The World Through the Eyes of a Product Fundamentalist | In Conversation with Li Auto SVP 范皓宇
Summary
范皓宇 reduces the job of an “originalist product manager” to observing, understanding, gathering enough elements, and then spending maximum effort compressing them into the optimal solution. He acknowledges that when growth is the goal, you can look only at the data, but warns against teams rushing through the most important thinking; the real edge comes from seeing more raw material than others and combining it into outcomes that surprise users: “Data and metrics are a crutch, but you can’t walk by leaning on a crutch.”
Chasing growth for its own sake traps products, content, and users in a feedback loop of ever-higher stimulation, until both sides of the market are exhausted. Usefulness is merely the “door opener”; long-term value comes from whether a product can stay with users and keep delivering the feeling of “Wow, it can do that too.” 范皓宇 is clear about the trade-off: if 10 things cannot coexist, cut several and keep only the parts where “useful and interesting coexist.”
AI’s biggest product opportunity is to turn standardized, persona-based supply into real-time service for each individual. When a system can combine location, companions, current state, and behavioral history to understand “I want baozi right now,” the boundaries between maker, user, and designer may dissolve, while new agents must earn long-term trust and feel like “it belongs to me.” For investors, that shifts value away from a single function or traffic入口 and toward contextual understanding, persistent connection, and end-to-end delivery.
China’s consumer market is not moving uniformly upmarket or downmarket; it is a mille-feuille of segments running entirely different races. Practical needs in first-tier cities are already oversupplied, while users lack understanding, care, and a sense that their lives have meaning; lower-tier markets still have large unmet needs for basic services. The opportunity therefore exists both in products that are “more reasonable, cheaper, but better to use” and in taking mature services further downmarket and overseas.
When data conflicts with taste, 范皓宇 trusts his own instincts, goes left, and accepts responsibility for the business outcome. His reasoning is that categories, metrics, and axes all contain subjective judgments, while products are ultimately shaped by a continuous sequence of decisions; the real danger is not getting one decision wrong, but repeatedly wavering and failing to build momentum. “You need some beliefs to believe; once you believe, you have the willpower to persist.”
In the AI era, the medium-term moat is more likely to be the organization than any specialized link in a traditional production line. As the path from need to outcome compresses dramatically, product managers and designers must own the work end to end; after seeing a designer independently complete everything from asset slicing to account registration, 范皓宇 immediately concluded, “Get in now.” His conclusion is blunt: “If you don’t revolutionize yourself, you’ll be revolutionized.”
范皓宇 remains intensely optimistic about China, based not on short-term data but on rising taste, supply-chain capabilities, and a persistent desire to change how people live. He believes Xiaomi, new-energy vehicles, and others have brought price, quality, and taste into mass-market conversation, and says plainly that “with so many valuable companies in the A-shares market trading at such low prices, we have a huge spread opportunity.” Short cycles will fluctuate, but the underlying drive of Chinese people—“light in their eyes,” willing to use new tools to “change the world”—has not disappeared.
Deep dive
1. A 25-year zigzag career, always anchored in product and design
范皓宇 started designing in his freshman year in 2000, then moved through feature phones, 3D MMORPGs, Unreal and Epic engines, self-media, Alibaba’s operating system, mobile UX, and automotive projects at Alibaba and SAIC before eventually joining what was then still called “Chehejia,” now Li Auto.
On the day Kung Fu Hustle premiered in 2004, he arrived in Beijing and entered a mobile-phone company the next day. He quit graduate school after 6 months because he could not stand merely teaching classes and doing academic work; he “really wanted to achieve something.”
He joined Li Auto first because he wanted to return to Beijing, and only second because of the company itself. The smart connected-car wave was already unmistakable, and he happened to know the team, so he brought his mobile, software, supply-chain, and user-experience capabilities into automobiles.
2. Beijing shifted from an idealist battleground to an efficiency lab, but the energy remains
To 范皓宇, Beijing’s greatest appeal is its diversity: people arrive with dreams and enter this battlefield, while even roadside barbecue conversations can cover investing, IPOs, and creation. “Shanghai is certainly better for living, but the people in Beijing are more interesting.”
Before and after the pandemic, café conversations shifted from “What project should we build?” to “How do we improve efficiency, and who do we lay off?” As the space narrowed and the sense of security deteriorated, contradictory demands for both consumption upgrades and downgrades emerged.
曲凯 noted that more and more people are leaving Beijing. 范皓宇 acknowledged that the pressure is real, but cautioned that perspective also changes with age and management responsibility. Every school-opening season, young people still stream in dragging suitcases, hoping to “create something impressive”—the possibility that major cities continue to offer.
3. One generation’s upper body embraced consumerism while its lower body still obeyed order
范皓宇 describes his generation as people who had “all been artistic, more or less.” They lived through 9/11, China’s WTO accession, globalization, and the sudden arrival of consumerism, while retaining the bodily memory formed by an era of planning and discipline.
His metaphor is that “the lower body remains deeply obedient to order, while the upper body has become a way to prove oneself by buying products.” Even today, if he sees a faucet running, he instinctively turns it off.
From Nokia feature phones and NEC point-and-shoot cameras to iPhone, Android, mobile internet, and the rapid jump from 2G to 4G and 5G, this generation has been riding a high-speed elevator. Rapid growth once concealed problems; after the slowdown, anxiety and an indistinct future became much more visible.
4. China’s past leap was powered by 3 layers of movement
At the bottom, rural workers moved into cities; the second layer was the “East Asian Mediterranean” supply chain formed by Taiwan’s coast, Japan, South Korea, and Southeast Asia; only at the top came the technology and internet stories that usually receive the attention.
During an internship in Wenzhou in 2003, 范皓宇 designed a lawn mower for a company that sold exclusively overseas. He had never used one himself but had to simulate overseas demand; the company was already shipping products around the world. It was a concrete snapshot of export-oriented manufacturing.
Overseas talent brought back internet technology and universalist values, while the expansion of domestic universities allowed more people to enter the technology sector directly. 范皓宇 also pointed to a new trend: Chinese management cadres are beginning to move overseas. They understand supply chains and coordination across hierarchies, and can even operate factories independently.
These 3 layers together shaped his enthusiasm for AI: when new technology appears, development should keep moving forward and make the world better. That is still the originalist spirit of the early internet—connect people and knowledge first, then talk about monetization and growth.
5. Vertical AI agents may follow feature phones toward unification
Early mobile phones were split into music phones, camera phones, and office phones, much like today’s agents built to handle one task each. The fragmentation was not the endpoint; it was the specialization forced by weak underlying capabilities.
From this, 范皓宇 concludes that once capabilities mature, these vertical products “will definitely converge before too long.” After convergence, costs fall and extensive customization is no longer necessary; software, platforms, and services can then provide differentiated capabilities for different users.
This is not merely a historical analogy but a product judgment framework: having experienced one round of category fragmentation caused by insufficient capability makes it easier to recognize the direction of the next capability integration.
6. The originalist product manager expands the raw material first, then compresses complexity to the limit
In his junior year, he designed a fingerprint-recognition device for a military research institute. Seeing cluttered desks in a secure office and equipment constantly going missing, he made the device into a teardrop-shaped roly-poly: “You flipped it over—smack—and the thing popped out by itself.” Users later found themselves playing with it.
The key was not the shape, but noticing everyday details that nobody had included in the requirements. Strong product people start with more elements than everyone else, then combine them into something useful, interesting, and intelligible.
By “originalist,” 范皓宇 means observing and understanding first, then spending maximum effort on compression to find the optimal solution. He rejects “It’s roughly good enough—just run an A/B test, find PMF, and scale” as a substitute for deep thinking.
His mind works like a detective wall: photos, clues, and connections keep accumulating until they produce an unexpected but coherent result. When people open a book, watch a film, listen to a podcast, or use a product, what they are really waiting for is that small moment of surprise.
7. The growth feedback loop drains creators and users alike
If growth is the only objective, 范皓宇 says you can simply look at the data. But version-testing cycles keep getting shorter, forcing suppliers to track retention, plays, and other metrics continuously, with the frequency rising all the time.
He observes that top bloggers with millions of followers and high endorsement fees are mostly not relaxed or enjoying work-life balance; they are busy and anxious. The closer they get to and the more they cater to a mass audience, the more they need the newest, strongest, and most differentiated stimulation to capture attention.
Users, meanwhile, are like people receiving one injection after another: satisfaction, numbness, then a demand for a stronger hit. Short videos and short-form dramas push the mechanism to its extreme. Eventually the supplier is exhausted, the audience is exhausted, and both sides may stop to reconsider.
曲凯 compares the process to a pendulum, arguing that people will eventually rebel against infinite stimulation and return to a healthier state. He emphasizes that A/B testing is a phenomenon; what matters is who uses it and how. 范皓宇 uses “circle” and “the old order will be back” to explain how order repeatedly returns.
8. Usefulness gets you through the door; interest determines whether a product stays
范皓宇 starts with a bottom line: if a product is useless, users will ignore it; even a toy has a clear purpose for a child. But quality is not determined by a feature list. It depends on whether the product can stay with someone over time and keep making them think, “Wow, it can do that too.”
Once an experience becomes internalized over time, the product forms a relationship with the user. If films were only about box office, finding a topic and hiring stars would be enough. Yet the film recently winning audiences’ votes may be the unnamed Journey to the West-themed animation that cared more about story and resonated with the present.
This creates 2 types of demand: one group says, “Stimulate me faster; give me something new,” while another is willing to savor the process. A good commercial product cannot choose only one side; it must use practicality to carry interest.
9. AI may collapse the industrial-era triangle back onto the individual
The British design museum’s recurring triad—maker, user, designer—made 范皓宇 realize that 100 years ago, the person who made a hammer, designed it, and used it might have been the same person. Industrialization first separated maker from user, and modern design then separated designer from maker.
The separation temporarily sanctified manufacturers and designers, leaving users to accept standardized supply. Users then began to “vote with their feet,” while makers and designers started asking, “What do you need? What should I make for you?”
AI could reshape the triangle further. Instead of opening Meituan and choosing baozi from a set of common options, the system could read the 3 people in the home, the current context, and their eating history over the past 3 days, then generate and fulfill that specific service directly.
范皓宇 believes design exists because products cannot serve individuals one to one and must instead target the largest personas, such as men or women. Once each person’s precise needs can be solved continuously, traditional design itself may dissolve.
10. After excess capacity, the direction turns toward the specific user
Bauhaus-style standardization arose from scarce resources and limited production capacity: manufacturers could bend steel tubes into reproducible chairs so ordinary people could first reach a basic standard. Today, China’s supply chain has “no problem manufacturing for the entire world; doubling capacity would be no problem either.”
Once excess capacity becomes the pressure point, the answer should not simply be another standardized product covering more personas. It should let each person receive different services based on their environment, state, and history.
范皓宇 stresses that this personalization is not expensive customization once reserved for kings and nobles. It is mass-scale, one-to-one supply supported by a universal framework, and it must be continuous, useful, and interesting at the same time.
11. Chinese users are a mille-feuille of needs running on entirely different schedules
范皓宇 cites a People’s Daily survey in which around 80% of respondents considered themselves part of a disadvantaged group, including bosses, executives, and leaders. The example reflects a shared feeling of pressure, busyness, and “my time does not belong to me.”
In first- and second-tier cities, supply on the usefulness dimension is already excessive: ask for a shovel and options from RMB5 to RMB5,000 appear instantly. After opening a food-delivery app, the real challenge is deciding what to eat. What remains scarce is being understood, cared for, and having one’s life affirmed.
Third- and fourth-tier cities remain embedded in family and social relationships, with higher floors of security and a higher minimum standard, but their upside may be constrained. First-tier cities pursue a higher ceiling, but may offer less presence and security at the floor. Everyone is like “a battery, working hard to generate power.”
In more downmarket areas, even basic services remain uncovered. 范皓宇 uses a race as the metaphor: first-tier cities are at 800 meters of a 1,000-meter track, while some people are still at 100 meters. The former need meaning and recognition; the latter first need products and services to truly move downmarket.
12. Emotional value exists even in scarcity; the difference is the meaning system
曲凯 uses pizza as an example: under economic pressure, parents may deconstruct it into flour and meat, seeing the use value of a RMB100 pizza as similar to cheap mantou stuffed with meat. 范皓宇’s rebuttal is that this does not justify labeling the previous generation as pure utilitarians.
His parents’ generation ate mantou and noodles in daily life, but still drank and enjoyed “small luxuries” when friends gathered. Old photo albums and toys likewise carried style and the flavor of a life. “There were ways of playing in an era of scarcity.”
People always pursue meaning that is “just a little bigger than themselves.” In an age of excess supply, a Muji-style return can emerge instead: stripping away decoration and branding, returning to raw materials. This is another swing of the pendulum in the opposite direction.
范皓宇 summarizes the cycle as “the old order will be back”: consumerism keeps pushing differentiated demand higher, but order, restraint, and a desire for things themselves are always brewing underneath.
13. Mass-market taste is rising; the economic cycle has not erased the urge to create
范皓宇 suggests lining up the best-performing holiday releases of the past 20 years. The content audiences recognize today is visibly different from what they accepted 10 years ago. Short videos, polarization, and populism do not prove that the public is mindless; films instead show that taste is rising.
Xiaomi and new-energy vehicles have brought price, quality, experience, and taste into the mass market at the same time. He even overheard 2 men leaving a gym seriously debating which car had no taste and which had more taste—something that could not have happened when supply was scarce.
He sees an economic downturn as a cycle, but believes Chinese people’s responsibility to their families, desire for a better life, and work ethic will not disappear. “There is light in their eyes,” they “want to get things done,” and want to “change the world”—the foundation of his optimism about China’s domestic product market.
That optimism extends to the capital markets. He says plainly that the A-shares market has many valuable companies trading at low prices, creating a “huge spread opportunity.” But this is an expression of conviction; the conversation did not discuss specific names or valuation methods.
14. Taste is a marker of distinction; truly scarce is original content
范皓宇 defines taste as a form of distinction. Saying you have no taste places you in the utilitarian camp; claiming taste signals that you understand life, design, or quality. People must build a self-narrative through differentiation.
Behind taste is styling, and behind styling is content. Climbing, home design, beauty, and travel on Xiaohongshu can all form a style; podcasts and scholars have styles too. But most people are merely consuming and reproducing styles that already exist.
His extreme assumption is that 10 years from now, 100,000 pieces of content may be produced every hour, while perhaps only 5 create genuinely new styling; 90% will reproduce existing styles. Eating san bei ji at Lugang Town for the first time, or learning to order a Grande at Starbucks for the first time, is “consuming taste once.”
15. When data conflicts with instinct, he chooses instinct and accepts the cost
曲凯 asks: if every survey and operating metric points right while your own taste tells you to go left, what should you do? 范皓宇 first answers, “Trust your own feelings,” then makes it explicit: “Still go left.”
His reasoning is that “all data is subjective.” How metrics are defined, how the y-axis is set, and how categories are determined all introduce the observer’s judgment before analysis even begins. The direct feeling that a product is wrong may appear before the metrics do.
He compares an experienced product person to a chef. A good chef sees cabbage, radishes, meat, scallions, ginger, and garlic, and already imagines the wok hei, the plating, and how to present the dish to the customer. An inexperienced person checks “to taste” and “a pinch” at every step, only discovering the deviation after burning the pan.
曲凯 points out that taste can be wrong, or so far ahead that it cannot improve short-term operations. 范皓宇 does not evade the implication: “You have to accept that outcome.” The value of judgment is not always being right, but accepting the consequences of the decision.
16. Products are shaped by continuous decisions; willpower gets you through short-term noise
范皓宇 believes people often decide by feeling first and then search for logical reasons afterward. He agreed to record the podcast because he found 曲凯 interesting after talking with him; explanations about “training his abilities” came later.
One decision receives feedback and leads into the next, eventually forming a “sequence of decisions.” Life is determined by a series of choices that decide who you become, while products are determined by a series of trade-offs that decide how they exist in users’ lives.
The real danger is constant wavering: buying a home today, going to graduate school tomorrow, taking a job the next day, and going into business after that. Time is consumed by oscillation, and nothing accumulates in any direction. 范皓宇 sums up the answer to pressure in 3 words: “Willpower.”
A post-IPO share price and short-term data accelerate feedback, but he compares the process to keeping an appointment: once the destination is set, you have to arrive whether the day is clear or a typhoon is coming. The weather changes the difficulty of the route, not the final direction.
17. Idealism must be built on pragmatism
范皓宇’s boundary is clear: “If you are not a pragmatist, you do not deserve to be an idealist.” Charging at windmills like Don Quixote on emotion alone is suicidal; if you are trapped in hardship yourself, it is difficult to help others for long.
Commercial reality requires considering current sales, costs, and user feedback, as well as whether a product will mean something and what it will leave behind over the long term. Real product design is therefore not an either-or choice, but “all of the above.”
He wants low cost, strong performance, full use of the technology, and easy user access, while also making people feel that they have become better. If the job were merely to complete a requirements list, development could deliver directly; product design would not be needed.
曲凯 adds that willpower is produced precisely between 2 goals. 范皓宇 agrees that reality and the ideal world contain a delta, requiring “one hand on the left and one hand on the right” to pull the imagined good into reality: “Make something happen.”
18. Software and hardware are both locked in by existing structures and habits
曲凯 believes hardware takes 1 or 2 years to build and is difficult to change after launch, making it more like film. 范皓宇 argues that the essential difference is smaller: once a software architecture is set, it can lock in subsequent choices just as tightly.
User habits are even harder to iterate than hardware. QWERTY may not have been optimal at the starting point, but once it formed a habit at scale, later entrants found it difficult to change.
His analogy is that hardware is the house itself, which the next generation can still replace; software is the object that has sat in the home for years. “Don’t move it casually,” or you will directly interrupt a way of living that users have already internalized.
19. User needs begin with a creator’s hypothesis, not with blank-slate research
Users are like a mirror, but they are not a unified subject. Some people go to the cinema just to laugh, some want quiet reflection, and some want a new world like Avatar. No product can satisfy every person in the world.
范皓宇 insists that the starting point remains the creator: first have something you want to express or change, then observe users with a hypothesis. Ask a woman selling roasted sweet potatoes whether she wants an AI assistant, and she may care only about selling more sweet potatoes. Research detached from context does not automatically generate a business.
The path is to first verify that the need exists, then optimize the solution, expand reach, and finally let the returns cover living costs and the cost of the next round of creation. Product detail works the same way: users in the distance are naturally blurry, so you move closer and gradually prove that the hypothesis can work.
He divides user relationships into 3 types: top-down, “I am God; you accept”; bottom-up, “Tell me what you want and I will give it to you”; and the healthier relationship between friends—an equal dialogue in which users can push back and product people can still hold their ground, with both sides improving through understanding.
20. A product’s 4 voices span acquisition through the end of its life cycle
范皓宇 does not reduce product methodology to a fixed formula. He prefers the master-apprentice model, demonstrating on specific projects how many times to swing the hammer and from what angle. Product work is “a fluid, messy business”; if a formula were sufficient, the role itself would not need to exist.
He nevertheless offers an end-state evaluation set. The first voice, “Oh,” represents differentiation: it makes someone willing to approach a product among a crowd of similar offerings. Without that reaction, the product may not even sell.
The second voice, “Mm,” is the smoothness and satisfaction of first use, like staying up all night installing apps after getting an iPhone. The third, “Oh, wow,” comes months later when the user discovers another capability, such as playing a fruit-slicing app for the first time.
The fourth, “Holy shit,” comes at the end of the life cycle: you replace your phone, connect to Wi-Fi, and an hour later everything has automatically returned. Four reactions means god-tier product; 3 means an exceptionally strong product, 2 can produce a major hit, and 1 is at least enough to sell.
21. AI will ultimately test end-to-end organizations and their ability to handle real complexity
范皓宇 imagines that 20 years from now, ordinary people may spend 80% of their daily speech addressing AI because it participates in co-work, retains context, and reduces misunderstanding and the burden of communication between people. He emphasizes that this is only the high-probability direction; the specific roles will take shape step by step.
The new role should be trustworthy and become a long-term source of support, while “ideally not turning into a hegemony.” It should not merely buy baozi once or complete a single task, but build a lasting connection by feeling like it belongs to me, continuously delivering something useful and interesting.
曲凯 proposes that organization is AI’s medium-term moat, and 范皓宇 “completely agrees.” When there is almost no process between need and outcome, the assembly line divided into PRDs, icons, and asset slicing will fail; product managers and designers must both deliver complete results end to end.
AI has not suddenly made the world more complex. It has made previously invisible chaos observable and actionable for the first time. History once recorded only kings and generals; now new capabilities can confront vast amounts of data and situations, such as traffic. “Chaos is always chaos.” What has changed is that we can finally see it—and do something.