Kimi Without Titles, Departments, or OKRs: How 300 People Carry a $120 Billion Valuation | BlueRun Ventures Family Headlines

This People magazine feature, "Undercover" at Kimi: 100 Hours, is the most fitting profile of Moonshot AI we've read in recent memory.

As an early investor in Kimi, BlueRun Ventures was the first to issue a term sheet in the A1 round. At the time, the technology still carried risks, commercialization had yet to materialize, and upstream resources could have been monopolized by competitors. BlueRun still chose to back a young man who practiced his "10,000 hours of passion" and possessed extraordinary talent magnetism — Kimi founder Zhilin Yang.

From 2023 to today, BlueRun has continuously increased its stake across four rounds starting from the Series A, watching Kimi grow from zero to an AI unicorn valued at over 100 billion RMB. We've kept asking ourselves: what makes this company so different?

Some details in the article happen to offer partial answers.

It describes an extraordinarily rare organizational form. No departments, no hierarchies, no OKRs. Founder Zhilin Yang's WeChat signature reads simply four characters: "Direct Communication." Here, to get anything done that requires someone else's help, you just "ask directly." This extreme flatness still functions at 300-plus people, sustained not by systems but by the self-drive and shared consensus of a "cluster of geniuses."

It captures the essence of these people. 80% are introverts, yet the HR dares to throw a candidate's soul-searching question directly at a co-founder in an all-hands meeting: "DeepSeek also made an offer, why should I come to Kimi?" Someone spent hundreds of hours livestreaming on Bilibili to learn skills for an internal transfer. Someone went through "three exiles to the repentance cliff" — a project rejected and returned three times. The article gets one word exactly right: resilience.

In these most frenzied years of the AI wave, few startups have maintained this state of continuous ascent. As an early-stage investment firm that similarly pursues "undefined, trailblazing" ideals, BlueRun is honored to witness this "moon landing" journey.

(This article is republished with permission from People, written by Liu Mo, edited by Jin Zha)

The "Undercover" Plan

The spring of 2026 has favored Kimi. From revenue, funding, and valuation records shattered in succession, to a paper with a 17-year-old high school intern as first author earning high praise from Silicon Valley heavyweights including Elon Musk, to being "wrapped" by Cursor — a US company valued at $50 billion — Kimi has nearly simultaneously completed a beautiful triple play of capital, technology, and commerce. This three-year-old startup, valued at over 120 billion RMB, is gradually emerging in the global AI narrative.

Yet Moonshot AI remains mysterious.

I was granted 100 hours of deep observation at company headquarters. As an independent writer, I could interview any employee willing to speak, sit in on any meeting not involving commercial secrets, create without interference, and receive no compensation — this is indeed the company's style.

When you stand inside the company, it's like being at the eye of a storm. Within the eye, all is still. Workstations are quiet, keyboard clicks sparse, occasional laughter drifting over. The outside noise — rumors, debates, adulation and imitation — finds no echo here.

300-plus people, average age under 30, each shouldering nearly 400 million RMB in valuation. 80% of colleagues here are introverts — people sit side by side yet prefer typing to talking. Here, introversion isn't a flaw but an organizational protocol.

I recall my first visit to this company in 2024, on that night when the storm began brewing. My first impression of Moonshot AI wasn't particularly favorable.

"DeepSeek Saved Us"

The evening of December 24, 2024, was an unremarkable Christmas Eve for most Chinese, but the darkest moment of Julian's life. At 26, just two years out of Peking University with no industry experience, she was already among Kimi's earliest employees. This young yet senior girl sat at the long table of the Radiohead conference room, facing thirty-plus colleagues, tears streaming uncontrollably.

She still couldn't deliver a holiday marketing plan that met co-founder standards. With a month until Spring Festival, to continue upgrading or even completely overturn the latest plan — already revised six times — while ensuring product and R&D team coordination was itself a low-probability event. But the company's growth expectations for Spring Festival 2025 were enormous: it was last Spring Festival when Kimi broke through with "2-million-character long context," surging in C-end users and even spawning "Kimi concept stocks" in the capital markets.

That weekly meeting was long and despairing: twenty young colleagues as inexperienced as Julian took turns presenting, from social media placement to user operations, domestic PR to overseas marketing, every detail discussed collectively, decisions made by co-founders. Kimi was like an adolescent, bewildered teenager — even with tens of millions in monthly ad spend, facing aggressive competitors, it was flailing.

The meeting ended promptly before 4 a.m.

No one knows if Julian's plan eventually succeeded. Because a month later, when the world first learned the name DeepSeek, none of it mattered anymore.

Hayley from the growth team returned to her hometown in Wenzhou; relatives and friends all asked: Do you know DeepSeek? As if Kimi were a name from last century. Hayley endured the hardest New Year of her life. She said, the company's silence was deafening.

The annual meeting, typically held in March after Spring Festival, allows all employees to challenge the bosses. That year's questions almost all revolved around DeepSeek. The sharpest came from the HR team, who with absolute sincerity pierced through the pretense: How should we answer candidates' question —

"DeepSeek also gave me an offer, why should I come to Kimi?"

But not everyone thought this way. Alex from the algorithms team recalls that if he felt any strong emotion during the DeepSeek moment, it was only one: excitement.

This excitement represented not just him but the entire algorithms team's mindset. They saw another possibility: lower-cost strategies, open-source approaches, and a fact no one had believed — with sufficiently leading technology and solid models, an unknown Chinese startup could earn global respect.

The product team wasn't anxious either. Kevin, among the earliest product hires, had clear and firm confidence: DeepSeek broke through on model capabilities, but once Kimi's model capabilities caught up, the embellishments they could add on the product end would be far greater.

No one knows what discussions the co-founders went through. But the company completed strategic adjustment and refocusing with remarkable speed, achieving genuine all-hands consensus. Ask any colleague what the company's most important priority is, and they'll tell you without hesitation: the model.

From then on, you could sense the respect for DeepSeek permeating the company — partly professional camaraderie from a peer perspective, partly, as Alex put it: "Actually, DeepSeek saved us."

Kimi K2.5 model performance

Taste Is All You Need

"Why are you wearing shoes like that?"

After Ezra asked this in surprise, I was more surprised than her. On her office floor, nearly everyone keeps a pair of slippers under their desk, because comfortable dress makes people more relaxed, more focused, more creative.

This is the dress code of the smart.

I've met many top students, but the "good students" here are very different. Ezra tried to hack her family's computer in elementary school, simply because her parents wouldn't tell her the password; in middle school she developed interest in Bitcoin, then priced at just 300 RMB per coin, and persuaded her mother for pocket money to invest — her mother told her it was a scam; the first time she took a taxi in high school, she sketched the product model for ride-hailing, though without AI tools at the time, she might have launched it much earlier; in college she finally had her own spending money and entered the stock market, losing 90% in A-shares.

This investing Waterloo made her deeply reflect on human limitations, unlocking her interest in AI.

Her understanding of AGI is simple: create N Albert Einsteins to solve all of humanity's hard problems. From then on, she resolved to find a company exploring AGI's limits. Though she had since made back her A-share losses.

With solid academic credentials, she received offers from major companies. She chose Kimi solely because during the interview, founder Zhilin Yang's deep technical understanding and meticulous attention to detail moved her. She believed Yang was someone who truly cared about models. He lacked the restlessness of smart people, the utilitarianism of businesspeople — she didn't even know he was the company founder until the interview ended.

Karen was rebellious from childhood, arguing with teachers, never listening to parents, insisting on going abroad for school, then insisting on starting a company after graduation, despairing of the stable, comfortable life at a big tech firm. He didn't want a life he could see the end of from the start.

I asked him: if 100% chance of 60 points, versus 1% chance of 100 points, which would you choose?

He chose the latter without hesitation. Not that he couldn't stand 60 points — he just hated that 100%.

Such entrepreneurial genes form a collective undertone. According to incomplete statistics, at least 50 people at Moonshot AI have founded or joined startups.

Clearly, Kimi likes to hire CEOs.

More precisely, here shelters wave after wave of wandering geniuses. Genius need not mean top student or good student — what matters is possessing, in some dimension, eyes that penetrate time.

In this company where 80% of employees hold degrees from "985" and "211" universities, Yannis's resume doesn't stand out, yet as early as 2023 he foresaw in engineer communities the rise of DeepSeek and Kimi — in an era when model companies had no products yet. This foresight was discovered by another post-00 colleague, who referred him into the company.

Karen says too many smart people are trapped inside the system. From family to school to the workplace, they unconsciously submit to the collective, blind to their own true needs. Only a small fraction try to escape, and they're often invisible to the world. One of Kimi's missions is to see them.

Without this kind of seeing, there would be no seventeen-year-old high schooler interning at Kimi, collaborating with the team, publishing a paper that earned Elon Musk's praise. The person who put him as first author was precisely the "talent scout" who had discovered him — Bob.

There's only a fine line between genius and madness. When a misunderstood madman arrives at Moonshot AI, he might suddenly become a genius who changes the world; or those geniuses not yet revealed to the world can only bloom wildly here. Bob told me that, in a sense, a big Ego isn't a problem — it might even be a good thing. Treating Ego as internal drive, convinced that one must participate in some great cause — that's the truly mad genius, and also the person they would never want to miss.

Genius is偏执 (obsessive).

On this team, training top-tier AI models is called "alchemy," and alchemy is essentially endless bug-fixing. After launching a Flagship Run, Bob and his colleagues developed an unshakeable habit: the first thing they do upon waking is refresh over a hundred thousand internal monitoring metrics. Any anomalous spike on the screen triggers immediate alert — is there an optimization problem? A flaw in architecture design? Or misaligned numerical precision?

They're as keen as well-trained animals. Some even filter out tokens with excessively large gradient values from the training corpus, print them out one by one, and interrogate them word by word: why do you fluctuate so violently?

Everyone who has truly participated in this "delivery" has lived through days of such tension they couldn't sleep at night. They're not anxious — curiosity drives them. This偏执 (obsessive) vigilance pushes the model to industry-leading levels.

Geniuses cluster.

Over the past year, more than 100 of Kimi's hires came from internal referrals — friends, or friends of friends. This hiring model is internally called "human-to-human transmission." Based on this deeply connected relationship network, trust becomes a natural organizational asset.

Essentially, Kimi transfers the difficulty of organizational management onto talent recruitment. People attracted through referrals "share the same scent," which echoes the keyword almost everyone emphasizes: TASTE.

One evening in September 2025, several engineers casually launched an internal side project, naming it "Ensoul" — meaning "to give a soul." The name itself reads like poetry: they wanted sleeping code files to "come alive," becoming a conversational intelligent assistant right in the command line.

This sensitivity to naming is no accident. They once had a framework called "YAMAHA" — actually an acronym for "Yet Another Moonshot Agent"; and the core underlying layer was named "Kosong" — meaning "emptiness" in Malay, taken from the Zen concept of "form is emptiness,"寓意 (implying) it is like a blank sheet of paper, presupposing no functions yet containing all possibilities.

It is precisely such taste that determines what a product looks like.

While others were cramming chat windows into command lines, they found this ugly: real programmers open terminals to input commands, not to chat. So Kimi CLI was designed more like a "smart Shell" — it understands the commands you type, but doesn't forcibly become a chat window.

This minimalism also shows in the code. The entire core logic is just 400 lines of Python, like a short poem, with all unnecessary ornamentation deleted. Modules are decoupled so cleanly that users can not only customize features themselves, but also disassemble Kimi into parts and assemble their own applications.

Even Kimi Agent once styled itself "OK Computer" in its early days — though the name was eventually forced to change due to its high barrier to传播 (communication). But the namer seemed to care nothing for internet rules of traffic maximization, obeying only private musical taste and linguistic洁癖 (fastidiousness).

Someone half-jokingly said that if ranked by the proportion of employees who play musical instruments, Kimi might place first among AI companies.

Taste becomes the highest, and hardest to achieve, hiring standard. It cannot be quantified, yet it is everywhere.

Kimi founder Zhilin Yang

---

Generalize, Then Evolve

You may never figure out what exactly each person at Kimi is doing.

The company likes to use the word "team" to describe division of labor. Broadly, directions like algorithms, product R&D, growth, strategy, and functional roles are roughly clear. But once you drill down to so-called "departments" or even specific responsibilities, no one can say for certain.

Because you're facing an organization with no departments, no ranks, no titles, no OKRs or KPIs, with reporting lines so simple they seem unreal.

For Brandon — who graduated from Tsinghua University for both bachelor's and master's, had served in management at Silicon Valley giants and Chinese "major internet companies," and built a unicorn startup — this was simply incomprehensible. He had immersed himself in the industry for years, excelled at technical management, once led teams of nearly a thousand people, and had wanted to leverage his experience to make a splash in AI. But he was told by co-founder Yutong Zhang that the company didn't work this way, because the number of people she could give him to lead was — about two.

Driven by some intuition about the future, he wanted to talk further.

So in January 2025, on a long night of spreading doubt and restless hearts, Brandon met his Tsinghua junior, founder Zhilin Yang. The former couldn't have known then that the latter's name would today be so frequently mentioned alongside Elon Musk and Jensen Huang in the media. All he remembered was the first thing this junior said after寒暄 (exchanging pleasantries):

"Senior, RL is the future."

After that, the conversation became more like Zhilin Yang's murmuring to himself — he was so deeply immersed in his own thinking that Brandon couldn't understand most of the Chinese. But he couldn't deny that for the first time, he realized the knowledge structure and思维模式 (thinking patterns) he had built over twenty years were collapsing on the eve of a revolution, along with all his Ego. As for what ultimately made him decide to join, he told me somewhat mysteriously: Zhilin Yang might become a great prophet, because he is sufficiently visionary, and sufficiently pure.

Later, when this company that doesn't use titles hesitated because they truly didn't know what role to give him, his firm response didn't sound like a joke: "Even if you have me clean toilets, I'll come. And I'll clean them to be the best."

Not all major company managers and specialists can thrive here. Phoebe is a post-00s girl who transferred from the growth team to product R&D, calling herself a "clueless punk kid." She told me earnestly that at this company, rich experience and deep credentials can become burdens — the AI industry is too new, changing too fast; a seasoned expert might not learn and grow as fast as she, this "clueless punk kid."

She has seen at least three mid-to-senior level people from major companies fail their landing here. One ultimately decided to spend his remaining years in another industry, because he found the people around him extremely young and extremely smart, and after being crushed again and again, he completely broke down — this wasn't his era or his industry, better to accept fate and lie flat.

After the DeepSeek moment, Phoebe also developed a deep sense of crisis, determined to completely abandon research on user acquisition, wanting instead to help the company on product and R&D. She began endless知识恶补 (frantic knowledge catch-up), even livestreaming her studies on Bilibili for hundreds of hours. But what surprised her was that the company gave her the transfer opportunity without any hesitation from the start.

In fact, among just the thirty colleagues interviewed, over half had their job responsibilities change multiple times. Compared to their previous jobs, this ratio probably reaches 80% — meaning almost everyone at Kimi is doing something completely different from before.

Kimi likes people with generalization ability.

In the context of AI, generalization ability refers to a model's capacity to remain effective in new scenarios beyond its training data — it's not rote memorization of answers, but capturing underlying structural patterns. Those mid-to-senior level people from major companies have been trained too long in giants' specific KPI systems, reporting jargon, and resource博弈 (gaming) rules; their algorithms have overfitted to local optima. When environmental variables fundamentally change, their abilities may fail in adapting to new distributions.

If traditional major company employees are like specialized models, then the individuals Moonshot AI pursues are like foundation models: mastering basic rules through SFT, then using RL to self-play across diverse tasks, ultimately acquiring cross-domain transfer ability.

Cursor's founder publicly apologized and acknowledged Kimi K2.5 as the "Strongest base model" in his evaluations, the soul of his core product.

James is a returnee from Silicon Valley, twenty-six years old, whose dream is "to give money to young people." As a devout and狂热 (fanatical) believer in artificial intelligence, he considers his physical body merely a sensor for his Agent to collect information. When playing League of Legends with friends, he simultaneously records audio and collects his own physiological data like heart rate and pulse, analyzing which teammate's words affected his emotional state and operational performance. His views are sharp to the point of near extremism:

Anyone over fourteen learning a completely new language cannot achieve native-level mastery. AI is the same.

Dan, who joined right after graduating, experienced "knowledge anxiety" for the first time. In school he had at most toyed with "toy-grade" models — 7B parameter small models, finished in days with 32 GPUs; now he must harness a hundreds-of-billions-parameter MoE architecture beast, facing an ocean of trillions of tokens — equivalent to jumping directly from a small pond into the Pacific. To conquer this hard bone, he entered a self-abusive learning mode, his作息 (schedule) completely disrupted: working on Beijing daytime following Silicon Valley's late night rhythm, then switching to Beijing's late night during Silicon Valley's daytime, staring at training monitoring screens for hundreds of hours without daring to blink, like a stock trader watching the market.

The real challenge wasn't the workload — it was that he had to wear three hats simultaneously: algorithm architect, designing optimal solutions in the labyrinthine maze of model structures; systems engineer, troubleshooting in the quagmire of distributed computing like repairing a pipeline system that spanned the globe; and data curator, "alchemizing" through oceans of data — delivering hardcore benchmark performance while keeping the conversational experience soft and pliable.

Mid-training, they suddenly had to perform "internal surgery": critical parameters had been stored in half-precision (bf16), but numerical anomalies spiked wildly, threatening to spiral out of control. The team made a snap decision to switch to full precision (fp32) halfway through training to stabilize the situation — like changing running shoes mid-marathon. Dan said that someone who only knows algorithms, or only systems, or only data cleaning — a specialist with single-point expertise — cannot build a top-tier model.

There's no "that's not my job" here. It demands that you synthesize three completely different worlds: algorithms, engineering, and data. It's like working several jobs at once. This cross-dimensional tempering allows a person to evolve in extremely short timeframes what would take others years.

So for anyone trying to join Moonshot AI, the test is brutal.

Though there are no OKRs or KPIs, no office politics or managerial PUA, not even clocking in or attendance checks, if you aren't "AI Native," if you can't generalize, if you can't do RL — then you won't find meaning in your own existence.

The night before the Kimi K2 model release

---

"No Corporate Bullshit Here"

Most brands want a story. But every Moonshot AI colleague kindly warned me:

Don't write about Pink Floyd, or that piano sitting at the company entrance.

They feel that those who get it already get it; those who don't needn't be told. From Moonshot AI to Kimi, neither name's origin has anything to do with technology or AI. But if a company overemphasizes its bond with rock art, it becomes pretentious. People prefer beauty that doesn't know itself.

Win is a post-00 who escaped Big Tech. He told me, it's so strange here — you can actually get work done without meetings.

At his previous company, days were for meetings; nights were for actual work. He arrived at a simple realization: if your main energy goes into coordinating production relations, there's little room left to improve productive forces.

This is characteristic of some AI Native organization. Over ten employees explicitly stated that they increasingly dislike interacting with real people, preferring to communicate and collaborate with AI because it's more reliable and simpler. This also relates to the company's overall introverted personality. Someone used a cuter word: shy. Everyone can be a social butterfly in group chats, yet remain taciturn in person. Moonshot AI doesn't have many cultural activities — besides the annual party, the most recent was organizing office massages.

Introversion doesn't mean rejecting communication or lacking vitality. Though no interview was anyone's task or obligation, I never received a single "no" throughout. In group chats, massive amounts of information flow rapidly every day, alongside various abstract emojis. No one's remarks are met with awkward silence.

If you want to accomplish anything at the company that requires others' assistance, it's simple: just ask. No need to go through managers, no approvals needed, no coordination meetings, no breaking down "department walls." Moonshot AI has no department walls — it doesn't even have departments.

Zhilin Yang's personal signature is just four characters: communicate directly.

But everyone admits the company has been changing since day one. Some changes are proactive, some reactive, some even feel like getting slapped in the face. From massive user acquisition to model-focused; from insisting on closed-source to rapidly open-sourcing; from Chatbot to Kimi Agent, Kimi Code, Kimi Claw; from consumer to enterprise back to consumer... Not all changes withstand scrutiny.

In Ezra's mind, one through-line remains constant: respect for facts. She knows that all changes have one reason and one purpose only — to make the company's development better conform to the demands of objective laws.

The company allows everyone to have ego, but dislikes recruiting people who place themselves above facts. From co-founders to every colleague, people are persuadable. As long as facts are clear enough, people are willing to acknowledge their limitations — and so is the organization.

Ezra said it's this extreme pursuit of reality, truth, and verity that lets people speak honestly. Because truly smart people don't have their pride wounded by honest words.

Another necessary condition for radical candor: there's no horse race mechanism here, no zero-sum games, no conflicts of interest. Every member willingly shares their research findings and technical details for free. Just as the company had its own community early on, it now advocates community culture. The sharing of information and knowledge accelerates collective learning and collective progress — and ultimately, everyone benefits.

Win said toxic culture is contagious. Good culture is too.

Someone used "solidarity" — an archaic word long unused to describe a company — to capture this state. In reality, Moonshot AI has always faced a harsh survival environment: external competition from giants, internal pressure from Big Tech, limited compute resources. But these adverse factors are intensifying the company's cohesion. At root, people are the organization's only important asset.

Recently, Florence was poached by a peer company offering double her salary. She refused without hesitation. Her reason for staying was simple: "No corporate bullshit here."

The company's new location (JD Technology Building)

"I Don't Know How She Made It Through"

Early in my interviews, facing some of the smartest, most AI-literate people in the world, I was incredibly nervous: as a humanities student, I'd never worked in tech, and my understanding of AI was superficial at best.

But when I actually started talking with young experts from the algorithm and product R&D teams, I discovered they were even more nervous: they were terrified I'd feel embarrassed for not understanding the jargon they used.

So everyone carefully translated English into Chinese first, then translated that Chinese into Chinese I could understand.

This protectiveness was deeply moving. I thought of the company's only request before interviews began: protect everyone.

So I avoided overly sensitive questions or ones that might sting. Even so, when Ty spoke with me by phone, he revealed a barely perceptible emotional tremor. When he'd first joined and was getting settled, he encountered major difficulties and even felt he couldn't persist, entertaining thoughts of leaving. Until one weekly meeting, when he saw Annie — a girl just two years out of college — who, after who knows how many rounds of setbacks and internal doubt, had finally pushed a certain project to substantial heights. He felt he couldn't give up either. After all, he was much older, with far more life experience, yet his mental resilience paled in comparison. He reflected: "I don't know how she made it through."

It wasn't just him who'd thought of leaving. Annie had too. For a considerable stretch, she needed to build something from 0 to 1 in an overseas segment, with no breakthroughs. To make matters worse, colleagues from other teams, meaning well, directly advised her to "abandon this meaningless effort." She said she'd cried more tears for Moonshot AI than in her entire life — never for another company, never for any ex-boyfriend.

She wasn't short on opportunities; she'd even received a better offer elsewhere. But she couldn't convince herself to sell her soul to someone else. She wanted to talk with Yutong Zhang one more time.

After they talked, she decided to stay. She didn't tell me what was said, only: Yutong is the toughest, fastest-iterating, highest-ceiling boss I've ever seen. Following her, I can reach higher limits. Then she added: "I don't know how she made it through."

When you've gathered enough information, you notice how often certain phrases repeat. And the most repeated phrases tend to sketch the team's shared character.

Bob was pulled back to China by Zhilin Yang to start a company, giving up his PhD opportunity in the United States. He joined on day one, representative of those who know the company best. When asked the question everyone gets — what do you think is the team's most important quality? — he thought for about two minutes, then answered with two words: resilience.

For a company just three years old, any emphasis on resilience is luxurious, but no less sincere. He believes that smart and brave are sometimes antonyms — the smarter you are, the easier you see risks, and thus the easier you give up. And stupid persistence cannot succeed. So only those who see the truth clearly, calculate the probability of failure, and still press forward — that deserves to be called resilience.

A story once circulated internally about "three entries into the Repentance Cliff."

In May 2023, Freddie and colleagues received a seemingly impossible task: make AI capable of reading 128K-length text in one go (equivalent to hundreds of pages), while the industry standard was roughly 4K. He quickly designed MoBA v0.5, but it required rewriting the underlying training framework and the main model was already halfway trained — too costly. The plan was shelved. This was his first entry into the "Repentance Cliff."

Half a year later he returned with v1, revised to allow continued training from the existing model. Small-model validation succeeded, but the large model hit loss spikes — training loss suddenly surged — and no amount of debugging worked. The project retreated to the "Repentance Cliff" a second time, for another half year, even missing the company's 200,000-character release milestone. But the team didn't disband. The company launched "saturation rescue" — mobilizing technical heavyweights for collective assault, rewriting underlying logic, finally allowing v2 to stably pass the "needle in a haystack" test.

Just as they were about to go live, the third blow came: during SFT, long-text summarization tasks performed poorly due to overly sparse training signals. By now the project had incurred massive costs, but engineers retreated once more to the Repentance Cliff to find solutions, finally resolving it by adjusting the attention mechanisms in the final layers.

Three retreats, three returns. At the interview's end, I posed Freddie the ultimate question: how should one describe this company?

He too answered with just two words: moon landing.

Why moon landing? He quoted that famous speech:

"We choose to go to the moon in this decade and do the other things, not because they are easy, but because they are hard."

"We choose to go to the moon in this decade and do the other things, not because they are easy, but because they are hard."

The company's meeting rooms are all named after bands.

Genius Swarm

In the end, I didn't disturb or attempt to pry into any of the co-founders. Externally, they remain in stealth mode — they don't like giving interviews, and have no interest in fame. But internally, they're present almost everywhere.

Under an极致扁平的架构, you need superbrains as support; otherwise, vitality devolves into chaos. With no middle management layer, each co-founder directly handles 40 to 50 colleagues, embedded deep in the technical and business frontlines, ensuring tight alignment between decision-making and execution.

Though all five co-founders hail from Tsinghua University, the bandwidth and managerial radius of carbon-based life forms have hard limits. When a company reaches a 120-billion valuation and expands to a team of 300-plus, even superbrains begin to overload.

The overload isn't limited to co-founders. This is an infinite game driven from within: every team member must justify roughly 400 million in per-capita valuation, meaning they need to deliver human efficiency far exceeding that of most companies.

The revolutionary variable is tooling. Working hours at Moonshot AI aren't extreme — employees are allowed to wake up naturally, and pulling all-nighters isn't expected. Leo, on the product team, says he commands a virtual army of thousands. Picture this scene:

10 a.m. Leo wakes up and walks into the office. His task: synthesize user feedback from five global markets over the past 24 hours, and set this week's iteration priorities — something that used to require three people and two days. Leo launches three Agents. A strategy Agent filters 3,000 pieces of feedback for high-priority needs related to "long-context interruption." A translation Agent parses Japanese dialect and Korean honorifics in real time, tagging true emotional intensity. A competitive-intelligence Agent simultaneously scrapes Cursor and ChatGPT's updates for the day, generating technical comparisons.

Leo does only three things: vetoes one misclassified sarcastic comment, flags one screenshot containing an unreleased UI, and confirms the top-three needs recommended by the Agent. By 11:30 a.m., the PRD is done. And his code Agent has already auto-generated 70% of the foundational framework based on the requirements, waiting for the afternoon's discussion with human engineers on creative solutions.

Humans set the rules; silicon executes them. The organization becomes a vessel for algorithms. The ability to fluently deploy Agents and deeply integrate them into workflows is what it means to be an AI Native company.

The model isn't just the destination; it's also the tool. Whether directly empowering productivity on the technical front, or fundamentally restructuring management models at the organizational level, AI's genes are already etched into this company's bones — like it or not. Just as with the Agent Swarm it develops, the team itself is essentially a Genius Swarm: each Genius working in parallel independence, seamlessly coordinated with the others.

Yet such a flat organization carries structural fragility. When asked, "Is this model sustainable as the company scales from 300 to 3,000 people?", interviewees mostly gave cautious answers. After all, historically similar experiments in radical flatness — Holacracy, Haier's micro-enterprise chain groups — tend to hit decision bottlenecks once scale exceeds 500. When information nodes multiply, "direct communication" devolves into information overload.

A more immediate pain point is the sense of weightlessness at the individual level. The absence of rank as a buffer means directional chaos transmits directly to everyone. One former employee who ultimately returned to a major tech company put it bluntly: there are no top-down OKRs or KPIs here. Sometimes you walk into the office in the morning not knowing what you should be doing, and no one proactively tells you how you're doing — this lack of feedback, this insecurity, made some people nostalgic for the clear reporting lines, defined review milestones, and quantifiable deliverables of big tech.

After all, those seemingly tedious rituals actually provide individuals with a baseline of certainty: where the goal is, what counts as done, how performance is evaluated — everything visible and clear. This isn't Stockholm syndrome; it's basic organizational mechanics.

If Alibaba resembles a precisely calibrated promotion assembly line, ByteDance a goal-obsessed military corps, and Tencent a more fault-tolerant professional academy, then Moonshot AI is a primeval forest: geniuses may find their hunting paths, but ordinary people may wander lost in the fog.

Kimi Agent conducting in-depth industry research

The Necessary "Dual-Vector Foil"

No departments, no ranks, no performance reviews — the AI Native organizational paradigm is anti-establishment, non-structural. On this, the tech giants can no longer pivot; smaller companies have missed their window for self-aggrandizement. This is asymmetric warfare.

In The Three-Body Problem, the Singer civilization casually tosses the high-dimensional weapon "dual-vector foil," flattening the solar system from three dimensions to two. Planets, stars, humanity — all collapse into a picture without thickness. Earth is destroyed.

Moonshot AI is actively releasing this dual-vector foil upon its own organization. Not to destroy rivals, but to push organizational efficiency to the extreme: no hierarchical depth, no departmental walls, no three-dimensional entanglement of office politics — only "model" and "intelligence" intersecting in the most brute-force manner.

In AI era's field of compulsion, every startup is forced to cast the dual-vector foil at itself. The surge of one-person companies is essentially a generational explosion of AI Native talent: when technological empowerment collapses organizational capability to an individual singularity, all intermediate managerial buffer layers evaporate instantly. The organization is flattened; no depth remains for maneuvering. Everyone is forced to confront the problem itself.

This is the iron law of organizational paradigm evolution across the entire business world: everyone will be folded.

When people are exposed on the same plane, one person's super-radiation to fifty ceases to be a management spectacle and becomes an organizational norm. The distance from center to periphery is redefined. Elites dependent on hierarchy and OKR coordinate systems suffocate immediately, while geniuses violently deconstruct intelligence on this exposed plane, with guardians sweeping away all entropic noise — not without humility styling themselves as pioneers expanding the boundaries of human civilization.

Yet from three dimensions to two, this process cannot stop, cannot reverse.

From here, the Kimis cannot look back. Every strategic adjustment is high-risk chaotic iteration. Rivals may still turn slowly in the labyrinth; if Moonshot AI attempts to accelerate organizational expansion, it will only cause internal structural tearing. And all this self-dimension-reduction serves only to complete one crazier dimensional leap.

The endpoint of organizational dimension-reduction is intelligence dimension-elevation.

Only by pushing model intelligence past the inflection point, elevating it to sufficient height to escape the gravity well of all carbon-based organization, can Moonshot AI flatten all competitors' organizational advantages in one stroke — granting ultimate legitimacy to this irreversible dimensional gamble.

By then, discussions of managerial radius or architectural form will lose meaning — just as the Singer civilization doesn't care what dimension it occupies, because the advancement of dimensional weaponry itself defines new rules of war.

At that point, "Moonshot AI" will transform from metaphor to reality: they become the high-dimensional light source illuminating the dark side of the intelligence universe, and all past organizational pain merely the ablative heat shield burning away as the lunar module punches through the atmosphere.

Either ascend to godhood through dimension-elevation, or be sealed away in collapse.

There is no third path.

(All English names in this article are pseudonyms.)

Moonshot AI's Yutong Zhang Shares Her AI Entrepreneurship "Chain of Thought" for the First Time | Buming Entrepreneurship Camp

Genspark CEO Kun Jing: Stop Asking How to Implement AI, First Give Every Employee Unlimited Token Access

Zhihui Jun: The Most Robot-Like Human Built a Robot | BlueRun Ventures Family Headlines