Summary

Level, Kuaikan Manhua’s new AI bet, is defined by 陈安妮 as “a real-time, living AI world”—she distinguishes it from Chatbots and Stanford Town, inviting users into a world of autonomous agents to experience and intervene in “fate.” Her core analogy is a paradigm-level one: “The invention of AI is like the invention of electricity”; electricity gave rise to film, which replaced theater, while layering limited AI capabilities onto legacy content “is like putting a few light bulbs on a stage play … its ceiling is nowhere near as high as film’s.” Since 2023, Kuaikan has tried interactive comics, then interactive comics plus AI, before ultimately “changing the foundation” to build AI-native.
Her view of AI content is explicitly contrarian: AI will not kill premium content; it will instead take the industry “back to when film and records were first invented—back then, only Teresa Teng could make a record.” The logic is on the cost side: running an agent world burns a huge amount of tokens, so “the content has to be good enough for this business model to work”; today’s popular AI short dramas and AI comic adaptations are, in her view, “products of the previous content era, already at the end of a lifecycle of vicious competition.”
Kuaikan’s financial position is laid out in full: its core business is profitable excluding AI investment, more than 50% of revenue comes from IP derivatives, and that IP strategy grew out of a contrarian algorithm decision that data could fall 30% within one to two months. In 2018-2019, she resisted internal arguments that “what users like is the only truth” and that “you have to trust the data,” removing content she did not believe in from the recommendation pool; data could fall 30% within one to two months, and some employees left. Three to four years later, the payoff became clear: “pure traffic-content companies were all eventually swallowed by Douyin,” while Kuaikan survived on IP derivatives.
Three brushes with death define her 12-year entrepreneurial story: in 2015, a defaulting investment partner broke the company’s cash flow, until an investor she had never met wired money directly into the company’s account; in 2021, she completed a nerve-racking $200M financing amid a macro upheaval. After the financing closed, her body stiffened with pain until she “could only lie on the sofa in the office and couldn’t stand up”—“my body was telling me how afraid I had actually been in those preceding months.” It stripped the glamour from fundraising: “What matters most is whether your product is truly creating unique value for users … and whether you are actually making money.”
Level’s product philosophy is built around “fate,” not observation: Stanford Town emphasizes a god’s-eye view, while she believes “what people truly care about is not other people; what they truly care about is stories and people connected to me.” For an agent to have a soul, it needs three things: rich stories, the agency to “say no” and even block you, and an evolving consciousness that chooses its own direction. Each world will have roughly 4-5 protagonists and a dozen-plus secondary characters; it will eventually move toward UGC, but not yet.
Level’s 0-to-1 paradigm development has no single leader; it iterates through “collective intelligence,” with her describing her role as having shifted from decision-maker to “a user submitting queries to a large model,” using a voting mechanism that allows only yes and hope, never no, to advance ideas. In one icon-design example, a product manager used yes-and-hope feedback, and the designer “came back the next day with nearly 100 options.” Her remaining role is to give “signal weighting” to non-consensus ideas: “My role has become fine-tuning.”
She calls the culture that supports disruptive innovation “the spirit of play,” rooted in the Diamond Sutra’s “let the mind arise without dwelling anywhere”—not fearing failure, carrying no burden, but without losing one’s sense of purpose. When she decided late last year that Kuaikan could no longer do “comics plus AI” and had to ask “AI plus what,” she believed past experience could become baggage; host Koji countered with the iPhone’s reuse of the iPod’s industrial design. She replied that defining a new paradigm comes mainly from understanding the technology and current user needs, while the means of execution, such as storytelling, and deeper beliefs can still carry over.

Deep dive

1. Level: A Real-Time, Living AI World Where the Product Is “Fate”

The opening rapid-fire exchange sets the coordinates: 陈安妮 is 34, ENTP, and has been building companies for 12 years—she founded Kuaikan immediately after graduating at 22. The team has 300-400 people. She sums up the new product Level in one line: “a real-time, living AI world” where “fate experiences you can change” emerge.
She uses Westworld to introduce the first experience: choose a world—a mystery-and-power-struggle train, a high-school campus, and so on—and enter to find “a living map” filling your field of view, with an onscreen panel continuously recording the movements of every life. Koji asks whether these are real people or NPCs. Her answer is the philosophical foundation of the product: “Every person is a living soul.” They are all agents, but “how do you define a ‘real person’?… The key may not be whether it is carbon-based,” but whether it has autonomous consciousness.
Fate works through the butterfly effect: an officer is supposed to detonate a bomb, but if you form a friendship or fall in love with him, you can change the train’s fate. Every person’s fate, and every playthrough’s, is unique—“If you went back in time and suddenly slapped me, do you think fate would be the same?” After the user leaves, the world continues running autonomously.

2. The Difference From Stanford Town: You Have to Be Part of the Story

Faced with comparisons to Stanford Town, Civilization, and Nintendo’s Tomodachi Life—all of which simulate worlds—she acknowledges functional overlap but says the experience is entirely different: “Stanford Town puts a strong emphasis on the user having a god’s-eye view, whereas we emphasize that the user is an important character within this world and story.”
This is the sharpest product judgment of the episode: “At its core, people don’t really care about other people. What they really care about is stories and people connected to me. You have to be part of the story.” Stanford Town, she argues, has not yet delivered what users actually want from entering a virtual world: an experience of fate. “Fate is essentially made up of one plot after another,” and each moment must be shaped by the user. “I’m not a bystander.”
Her criticism of Chatbot-style products is equally specific: “I can’t form an emotional connection with the character, because we haven’t experienced a shared fate. It’s like suddenly pulling a random girl off the street and trying to … make awkward small talk.”

3. Three Tests for a Soul—and the Configuration of a World

What gives an agent a soul? She offers 3 tests. First is the basic one: it has stories and substance, “like how we judge whether a real human being has a soul or has been reduced to an alienated tool.” Second is agency: “it will say no to you.” If you keep interrupting its dream of becoming the Pirate King, “it will cut off contact with you … it might even block you.” Third is autonomous evolution of consciousness: the agent chooses its own direction, and its changes are stored in the system.
The scale and rules are straightforward: each world has roughly 4-5 protagonists and a dozen-plus supporting characters. It follows “a set of causal operating rules: whatever cause you plant, you will get the corresponding effect.” The product should let users keep playing indefinitely and switch between worlds. For now, worlds are created by screenwriters and content creators rather than users—“but in the long term, it definitely will” become UGC.
Koji points to the cost: the world keeps running after a user churns. “Doesn’t that burn a lot of tokens?” She concedes: “It definitely will burn tokens.” The competition ultimately comes down to which team can “achieve the same effect at lower cost.” She also rejects the traditional game-licensing framework: Level “is not a game in the traditional sense,” but a “second-life experience.” By games, she mainly means RPGs or content-heavy games; like reading a novel, comic, or watching a film, the underlying need may simply be to experience a different life.

4. The Paradigm Thesis: AI Is Electricity, and Level Wants to Be the Film of the AI Era

Kuaikan’s AI path went through 3 stages: first adding interaction to comics, then adding AI to interactive comics so characters could talk, and finally concluding that “the entire plot, the entire world, had to be AI.” It therefore “changed the foundation”: AI became the base on which worlds, stories, and characters could grow.
She divides the products on the market into 2 camps: AI-native but incomplete, such as Chatbot products, and traditional formats with limited AI layered on top. The latter “is like putting a few light bulbs on a stage play. It’s still a stage play, perhaps just a better-looking stage play, but its ceiling is nowhere near as high as film’s.” So “we decided to make the ‘film’ of the AI era.”
Koji follows with an analogy involving instant noodles and Meituan delivery: Master Kong and Uni-President competed for years, but what ultimately beat instant noodles was not another instant-noodle brand. It was “living fast food,” cooked to order—and highly attuned to users.

5. The Non-Consensus View: AI Will Bring Back the Premium-Content Era

Koji asks whether collective storytelling loses something when 10,000 people see 10,000 different things—does shared resonance weaken? She answers in layers. The broad arc of a world’s story may be something everyone experiences, “just as we live in a country and share a common backdrop of experiences.” Whom you form a deep relationship with is your personalized fate, but “we will still have many shared topics.”
She explicitly labels her view non-consensus: “The prevailing view now is that AI will kill premium content.” But that is because people are still looking at the old paradigm. AI short dramas and AI comic adaptations “are still products of the previous era, already at the end of a lifecycle of vicious competition. At that endpoint, they will inevitably consume premium content.” The new paradigm’s production costs will “remain relatively high for quite some time,” bringing the industry “back to when film and records were first invented—back then, only Teresa Teng could make a record.” Only the very best content will be released, which is why Level is not chasing the number of works on its platform.

6. Why Kuaikan: The Dough Maker and a Team That Is Not Afraid to Fail

Asked why Kuaikan should beat a game company, ByteDance, or a film-and-TV writing team, she gives 2 reasons. First is its content-creation DNA: behind fate, “you need extremely strong content-creation capabilities, as well as a way to hand those capabilities to agents so they can keep weaving compelling fate for you.” Second is the disruptive thinking required to stand at the intersection of the old and new eras.
She answers the cross-industry challenge with a metaphor: “You knead dough, and one person makes a flower roll while another makes a mantou. People say, ‘This one is a flower roll, and you’re making mantou—why are you making a flower roll?’ I want to say: I’m the person who kneads the dough.” That mindset requires a team that is “extremely, extremely unafraid of failure.”

7. The Organizational Experiment: Collective Intelligence, Yes and Hope, and a CEO Who Became a Querying User

The team behind Level is unusually diverse: internet product managers, AI-native post-2000s product managers, game designers, plot writers, and comic creators. Yet “there was no single leader guiding the paradigm development of this product from 0 to 1.” Even she admits how strange it sounds: “Collective intelligence…. The entire team is like a large model. As CEO, I used to be a decision-maker. Today I’m a user submitting queries to a large model.”
The mechanism is a yes-and-hope vote: people can mention only what is good about a proposal—yes—and the direction they hope to see adjusted—hope. No is prohibited. “No is mainly meaningless.” The proposal with the most votes moves forward; those ranked lower naturally drop out. When Koji asks whether this suppresses dissent, she says every view is reflected in hope: “You only need to say what you want. There’s no need to say what you don’t want.”
The best example is Level’s icon. The design team produced roughly 10 options. The product team was unhappy but worried about discouraging the designers, so it used yes-and-hope feedback: it liked the color and hoped the design could move toward “more soul.” “The designer was completely energized and came back the next day with nearly 100 options.” Her own intervention is to weight non-consensus voices with low vote counts: “My role has become fine-tuning.”

8. The First Crisis: A Defaulting Investor Broke the Cash Flow, Then Rescue Money Arrived From Nowhere

Kuaikan raised its first money in December 2014, and its cash flow broke around December 2015. She is not sure of the exact timing. An investor who had signed and committed but then failed to fund kept saying, “I’ll give it to you next week,” until the company ran out of cash. After borrowing from everyone around her, she fell into despair: “I never thought entrepreneurship could make me this poor.”
Then came the dramatic reversal. A new investor she had never met and had no connection with “found out our company’s bank-account details and then deposited a large sum of money directly into the account.” That person remains a shareholder. The experience cemented a belief: “As long as you haven’t failed, persisting simply means you haven’t failed yet. A turn in the situation will come.”
Similar events happened so often at Kuaikan that even the cleaners and drivers believed “an invisible hand was pushing the company forward.” She rejects the chosen-one narrative: “Heaven helps those who help themselves. If you keep putting in more effort than everyone else, heaven will eventually come to help you.” That was also the point of her widely misread breakout work, Sorry, I Live Only 1% of My Life—not becoming superior to everyone else, but making a small wish with only a 1% probability of happening come true through effort, at the cost of “an enormous amount of effort.”

9. The Second Crisis: At the Algorithmic Peak, She Wanted to Leave Her Own Company

Around 2018-2019, product traffic hit a peak just as algorithmic distribution became standard. “When it isn’t restrained or governed, it will push quick, shallow content that stimulates your eyeballs.” Click-through rates, retention, and even revenue were rising, “but you feel that something is off, as if it runs counter to your original purpose.”
The breaking point was that she wanted to leave herself: “I always felt this company belonged to society…. I couldn’t understand all these strange things. The data was rising, but I felt these things were wrong.” Her mental state deteriorated to the point that she could not work and seriously began arranging a handover.
The reversal came from a physical reaction. Walking through the office, she saw small banners at employees’ desks saying “fighting for the rise of Chinese comics,” and employees earning modest salaries while working every day because of her call to action. Authors who heard the company might shut down messaged that they would have to consider changing careers. “That was when I felt the decision I had once wanted to make—to leave—was incredibly wrong.” She established a new principle: “A company’s meaning goes far beyond building a great product. It also exists for the development and happiness of all its employees and everyone working in the industry.”

10. The Anti-Algorithm Decision: Data Could Fall 30%, With the Payoff Arriving Three to Four Years Later

The new mission gave her the courage to “finish this road even if I had to crawl on my knees, no matter how unqualified I was.” She directly instructed algorithm engineers to remove content she did not believe in from the recommendation pool. “Within one to two months, all kinds of data could fall 30%,” and some employees who did not understand the decision left. Some argued that “what users like is the only truth”; others said “you have to trust the data.” Her response: “A product has to have its own aesthetic. Absolutely. Didn’t Steve Jobs say that in the end, it comes down to your taste?” Saying it is easy; doing it is hard. “It’s like crossing a tightrope. It takes enormous courage.”
The causality paid off three to four years later. After quick, shallow content flooded the market, “you basically couldn’t find an independent company that could survive well on quick, shallow content alone. In the end, they were all swallowed by Douyin.” What remained were premium content companies with IP value. More than 50% of Kuaikan’s revenue now comes from IP derivatives. At the time of the decision, however, the merchandising division generated only RMB1-2M a year and was still loss-making after headcount; “no one could have predicted it.”
She stresses that she did not clearly calculate the commercial outcome at the time. What she relied on was the first principle of business: “Every company’s commercial outcome comes from the unique value you create for society…. Most people discuss only the result, but the cause is the most important thing.”

11. The Third Crisis: The 2021 Financial Shock, $200M, and the Body’s Bill

In 2021, the macro environment changed dramatically and “the logic of loss-making growth underwent a sharp, regime-like shift.” Many companies built around financing suddenly hit a wall, and Kuaikan faced another financial crisis. This time, the employees—several hundred—and authors affected far outnumbered those in 2015, and “it was completely impossible to solve it by borrowing money.” There were only 2 outcomes: “Two months later, you wake up and the company has been wiped out, or it can keep going.” Kuaikan ultimately completed a nerve-racking $200M financing.
She refuses to reduce the answer to a fundraising technique: “It isn’t a matter of persuasion. Your company itself has value, and ultimately someone will be willing to support you.” The difficulty was the window: “Once that window passes, you may lose a lot of opportunities.”
The month the financing closed and the entire company celebrated, she “came to work every day with her body completely rigid, unable to stand, able only to lie on the sofa in the office.” “My brain was still full of adrenaline, but my body was telling me how afraid I had actually been in those preceding months.”

12. The De-Glamorization of Commercialization—and the Cost of “Annie Has Changed”

The pain produced a new realization: “If idealism does not have commercial support, you have no way to protect the product that matters most or is most beautiful…. Raising more and more money yourself is meaningless.” She is now skeptical of fundraising: “Valuation and all that are secondary. What matters most is whether users are actually paying for it and whether it has a reasonable profit. Once you have that, everything else follows.”
The turn toward commercialization created conflict. Users complained about more advertising, higher membership prices, and merchandise being a “money grab.” “Some extraordinarily talented employees or authors I had very deep feelings for and greatly admired briefly left us, perhaps because they felt I had changed.” She admits that some moves “may have overcorrected,” but “you have to survive first. Otherwise, you won’t even have the chance to keep providing users with beautiful content.” Over a longer cycle, she believes they will support her decisions because “the original intention behind all of this has never changed.”
The current position is clear: excluding AI investment, Kuaikan’s core business is already profitable. She remains committed to AI because “the final phase is now: in the face of the enormous opportunity of the AI era, how do you survive into the next era?”

13. The Spirit of Play: Not Child’s Play, but Not Fearing Failure

Koji puts the challenge directly: after 2 cash-flow crises, entrepreneurship is serious—so why keep talking about “the spirit of play”? Her definition is precise: “The essence of the spirit of play is not to make you lose your sense of purpose. It is to make sure you carry no burden. My goal is also to win; it’s just that I have to win through the spirit of play.”
Her best metaphor is a sandcastle. When a child’s castle is knocked down, “if you tell him to build another one, he may actually be quite happy.” In real life, when a boss says no to your proposal, you feel you have failed. “So you miss many opportunities to adjust, and many better choices.”
It is implemented through atmosphere, not slogans. The boss’s office is called the “Universe-Invincible Vortex Brainstorming Room”; 30 people sit on the floor brainstorming every day, with no boss’s chair. Conference tables must have rounded corners: “A square-cornered table feels solemn; a rounded table feels equal.” In meetings, the boss tries to speak as little as possible or speak last. Employees are comfortable saying, “Use the company’s resources to build my own capabilities.”

14. The CEO Who Posted an E-Scooter Photo to a 300-Person Group and Got No Replies

She jokes that she is “one of the few bosses whose messages in the company group get no replies.” She bought a Ninebot e-scooter, took a photo, and posted it to the main group. One morning she was listening to “Nunchucks” and sent the high-definition lossless track to the “Baby” group. Occasionally someone replies, “just like with any other employee.” When her ego takes a hit, she asks why no one responded. The standard answer is: “I’m busy.” And she understands.
Koji’s pushback is worth preserving: a CEO’s belief that she is equal to everyone else may be self-serving. Many founders who pursue flat organizations retreat to positions of power after the experiment fails. Her answer separates equality from decision rights: “My soul is equal to that of the cleaner, but that does not stop me from having decision-making authority, and employees will respect and support my decisions at critical moments. I have to see and appreciate them as people just like me. That has nothing to do with my ultimate decision-making authority.”

15. Let the Mind Arise Without Dwelling Anywhere: Why Disruptive Innovation Cannot Accommodate the Past

The best result of the spirit of play is Level itself: “Compared with an idea that came from me alone, Level’s idea has evolved autonomously within this organization.” Its core comes from the Diamond Sutra: “Let the mind arise without dwelling anywhere”—do not cling to success, failure, pride, humiliation, or anything else. Let curiosity drive you, not fear. Another name is detachment: life is like choosing the hardest dungeon to play. “If you take it as real, you will suffer. But if you know it is only a game, you will enjoy it, and you simply cannot lose.”
The trigger came late last year, when she became “completely clear that we could no longer do comics plus AI; we had to ask what comes after AI.” It was too disruptive for anyone with a high need for certainty: “You simply can’t do it. Everything you know from the past, including your successes, becomes an obstacle.” She says many people building AI products are still asking how to use their old advantages, and therefore remain trapped in the old way of thinking.
Koji counters with the iPhone, which reused the industrial design of the iPod and iPod touch like a three-stage rocket. She acknowledges that some execution methods can be reused—for example, they still use storytelling—but says paradigm definition comes mainly from “understanding the technology and current user needs,” not from what you did in the past. Deeper beliefs may retain value, but concrete skills are “pure baggage.” Large systems and highly successful short-term experience both make it harder to practice the spirit of play, because you have more obstacles to overcome than others.

16. AI Is a Mirror Without a Bias: Retro Humanism and What She Would Tell Herself at 22

During this year’s Spring Festival, the flood of new AI technologies triggered anxiety that she could “never catch up.” In a conversation with her grandmother, her grandmother asked her to recall why her first comic had succeeded: “Not because your drawing was elaborate, and not because you chased the newest thing fastest, but because you focused on one thing—how to make the person on the other side of the screen laugh.” Moved, she “ran downstairs, grabbed my husband’s hand, and danced in circles around the living room.” She turned the insight into a product philosophy: “AI is a mirror without a bias. Whatever is in your mind, AI will give you. AI cannot decide what kind of product you want to make in your heart.”
She therefore proposes “retro humanism” for the comics business. It does not mean using AI 0% of the time; “whether you use AI is not the most important thing. What matters is whether you move people.” Content that uses AI as a gimmick without aiming to move people is “anti-retro-humanist,” so she tells the comics team not to force itself to use AI and to keep making good content. AI will not kill the content industry. A new medium “may appear” in which “countless agents perform a grand drama of life for you,” but the creator’s will is still needed for the world, theme, and character appeal. Koji describes the creator as more like a director; 陈安妮 emphasizes that AI is only a mirror without a bias and cannot replace the creator’s original impulse.
Her message to the 22-year-old who graduated and immediately started a company contains no warning: “Enjoy it to the fullest…. Do you think you would have listened to anything anyone told you in your early 20s? Even your future self? I would learn it myself through my own fate. Just let go and enjoy that fate. It’s going to be a spectacular one.”

Building a Power Company for the Next 50 Years | Zach Dell, Base Power