

请结合实验聊聊这一点,以及我们的观众该怎么换个角度看待自己的角色——不只是角色,还有 agency 上的赋能和解锁,对吧?

Because they can now do so much more.

因为他们现在能做的事多太多了。

Yes, I would also just add one more word autonomy.

对,我还想再加一个词:autonomy(自主权)。

Because yeah, I love that.

因为——对,我太喜欢这个词了。

You can do so much more by yourself, and uh this is where AI really shines.

你自己一个人就能做多得多的事,呃,这正是 AI 真正大放异彩的地方。

um, I have this concept I heard from Robbie McTavish.

嗯,我有个概念,是从 Robbie McTavish 那里听来的。

But I truly believe that AI, I know it's artificial intelligence, but I actually think it's the best average intelligence.

但我真心相信,AI——我知道它是 artificial intelligence(人工智能),但我其实觉得它是 best average intelligence(最强平均智能)。

uh, You need your own context to make it special, but imagine all of a sudden having at your fingertips.

呃,你需要用自己的上下文才能让它变得特别,但想象一下,突然之间你手边就有——

Average designer, average marketer, average product manager, average engineer.

一个平均水平的设计师、平均水平的营销、平均水平的产品经理、平均水平的工程师。

Will it make an exceptional result by itself?

光靠它自己能做出超凡的结果吗?

No, but add yourself into the mix, and you have your vision, you have your craft, you have your expertise that you're layering into it.

不能,但把你自己加进去——你有你的愿景、你的手艺、你的专业积累,一层层叠加进去。

And also having that at your fingertips, you can create something extraordinary.

再加上这一切随手可得,你就能创造出非凡的东西。

You can create so much more by yourself, um, than ever before.

你一个人能创造的东西,嗯,比以往任何时候都多得多。

Now, again, it only happens in a proper way if there is a trust in organization that is Giving you all of the right information to make right decisions,

不过,还是那句话,这一切只有在组织里存在信任时才能正常发生——组织把所有正确的信息都给你,让你做出正确的决策,

and giving you autonomy and agency without constantly having to travel decisions up and down or side to side in order for you to just go.

并且给你 autonomy 和 agency,让你直接往前跑,不用为此不停地把决策上上下下、左左右右地传来传去。

Because there's a very important shift that. happens when that is enabled in the company.

因为当公司真正放开这一点时,会发生一个非常重要的转变。

Which is velocity increases, and that can feel scary, because the major pushback usually is, well, what if you, how do you know you're running in the right direction?

那就是速度(velocity)会提上来——这可能让人害怕,因为最常见的反对意见通常是:呃,万一你……你怎么知道自己跑的方向是对的?

How do you know if you're doing the right things?

你怎么知道自己在做对的事?

And before we had a fear of doing the right things, which is why all of the decisions were travelling up and down, which we had such a big management level levels, is because cost of a mistake was really high.

以前我们总怕做得不对,所以所有决策都要上上下下地走流程,我们才有那么多层管理层级——因为犯错的成本真的很高。

You make a mistake.

你犯一个错。

It takes you months sometimes, maybe even years to get out of it.

有时要花好几个月、甚至好几年才能爬出来。

But if you accelerate velocity in terms in the organization, then cost of the mistake goes down.

但如果你把组织的速度提上去,犯错的成本就降下来了。

Because you made a mistake, OK? We have a high velocity, we can really quickly overcome it.

因为你犯了个错,OK?我们速度快,很快就能翻过去。

So you don't need to prep so much for any given initiative, you don't need to Be nervous so much about making some decision wrong, because then you can just evolve really quickly out of it. if you need to.

所以你不用为每个项目做那么多准备,也不用那么紧张某个决策会做错,因为真有需要的话,你可以非常快地调整出来。

But again, trust is super important.

但再说一遍,信任极其重要。

And that AI nativeness, and you'll be able to leverage that average expertise out of every single function out there, becomes really powerful and important.

而那种 AI native 的状态——你能把外面每一个职能的平均专业能力都调用起来——会变得非常强大、非常重要。

We've covered a lot of ground, Elena.

我们聊了很多内容,Elena。

This has been incredible.

这场对话太精彩了。

So, maybe just wrap it up, give us like the three takeaways that you want us to remember walking out of here.

那么,或许来收个尾吧,给我们讲讲你希望大家走出这里时记住的三条 takeaways。

Yes, uh, the first one, we'll start with experimentation.

好,呃,第一条,我们从实验说起。

Please experiment on your pricing and packaging, um, set up the right infrastructure for it now.

请在你的定价和打包方式上做实验,嗯,现在就把对应的基础设施搭起来。

Start to get the buy in whether it takes you to create some sort of monetization councils to get agree on the ideas.

开始去争取支持(buy-in)——哪怕这意味着你要成立某种商业化委员会,来让大家就想法达成一致。

Whether it just takes you to get the right talent and right owner for it to drive in the organization.

或者只是意味着你要找到合适的人才、合适的 owner,在组织里把这件事推起来。

But everybody who has any AI functionality in their product will have to embark on this journey.

但凡产品里带任何 AI 功能的人,都必须踏上这段旅程。

Start now, you don't wanna be left behind in it.

现在就开始,你可不想在这件事上掉队。

Especially as agents is gonna start coming in as our users, and we need to figure out how to monetize agents.

尤其是 agent 马上会开始作为我们的用户进场,我们得想清楚怎么对 agent 做商业化。

So, get into it right now.

所以,现在就动手。

Number two, experiment with yourself of what is it that AI nativeness means to you?

第二,拿自己做实验:AI native 对你来说到底意味着什么?

How much can you push it forward?

你能把它往前推多远?

And I really ask you to Designate maybe an hour a week, if you can, to just try tools over and over again.

我真心请你,如果可以的话,每周留出大概一小时,反反复复去试各种工具。

Because what was not possible three months ago might be possible now.

因为三个月前还做不到的事,现在也许就能做到了。

What does in AI native this means to you?

AI native 对你来说意味着什么?

How far can you get with AI on any given project without having to pull other stakeholders and uh and people into it in order to get it done?

在任何一个项目上,你能靠 AI 走多远,而不必为了完成它把其他利益相关方、呃、其他人拉进来?

And then the last one I would say is just really think about how you can give more of your product away to land grab as much as you can.

最后一条,我想说的是:认真想想怎么把你的产品更多地免费送出去,尽可能多地抢占地盘(land grab)。

So when you're doing monetization or experimentation or any of those initiatives, it's really so much about freemium now, and strategic giveaways, and unlocking engagement as much as possible,

所以当你做商业化、做实验或者任何这类动作时,现在的重头戏真的在于 freemium、战略性赠送,以及尽可能地解锁参与度,

because right now is when everybody is adopting it.

因为现在正是所有人都在采用它的时候。

Right now is not the time to ARPU-optimize or average revenue per user optimizations, and to be done.

现在不是做 ARPU 优化——也就是单用户平均收入优化——然后就收工的时候。

Right now is the time to get as many users to use your product as possible So how can you do that.

现在是要让尽可能多的用户用上你产品的时候。那你要怎么做到?

Amazing thank you so much

太棒了,非常感谢。

Like genuinely this was actually really fun to have this conversation with you

就是,真心的,和你进行这场对话真的特别开心。

You've given us so much to think about Elena

你给了我们太多值得思考的东西,Elena。

And I imagine this is gonna be one of those sessions that you know it's not just today but people are gonna come back to.

我想这会是那种,你知道,不只是今天,人们以后还会反复回看的场次。

And reference even months from now, I know I'm gonna listen to this a few times over for all the insights you've given us.

甚至几个月后还会引用——我知道我自己就会把这期反复听上好几遍,消化你给我们的所有洞察。

I think the uh piece that I wanna thank you for sincerely is also like how open you've been.

我想我要真诚感谢你的,呃,还有一点,就是你有多么开放。

In sharing some of the learnings, right, as you're seeing it happen in real time.

愿意分享这些经验教训,对吧,而且是在你实时看着它们发生的当下。

So, it's incredible for us to be able to learn from you.

所以,能向你学习,对我们来说太难得了。

Super fun to have you again, Elena.

再次请到你太开心了,Elena。

Thank you so much for joining us today.

非常感谢你今天的参与。

Thank you for having me. It's been a pleasure.

谢谢你们邀请我。这是我的荣幸。