做垂直 agent 的决策里,有 10% 是恐惧
"if I am honest with myself, about 10% of that decision came from fear — fear of entering a hyper-competitive market."
on why she chose a vertical agent over a general one — and how fear hides inside rational analysis
看似理性的"差异化"分析,底层可能是逃避不确定性。恐惧从不以恐惧之名出现,它伪装成分析。
选原生 agent 而非 workflow,代价是放弃企业市场
"I made a simple bet: choose the path that benefits from models getting stronger."
head-to-head 测试 workflow 准确率更高,她仍下注上限更高的 native agent
企业要可预测、稳定、准确——而这恰好是 agent 架构最不擅长的地方。事后才看清。
SaaS 是轨道,agent 是旷野
"SaaS gives you a track. Agents give you a wilderness."
用 Norbert Wiener 1948 年的控制论框定 SaaS 与 agent 的哲学对立
大多数用户被丢进旷野会僵住。这是 agent 公司难做 to-B 的根因——SaaS 赢确定性,agent 赢探索性。
想要按钮的是执行者,想要结果的是创始人
"the first camp was operators and executors. The second was founders."
GUI + LUI 混合产品上线后,反馈分成两个完全不同的阵营
满足第一类等于把 AI 装回 SaaS 老路。她最终选了拒绝大多数 feature request、服务想要"自主权超过工具"的那群人。
把"理论正确"当避风港,是分析型创始人的诅咒
"you mistake explanatory power for commercial traction."
复盘自己必须承认的四个错误之一——理论解释力 ≠ 商业牵引力
框架越优雅越危险——你以为在解释世界,其实只是在满足自己的虚荣,而不是用户的付费意愿。
AI 创业是明牌局:基础设施被垄断,应用层被民主化
"Monopolized infrastructure, democratized applications. Startups get squeezed from both ends."
旧金山黑客松走一圈后的悲观结构判断
移动时代有十亿级新用户增量,AI 时代没有。Foundation 模型公司亲自做应用、Claude Code 让谁都能周末出产品——创业公司三头夹击。
Lovable 用 18 万付费用户做到 2 亿 ARR——歼灭战该让位游击战
"find 10,000 power users, serve them deeply, and you are a $20M ARR company."
当 agent 的定价基准从 $20/月 SaaS 上升到"一份人力薪水",小团队也能做大公司
不需要堆 DAU、打广告战。几个有 taste、有执行力的人,深度服务一万付费用户,就能活得很好、长得很快。
"San Francisco has been raining for three straight days."
旧金山连下了三天雨。
"A cold-rain night like this feels like the right time to honestly review the past six months — the traps I walked into, why I'm pessimistic about the structure of AI entrepreneurship, and why I'm still full of fight for 2026."
在这样的冷雨之夜,适合诚实地回望过去六个月——我踩进的陷阱、我为什么对 AI 创业的结构感到悲观、以及我为什么仍然准备在 2026 年继续战斗。
"Six months in. A lot of tuition paid."
六个月过去了。学费交了不少。
"Fortunately, relative to our funding, the cost was manageable — but if we had been a YC company with a smaller runway, we would already be dead."
幸运的是,相对于我们的融资规模,代价还在可承受范围内——但如果我们是一家 YC 公司、runway 更短,我们这会儿已经死了。
"I want to lay out my decisions honestly, open-source my mistakes."
我想诚实地把这些决策摆出来,把我的错误开源。
"In hindsight, right and wrong are easy to assign."
事后回看,对错都很容易判断。
"But I want to recreate the fog of the moment, because the hard part of startup decisions is never the retrospective attribution — it is the ambiguity of the present."
但我想还原当下那种迷雾感——因为创业决策最难的部分从来不是事后归因,而是身处其中时的那种含糊不清。
"Our first big call was to build a vertical agent, not a general one."
我们的第一个大决策,是做一个垂直 agent,而不是通用 agent。
"The logic was clean: general AI is a big-company game, and a startup charging in has no advantage."
逻辑很干净:通用 AI 是大公司的游戏,创业公司硬冲毫无优势。
"I still believe this was rational."
我至今仍然相信,这个决策是理性的。
"But if I am honest with myself, about 10% of that decision came from fear — fear of entering a hyper-competitive market."
但如果我对自己诚实,这个决策里大概有 10% 来自恐惧——害怕进入一个超激烈竞争的市场。
"That fear is insidious."
这种恐惧是阴险的。
"It never appears in your consciousness as “I'm scared.”"
它从来不会以"我害怕"的样子浮上意识。
"It disguises itself as reasonable analysis: “this direction isn't differentiated enough,” “big companies will crush us here.”"
它会伪装成理性的分析:"这个方向差异化不够"、"大公司会在这里碾压我们"。
"But underneath, it is just helping you avoid uncertainty."
但骨子里,它只是在帮你避开不确定性。
“That fear is insidious. It never appears in your consciousness as 'I'm scared.' It disguises itself as reasonable analysis.”
"After going vertical, we faced a sourcing use case."
选定垂直之后,我们要做的是一个 sourcing(寻源)场景。
"Agent accuracy was low."
Agent 的准确率很低。
"The temptation was to build a hybrid — borrow from traditional software engineering, wire up a workflow, and get accuracy up."
当时的诱惑是做一个混合体——从传统软件工程里借经验,搭一条 workflow,把准确率拉上去。
"We compared the two approaches head-to-head: the workflow won on accuracy."
我们把两种方案做了正面对照测试:在准确率上,workflow 赢了。
"But I made a simple bet: choose the path that benefits from models getting stronger."
但我下了一个简单的赌注:选那条"模型变强会带着它变强"的路。
"As models improve, agents improve."
模型变强,agent 也跟着变强。
"A workflow's ceiling is the flow you hard-coded."
而一个 workflow 的天花板,就是你硬编码出来的那条 flow。
"So we chose native agent."
所以我们选了原生 agent。
"In hindsight, this decision meant we could never serve enterprise customers well — they want predictable, stable, accurate."
事后看,这个决策意味着我们永远没法服务好企业客户——他们要的是可预测、稳定、准确。
"That is exactly where agent architecture is weakest."
而这恰恰是 agent 架构最弱的地方。
"But at the time, this was not clear."
但在当时,这件事并不清楚。
"Traditional software is built on cybernetics — Norbert Wiener's 1948 theory."
传统软件的根基是控制论(cybernetics)——诺伯特·维纳(Norbert Wiener)1948 年的理论。
"A feedback loop: measure, compare, adjust."
一个反馈回路:测量、比较、调整。
"Same input, always the same output."
同样的输入,永远是同样的输出。
"SaaS is entropy reduction: a product manager talks to a hundred users, abstracts a workflow, locks the path down."
SaaS 是熵减:一个 PM 跟一百个用户聊完,抽象出一条 workflow,把路径锁死。
"Users follow the track."
用户沿着轨道走。
"Agents are the philosophical opposite."
Agent 在哲学上正好相反。
"They are stochastic, not deterministic."
它们是随机的,不是确定的。
"The path is not pre-written — it emerges from the agent's interaction with its environment."
路径不是预先写好的——它是 agent 与环境互动时涌现出来的。
"Every execution may differ."
每一次执行都可能不一样。
“SaaS gives you a track. Agents give you a wilderness. SaaS wins on certainty. Agents win on exploration.”
"SaaS gives you a track. Agents give you a wilderness."
SaaS 给你一条轨道,agent 给你一片旷野。
"SaaS wins on certainty. Agents win on exploration."
SaaS 赢在确定性,agent 赢在探索性。
"But here is the truth: most people, if you take them off the track and drop them into a wilderness, will freeze."
但真相是:大多数人,如果你把他们从轨道上拉下来、扔进旷野,他们会僵住。
"I am not dismissing SaaS — it is still a worthy game."
我不是在否定 SaaS——它依然是一个值得玩的游戏。
"I just do not find that game exciting."
我只是觉得那个游戏不令我兴奋。
"After shipping a GUI + LUI hybrid product, feedback split into two camps."
在我们做完一个 GUI + LUI(图形 + 语言界面)混合产品之后,反馈分成了两个阵营。
"The first wanted a clear workflow, more buttons, more dashboards — basically a SaaS."
第一类要的是清晰的 workflow、更多按钮、更多 dashboard——本质上就是 SaaS。
"The second said: just tell me the goal and send me the result."
第二类说:你告诉我目标就好,把结果发给我。
"A pattern emerged: the first camp was operators and executors. The second was founders."
一个 pattern 浮出来了:第一类是操作员、执行者;第二类是创始人。
"When I catalogued every feature request from the first camp, I realized I was echoing five-year-old SaaS products in our space."
当我把第一类用户的每一条 feature request 列出来时,我发现自己正在重复我们这个赛道五年前的那些 SaaS 产品。
"Those incumbents were asking how to add AI to their SaaS."
那些老玩家在问:怎么把 AI 装进我的 SaaS。
"We were asking how to add SaaS to our AI."
而我们在问:怎么把 SaaS 装进我的 AI。
"Different starting points — converging features."
起点不同,功能却在收敛。
“I had been repeating someone else's path while telling myself I understood AI better.”
"This observation was extremely painful, and took courage to admit: I had been repeating someone else's path while telling myself I understood AI better."
这个观察极其痛苦,而且承认起来需要勇气:我一边走着别人的老路,一边告诉自己"我比他们更懂 AI"。
"So we made a pivot: we rejected most feature requests, blocked the button-and-dashboard demands, and focused on serving users who wanted autonomy over tooling."
所以我们做了一次 pivot:拒绝绝大多数 feature request、挡掉"加按钮加 dashboard"的诉求、专注服务那些"想要自主权,而不是工具"的用户。
"I think this is a curse."
我觉得这是一种诅咒。
"Founders with strong analytical ability often build the worst products — because when a framework explains things elegantly, you mistake explanatory power for commercial traction."
分析能力越强的创始人,往往做出来的产品越糟糕——因为当一个框架能优雅地解释一切时,你会把"解释力"误认成"商业牵引力"。
"You end up satisfying your own vanity, not the user's willingness to pay."
到头来你满足的是自己的虚荣心,而不是用户的付费意愿。
"A repeatable, fast-to-deliver, word-of-mouth single-point product — I severely underestimated this."
一个可重复、能快速交付、能口口相传的单点产品——我严重低估了它。
"Meituan started as a Groupon clone."
美团一开始就是一个 Groupon 的 copycat。
"ByteDance started as a news aggregator."
字节跳动一开始是一个新闻聚合器。
"They grew into super-applications not because the starting point was sexy, but because it was sharp enough to pierce a real need."
它们最后长成超级应用,不是因为起点很性感,而是因为起点足够尖锐,能戳穿一个真实的需求。
"I was too allergic to building anything that looked like a SaaS tool."
我对任何"看起来像 SaaS 工具"的东西过敏太严重。
"The result: I never secured a starting point at all."
结果是:我根本就没有真正拿到一个起点。
"This is the most hidden trap."
这是最隐蔽的一个陷阱。
"When I discussed product positioning, a huge amount of self-narrative was embedded in the argument."
当我在讨论产品定位的时候,论证里嵌入了大量的"自我叙事"。
"Why did I hate toolification? Why did I want the product to have a global view?"
我为什么讨厌工具化?为什么希望产品能有一个全局视角?
"How much of that came from user needs and how much from my own aesthetic preferences and identity?"
这里面,有多少来自用户需求,有多少来自我自己的审美偏好和身份认同?
"I found myself “preserving optionality” to avoid the pain of choosing."
我发现自己一直在"保留选项",以此回避选择本身的痛苦。
"But startups demand the opposite: pick a direction, test it, accept the result, move on."
但创业要求的恰好相反:选一个方向、去测、接受结果、继续往前走。
"The visionary's disease: because you can see an opportunity logically disappearing within a year, you refuse to pursue it."
远见者的病:因为你能在逻辑上看到一个机会在一年内会消失,所以你拒绝去做它。
"But you underestimate distribution friction and trust."
但你低估了分发摩擦和信任成本。
"Users do not care that your product might theoretically be replaceable in twelve months."
用户不会在意你的产品"理论上"十二个月后会被替代。
"They have a pain point today, and they will pay today."
他们今天就有痛点,今天就会付钱。
"Similarly, many smart founders see market homogeneity and interpret it as “no opportunity.”"
同样地,很多聪明的创始人看到市场同质化,就解读为"没机会"。
"But homogeneity means the opposite — everyone has spotted the need, but no one has found the deepest factor yet."
但同质化的含义恰好相反——每个人都看到了这个需求,但没有任何人找到最深的那个因子。
"Five thousand group-buying websites were homogeneous to the extreme. Meituan still emerged from that field."
五千家团购网站同质化到了极致,美团依然从那片战场里冒出来了。
"Too little verification. Too much extrapolation. Too much love for narrative beauty. Too much fear of an unsexy starting point."
验证太少。外推太多。太爱叙事的美感。太害怕一个不性感的起点。
"The result: I never even secured a starting point."
结果是:我连起点都没拿到。
"On this visit to San Francisco, I finally went deep into the local startup ecosystem — hackathons, incubator events, Founders Inc.'s Artifact Festival."
这趟来旧金山,我终于深入走进了本地的创业生态——黑客松、孵化器活动、Founders Inc. 的 Artifact Festival。
"Walking the floor, I had one overwhelming feeling: I could have guessed every project here without leaving my apartment."
在场子里转了一圈,我有一种压倒性的感受:我不出门也能猜到这里的每一个项目。
"Marketing agents, sales agents, PM agents, browser agents, agents that cancel your appointments."
营销 agent、销售 agent、PM agent、浏览器 agent、帮你取消会议的 agent。
"A fatigue set in."
一种疲惫感升上来。
"Are we hitting an imagination ceiling? Is every card in this game face-up?"
我们是不是撞上了想象力的天花板?这局游戏里,是不是每一张牌都已经翻在桌面上了?
"I looked to the mobile internet era for answers, because it was strikingly similar — all cards were face-up there too."
我去移动互联网时代找答案,因为那个时代和现在惊人地相似——当时的牌也都翻在桌面上。
"Yet from that ocean of obvious opportunities, ByteDance, Meituan, Instagram, and Snapchat emerged."
但就是从那片"机会一目了然"的海洋里,字节跳动、美团、Instagram、Snapchat 冒了出来。
"ByteDance survived because it found the first-principles factor: recommendation algorithms powered by a data flywheel."
字节跳动活下来,是因为它找到了第一性原理的那个因子:由数据飞轮驱动的推荐算法。
"Users bring better content, not the other way around."
是用户带来更好的内容,而不是反过来。
"By the time Baidu shipped its news feed in 2017 and Tencent launched Tiantian Kuaibao in 2015, ByteDance already had three to five years of behavioral data and model iteration."
等到百度在 2017 年上线 feed 流、腾讯在 2015 年推出"天天快报"时,字节跳动已经累积了三到五年的用户行为数据和模型迭代。
"Late entrants could not buy back that time."
晚到的玩家用钱买不回那段时间。
"Meituan won the Thousand Groupon War not by outspending but by out-operating."
美团赢下"千团大战",不是靠烧得更多,而是靠运营得更狠。
"Wang Xing used KFC locations to gauge a city's spending power, Taobao indices for online buying habits, cinema counts for local activity."
王兴用肯德基的门店数判断一个城市的消费力,用淘宝指数衡量线上购物习惯,用电影院数量看本地活跃度。
"He only entered cities that met the threshold."
他只进入达到阈值的城市。
"Competitors blanketed hundreds of cities, burning cash in markets that generated no meaningful transactions."
竞争对手把战线铺到上百个城市,在根本不会产生有意义交易的市场上烧钱。
"When the funding winter hit in late 2011, 96% died."
2011 年底融资寒冬一来,96% 的玩家死了。
"Instagram found the deepest factor in a crowded photo-sharing market: mobile-first simplicity."
Instagram 在拥挤的图片分享市场里找到了最深的那个因子:Mobile-first 的极简。
"Four steps — shoot, filter, caption, share."
四步——拍、滤、写、发。
"It deliberately ignored the web, betting that phones would become the center of everything."
它故意忽略了 web,赌手机会成为一切的中心。
“The winners did nothing new. News aggregation, group buying, photo sharing, instant messaging — all existing categories.”
"The winners did nothing new."
这些赢家什么新东西都没做。
"News aggregation, group buying, photo sharing, instant messaging — all existing categories."
新闻聚合、团购、图片分享、即时通讯——全是已有的品类。
"They won by finding the deepest factor in each category and executing it beyond anyone's reach."
他们之所以赢,是因为在每个品类里找到了最深的那个因子,并且把执行做到了无人能及的程度。
"Honestly, the more I think about it, the more pessimistic I become."
坦白说,越想越悲观。
"The mobile era had one massive structural advantage that does not exist today: a billion-scale new-user increment from device adoption."
移动时代有一个今天根本不存在的巨大结构性优势:由设备普及带来的、十亿级别的新增用户增量。
"AI has no equivalent."
AI 没有对等物。
"There is no new device category penetrating rapidly."
没有什么新设备品类在快速渗透。
"All AI products are fighting over existing internet users — competing for attention from a fixed pool."
所有 AI 产品争夺的,都是存量互联网用户——在一个固定池子里抢注意力。
"Worse, the foundational layer is feudal."
更糟的是,基础层是封建的。
"Training a frontier model costs approaching billions."
训练一个前沿模型,成本正在逼近十亿级。
"Foundation model companies are not neutral platforms like Apple — they build competing applications themselves."
基础模型公司不像苹果那样是中立平台——它们自己也在做与你竞争的应用。
"Your infrastructure provider is also your biggest potential competitor."
你的基础设施提供商,同时也是你最大的潜在竞争对手。
"Meanwhile, engineering capability has been fully democratized: anyone with Claude Code can ship a product in a weekend."
与此同时,工程能力已经被彻底民主化:任何有 Claude Code 的人,都能在一个周末做出一个产品。
"Monopolized infrastructure, democratized applications. Startups get squeezed from both ends."
基础设施被垄断,应用层被民主化。创业公司从两头被挤压。
“Monopolized infrastructure, democratized applications. Startups get squeezed from both ends.”
"Want to build network effects? Scale largely follows capital — big companies always have more."
想做网络效应?规模基本跟着资本走——大公司永远比你多。
"Want a data flywheel? ByteDance succeeded because incumbents were three to five years late to react."
想要数据飞轮?字节跳动能赢,是因为老玩家迟到了三到五年才反应过来。
"Today, OpenAI sees a promising direction and ships in three months."
而今天,OpenAI 看到一个有戏的方向,三个月就能上线。
"Want technical moats? You use the same APIs as everyone else."
想要技术护城河?你用的是和别人一模一样的 API。
"In this era, startups have no structural advantage."
在这个时代,创业公司没有任何结构性优势。
"Almost nothing constitutes a moat — except taste and community."
几乎没有任何东西能构成护城河——除了品味(taste)和社群(community)。
"When features can be pixel-perfect copied — Instagram Stories was a pixel-perfect copy of Snapchat — what keeps users is not the feature but the belonging."
当功能可以被像素级复制——Instagram Stories 就是 Snapchat 的像素级翻版——能留住用户的不再是功能,而是归属感。
"Facebook had two billion users and still could not replicate Snapchat's place in Gen Z's heart."
Facebook 有二十亿用户,依然复制不了 Snapchat 在 Z 世代心里的那个位置。
"In the AI age, the feeling that a product gets you may be the only card that cannot be cloned in a weekend."
在 AI 时代,"这个产品懂我"的那种感觉,可能是唯一一张无法在一个周末被克隆掉的牌。
"After all that pessimism — I am not actually a pessimist."
说完这么多悲观——但我其实并不是悲观主义者。
"If I were, I would have stopped."
如果我是,我早就停下来了。
"I am pessimistic about the old rules. I am optimistic that the rules are changing."
我悲观的是旧规则,我乐观的是规则正在改变。
"And entrepreneurship is betting that you are an outlier."
创业本身就是一个赌注——赌你是那个 outlier。
"I have never doubted that about myself."
关于这一点,我从来没有怀疑过自己。
"The old playbook chased millions of users, tens of millions of DAU, billion-dollar GMV."
旧的玩法追求百万用户、千万 DAU、十亿 GMV。
"That was a “traffic tax” model: capture users first, monetize later."
那是一种"流量税"模式:先把用户圈进来,以后再变现。
"But if you truly believe agents can do extraordinary things, the pricing benchmark is not a $20/month SaaS subscription — it is a human salary."
但如果你真的相信 agent 能做出非凡的事情,定价的参照系就不再是 20 美元/月的 SaaS 订阅——而是一份人的工资。
"Lovable reached $200M in ARR with only 180,000 paying users — over $1,000 per user per year."
Lovable 用 18 万付费用户做到 2 亿美金 ARR——人均每年超过一千美元。
"You do not need to win an annihilation war."
你不需要打一场歼灭战。
"You need a guerrilla campaign: find 10,000 power users, serve them deeply, and you are a $20M ARR company."
你需要的是一场游击战:找到 10,000 个深度用户,深度服务他们,你就是一家 2,000 万美金 ARR 的公司。
"That is enough to live well and grow fast."
这足以活得好,长得快。
“Find 10,000 power users, serve them deeply, and you are a $20M ARR company.”
"This creates an entirely new company archetype: tiny team, extraordinary output per person."
这催生出一种全新的公司原型:很小的团队,每个人的产出却极其惊人。
"Lovable had fewer than 100 people when it crossed $100M ARR."
Lovable 跨过 1 亿 ARR 的时候,团队还不到 100 人。
"You do not need a massive fundraise to support a massive team."
你不再需要靠一轮巨额融资去养一支庞大的团队。
"A few people with taste, execution, and user empathy can build a product serving tens of thousands of power users."
几个有 taste、有执行力、有用户同理心的人,就能做出一款服务数万深度用户的产品。
"Meituan won the Thousand Groupon War not because of better UI but because it built merchant settlement infrastructure — letting merchants see their accounts in real-time, withdraw anytime."
美团赢下千团大战,不是因为 UI 更漂亮,而是因为它做了商家结算的基础设施——让商家能实时看到自己的账户、随时提现。
"That solved supply-side trust."
这解决了供给侧的信任问题。
"Five thousand competitors could build a pretty consumer page; only Meituan bothered with the unsexy merchant plumbing."
五千个竞争对手都能做出漂亮的消费者页面;只有美团愿意去搞商家这套不性感的管道工程。
"In the AI era, this logic is more important, not less."
到了 AI 时代,这套逻辑更重要,而不是更次要。
"Big companies excel at horizontal tools — ChatGPT, Gemini, Claude are all general-purpose."
大公司擅长做横向工具——ChatGPT、Gemini、Claude 都是通用型的。
"But they will never send someone to a WeWork to talk through a solo founder's growth anxiety, never understand a dental clinic's scheduling pain, never learn a cross-border seller's compliance details."
但它们永远不会派人去一个 WeWork、跟一个单人创始人聊他的增长焦虑;不会理解一家牙科诊所的排班痛点;不会摸清跨境卖家的合规细节。
"Domain knowledge and relationships are ground out one by one. No amount of compute solves that."
领域知识和关系网,是一个一个磨出来的。再多的算力也解决不了这件事。
"Finite games end with winning or losing. Infinite games end only when you stop playing."
有限游戏以输赢结束;无限游戏只有在你停下来的那一刻才结束。
"If you treat entrepreneurship as a finite game — funding as victory, IPO as the finish line — every setback is a loss and you crack under the pressure."
如果你把创业当成有限游戏——把融资当胜利、把 IPO 当终点线——每一次挫折都是一次失败,你会在压力下崩掉。
"If you treat it as an infinite game, what you pursue is your own iteration, and success is a byproduct."
如果你把它当成无限游戏,你追求的是自己的迭代,而成功只是副产品。
"A company I thought had fallen behind six months ago just shipped a model that stunned me."
一家六个月前我以为已经落后的公司,刚刚发布了一个让我震惊的模型。
"They never left the game."
他们从来没有退出过游戏。
"In this industry, no one is permanently sentenced, but no one can rest on past glory."
在这个行业里,没有谁被永久判刑,也没有谁能躺在过去的荣光里休息。
"The deck reshuffles every six months."
每隔六个月,牌就重新洗一次。
"I believe entrepreneurship is the best way to project your worldview onto reality."
我相信创业是把你的世界观投射到现实上最好的方式。
"You have convictions about the world — what matters, what is right, what deserves to exist."
你对这个世界有一些信念——什么是重要的、什么是对的、什么是值得存在的。
"Building a company is the only way to test those convictions with your own hands."
做一家公司,是唯一能用你自己的双手去检验这些信念的方式。
"Not a social media take. Not a podcast opinion."
不是一条社媒发言,不是一次播客上的看法。
"Real work, real friction, real arm-wrestling with reality."
是真实的劳动、真实的摩擦、真实地和现实掰手腕。
"And building has made me a better person."
而且创业让我成为一个更好的人。
"The growth I have experienced — in character, communication, even in how I love — so much of it comes from facing pressure, uncertainty, and the daily need to convince others to believe in what I believe."
这些年里我所经历的成长——在性格上、在沟通上、甚至在我如何去爱上——绝大部分都来自于面对压力、面对不确定性、每天都要去说服别人相信你所相信的东西。
"Drop the illusion. Prepare for battle. Compete in the obvious markets. Do the thing that looks dumb. Bet on the future."
放下幻觉。准备战斗。冲进显而易见的市场。去做那件看起来很蠢的事。下注未来。
"Never go back to patching the old world — go to the new world and fight head-on."
不要回去给旧世界打补丁——到新世界去,正面开战。
"A researcher I admire once said: even if you fall, fall at the front of the line — at least the people behind you can pull you back up."
我尊敬的一位研究者曾说过:就算你倒下,也要倒在队伍的最前面——至少身后的人能把你拉回来。
"But I cannot romanticize the idea that some hidden card will magically appear one day, dealt only to me."
但我不能浪漫化地以为,某一天会神奇地从牌堆里翻出一张只发给我的暗牌。
"Innovation is not born from daydreaming."
创新不是从白日梦里诞生的。
"Innovation is born from vision married to craft."
创新诞生于愿景与手艺的结合。
"There is a quote that every Chinese student has used in an essay at least once: “There is only one heroism in the world: to see the world as it is, and to love it.”"
有一句话,每个中国学生至少在一篇作文里引用过:"世上只有一种真正的英雄主义,那就是认清生活的真相之后,依然热爱生活。"
“That is the heroism of entrepreneurship.”
"I want to offer that to every AI founder."
我想把这句话送给每一个 AI 创始人。
"When you have recognized that you have no structural advantage in this war — when you accept that as a gambler you may walk away with nothing — and you still charge in with full conviction and guerrilla spirit —"
当你已经认清这场战争里你没有任何结构性优势——当你接受作为一个赌徒你可能空手而归——而你仍然带着满腔信念和游击精神冲进去——
"That is the heroism of entrepreneurship."
那就是创业的英雄主义。
"Have fun out there."
外面战场凶险,玩得开心。