---
title: "Why the More Powerful AI Becomes, the More You Need to Study Literature, Understand Aesthetics, and Know How to Tell Stories"
author: "Unique Research"
sourcePublication: "Unique Research Substack"
originalPublishedAt: "2026-04-22T10:31:08+00:00"
canonical: "https://ffcap.cn/en/research/src-20260422-02html"
source: "https://uniqueresearch.substack.com/p/src-20260422-02html"
language: "en"
---

# Why the More Powerful AI Becomes, the More You Need to Study Literature, Understand Aesthetics, and Know How to Tell Stories

_Original · Unique Research · 2026-04-22_

_Editor's note: The first-person report and its judgments belong to the original Chinese author. This English rendition retains the opening essay, five narrative sections, and the full panel transcript. Product claims, technical descriptions, market figures, and cultural comparisons are source or speaker claims, not independently audited findings. Company, personal and product names are transliterated where official English forms remain unverified. The source is dated April 22, 2026._

Unique Research

After AI Flattens the Technical Barrier, What Does Competition Return To?

The answer is: people.

"

A Chinese student in Europe was troubled by stolen takeout food, so he stuck a Taoist talisman on the takeout box—from then on, no roommate dared to touch it.

This is a joke in China, but Ding Wei, content director of Xinshijue, told this story to illustrate a serious point: the mystery of Eastern culture naturally has charm in a Western context—it works without needing explanation.

Then he said: "With the help of AI, it's also possible for a white person to be using Taoist methods."

At a Hangzhou AI WEEK roundtable, four guests discussed a topic that on the surface was "How AI Makes Comic Short Dramas Born Global"—but by the end, I found they were actually discussing the same question:

After AI flattens the technical barrier, what does competition return to?

The answer is: people.

Specifically, it is your understanding of human nature, your understanding of users, and your understanding of the specific person in that market you want to move.

Technological Equalization Has Arrived, and Aesthetic Hegemony Has Returned

Wang Jian, CTO of Jingying Technology, is a typical technical background—he used to do NLP, then felt the video era was coming, so he switched to AI video, and now specializes in overseas AI short dramas.

He said: "Technical people have already stepped down from the center of the stage. Now standing in the middle of the stage should be the creative people."

He explained it this way: previously AI video was more like a technical carnival, unable to produce commercialized products, and "technical hegemony" dominated the industry direction. But now this turning point has passed—AI technology is stable enough to produce real short-drama products, and the gap is no longer who uses a better model, but whose story is more gripping and whose aesthetics are more accurate.

"The importance of technology has given way to content, and aesthetic hegemony has re-emerged."

Tong Chao, co-founder of MOOTION, gave a complete framework. He believes the emergence of this dividend period requires three conditions:

First is model capability—from 2023 to now, progress has been so fast you can't believe it yourself. "Looking at 2023 technology with today's eyes, you think it couldn't be worse, but at the time no one expected it would be in this state now."

Second is distribution channels—platforms like TikTok and YouTube have already built very robust content distribution infrastructure, and with content it can quickly spread to the people it should reach.

Third is user mindset—this point is easily overlooked. In 2023, when users saw AI content, that sense of strangeness and novelty determined how they reacted; now everyone no longer cares whether content was made by AI, they only care "how are the narrative and expression capabilities."

Only when all three conditions are in place does the dividend period truly begin.

The Logic of Globalization Has Reversed in the AI Era

"Previously when doing globalization, content had to be the greatest common divisor, then you'd add subtitles in different languages and send it out to harvest box office."

Wang Jian said that was an approach constrained by cost.

Now it's completely different. AI has removed dependence on the physical world—you don't need to actually find a Brazilian actor, you don't need to film on location in Brazil, you don't need to assemble a local team that understands Portuguese. After technology dismantles the constraints of the physical world, what was previously impossible becomes feasible:

Using the same core, do extreme localization for different audiences.

He broke it down: North America likes werewolf castles, China likes domineering CEOs, Southeast Asia likes wealthy-family feuds. The themes seem very different, but the core is almost the same—humiliation, revenge, reversal, comeuppance—stimulating the same human emotions. Previously constrained by cost, you could only make one film that ten million people all found "okay"; now with AI, you can make ten thousand people "really, really like it."

"Use China's content production capability to tell each place's own stories well, each beauty its own beauty, and incidentally we make the money."

Liu Meiling, business director of Haiyi, has a more direct understanding—she calls it "whichever piece of meat is fattier, we eat that piece first." Haiyi's method is data-driven: doing numerous small-scale paid-traffic tests in different markets, looking at click-through rates, payment rates, and conversion rates, and using the data to decide whether to bet on a theme.

"For example, Brazilians prefer passionate, uninhibited local stories accompanied by armed forces; Europeans inherently have a sense of nobility and prefer the castle, vampire, and lord set."

This is not a judgment made off the top of one's head—it's data run with real money.

Is "Respecting Local Culture" Something Said, or Something Done?

In Haiyi's global-expansion story, there is a section I find particularly interesting.

When they first went overseas in 2023, they made over a dozen language versions. At that time AI products were basically oriented toward English, and no one seriously did localization. Their Portuguese translation may not have been that precise, but Brazilian bloggers discovered it: a Chinese company had come with real Portuguese—even if imperfect, "they felt the respect and understanding of the Chinese company."

It was this thing that got Haiyi its first 1 million MAU in Brazil.

Later they entered Japan, doing many pixel-style activities that Japanese users like. The Japanese market is to this day their long-term #1 market in AI image and video generation products.

"Many overseas products don't do enough details, don't genuinely stand from the user's perspective," Liu Meiling said. "For example, many beginners don't understand prompts, don't understand depth of field, neon colors—Haiyi launched one-click same-style—even a beginner who understands nothing can generate a work with one click, and the platform retains the user."

Doing markets first of all requires being diligent, having a keen nose, and being willing to test.

Tong Chao of MOOTION raised a tool-making perspective, which I think is another layer of understanding: tools should remain Universal—you can't make Japanese the core orientation just because you're hitting the Japanese market, because then when you go to the US users will be confused. "Use a universal solution to teach users how to make localized content"—finding a consistent core logic to hold on to is the essence of product survival.

Incidentally, they also tried the Japanese market.

Found users were relatively conservative, couldn't break in for now, so they withdrew.

"Can't do five things at once. If it doesn't work for now, we'll wait and see."

This sentence is worth noting. What is the easiest mistake to make in global expansion? Chasing five markets simultaneously, each one superficially tried, none done deeply. Tong Chao's choice looks like "admitting defeat," but it's actually clear-headedness.

What Exactly Is Technology Doing?

I want to pause here for a moment, because there was a conversation in this roundtable that I think laid bare the underlying logic.

Abner asked Wang Jian: what unique advantage does Jingying Technology's "narrative video generation technology" have compared to pure visual models like Sora, Kling, and Jimeng?

Wang Jian's answer was very calm:

"We and model vendors solve problems at different levels. Model vendors make shots realistic, emotionally powerful, close to prompts, approaching physical simulation—their generation unit is a 5 to 10 second single shot. What we solve is how to generate an entire drama well."

He gave an example: the first paragraph of the script says "I took out my phone to make a call," later it says "hearing shocking news, the phone fell to the ground"—the system has to recognize that both paragraphs are talking about the same phone. In the script, at the company everyone calls "Mr. Wang," at home the son calls "Dad," the wife calls "Husband"—the system has to know this is the same person.

This is cross-entire-drama dimensional context understanding, an engineering problem at the narrative logic level, not a single-shot visual quality problem.

The two things solve completely different dimensional difficulties. Model companies and content tool companies are not in a competitive relationship—more like two floors of a building.

Once this distinction is clear, many people's questions about AI content production are resolved—why do you need Jingying when you have Kling? Because you want to make a drama, not generate a single shot.

The Most Scarce Capability—They Started Arguing

For the final question, Abner threw out: if the era of AI global blockbuster films has arrived, what is the most scarce capability in this industry?

Wang Jian directly rejected the premise. He said, an idea contributed by one person can probably only strike a small group of people's hearts—it's not that easy to move everyone globally. If there really were such a thing, "it would more likely require luck, or happening to ride a big hot trend—it's not easily designed."

I think this is the most sincere sentence of the entire session, because it was said by someone who has made blockbusters.

Ding Wei didn't refute it, but gave a different angle:

"AI has given new opportunities. In the AI era, who is the director? A director who understands AI is more likely to become the dominant one."

He listed three core capabilities:

First, the capability to study human nature. Go take some supplementary literature classes—literature is the best way to study people.

Second, the capability to persist in aesthetics. Find confidence from Eastern aesthetics, and use AI to make romantic, elegant global content.

Third, the capability to tell stories. "In the future, a child only needs to give AI an instruction, tell a story he believes in, and the film may come out."

These three capabilities have one thing in common: none of them has anything to do with technology.

This was said by a content director—director, creativity, storytelling. And Wang Jian, who comes from a technical background, said "technical people have stepped down from the center of the stage." Two people from different angles said almost the same thing.

AI tools are getting more powerful, easier to use, and cheaper, but there is one thing AI cannot replace: your understanding of people.

You know what that Brazilian user wants to see in the late night while scrolling his phone.

You know the European user's inherent special feeling for castles and nobility.

You know why Eastern mysticism made that foreign roommate afraid to touch the takeout box.

Technology flattens barriers, but understanding people has never had a shortcut.

More Conversation Details

Unique Awards · Hangzhou AI WEEK Trends Roundtable Panel

"New Dividend Period for Global Content: How AI Makes Comic Short Dramas 'Born Global'"

Guests:

Haiyi AI — Business Director — Liu Meiling

Xinshijue AIGC — Content Director — Ding Wei

Jingying Technology — CTO — Wang Jian

MOOTION — Co-founder — Tong Chao

Moderator: Unique Capital — Partner — Abner

Abner: Hello everyone, I'm Abner (Zhao Liang), from Unique Capital. The guests on this Panel have very diverse backgrounds: Meiling comes from a business background, with over a decade of experience in the gaming industry; Director Ding has always been a professional director, from film, short dramas, comic dramas, to now doing AI comic short dramas and AI live-action dramas; Mr. Wang of Jingying Technology is purely from a technical background; Mr. Tong of MOOTION is from a product background. Next, let's discuss together what characteristics "born global" content has, especially in the current AI era.

First question, please give a brief self-introduction, about one to two minutes each, and at the same time use one keyword to summarize what you think "born global AI content" is. Starting with Meiling.

Liu Meiling: Thank you for giving us the opportunity to stand on such a stage. I'm Liu Meiling from Haiyi. Haiyi ranked #1 on the a16z list in December 2023, and was #1 on the list in both 2024 and 2025. Currently we have 30 million users overseas.

Ding Wei: Hello everyone, I'm Ding Wei from Xinshijue. We are a content company founded in 2011. From the founder to myself, we are both from director backgrounds, so we deeply understand that in content creation, good things are especially precious and valuable. We have always been doing content, and recently we are focused on AI short dramas and films.

Wang Jian: Very glad to attend this event. I'm Wang Jian from Jingying Technology. The host just introduced that I'm from a technical background. Before 2020 I mainly did natural language processing (NLP). At that time I felt the commercialization prospects for NLP were not particularly clear, and felt the future would be a video era, so I switched to AI video, and naturally started doing AI short dramas. We are now mainly targeting the overseas market with high-quality AI short drama business, and also produced several blockbusters last year.

Tong Chao: Hello everyone, I'm Tong Chao, co-founder of MOOTION. In 2023 we made an AI video creation tool targeting overseas, which was relatively early. To date we have accumulated a base of overseas users, and also seen some interesting market user changes, including interesting phenomena in content itself as technology changes—we can share with you later.

The keyword I want to give is "each beauty its own beauty." Regarding globalization, I think there is another keyword in video content: "semantic irrelevance." Video is actually a very universal visual expression, but often in different markets and among different users the biggest obstacle is language. If semantic or language irrelevance can be achieved, globalization becomes very interesting in the video content form.

Abner: Following up on Mr. Tong's sharing and topic, the differences between China and global markets are very large—not only language, but also cultural issues. But now everyone generally believes we are in a global dividend period for AI visual content. Please share your views: how long will this dividend period roughly last? For Chinese AI visual content or tool startups, what are the three most critical strategic measures to seize this wave of dividends? What things must be done now, that you'll regret if you don't?

Tong Chao: Regarding dividends, we ourselves think there are three necessary conditions:

The first is model capability. From 2023 to now, technological development has been compressed into a very fast rhythm. Looking at 2023 technology with today's eyes you think it couldn't be worse, but at the time no one expected it would be in this state now. So the evolution of model capability is a very important condition.

The second is distribution channels. We are at a very advantageous point in time—whether domestic Douyin, Xiaohongshu or overseas TikTok, YouTube, video content distribution channels are already very robust, and after having content it can be distributed and deliver value very quickly.

The third is user mindset, which is overlooked by many product people. In 2023, when users saw content with AI, their mindset was different from now; now everyone no longer cares whether this thing is AI, they only care about how the narrative and expression capabilities are.

These three conditions have brought about the beginning stage of the current dividend period. For all entrepreneurs, when AI flattens technical expression to a language-agnostic dimension that only cares about expression, you inevitably have to do something global. You can no longer stay in one place, otherwise entrepreneurship may be meaningless. As for what must be done now, it depends on the stage of different products. MOOTION has been operating for two years with a relatively solid user base, and at this node it is indeed time to do large-scale commercialization.

Wang Jian: I think there are several significant dividends in the current AI era:

First, now is a turning point. Previously AI video from a technical perspective was more like a toy or technical carnival, unable to make good commercial products. But now the turning point has passed—current AI technology can make good commercial short dramas. This marks a shift from technical hegemony to technological equalization, where the importance of technology gives way to content, and aesthetic hegemony re-emerges. We technical people have already stepped down from the center of the stage; now standing in the middle of the stage should be the creative people.

Second is cross-cultural global expansion. Previously, when targeting users in South America, you had to find actors who understood Brazilian Portuguese, which was hard to find; now AI has dismantled the difficulty, letting everyone together do things that couldn't be done before.

In terms of levers, I think a huge adjustment in organizational structure is an important lever. Traditional film and television faces numerous organizational and physical-law-confronting problems: to shoot things that don't objectively exist, the production manager has to manage hundreds or thousands of extras, arrange shooting schedules, etc. Now none of this exists—the entire AIGC organizational form, shooting process, and skill requirements are different, and transformation in this regard is a very important lever.

Ding Wei: First of all, the opportunity is certainly very large. My keyword is "breaking boundaries"—because where there are boundaries, there are actually opportunities. We used to have very clear content divisions: film, web drama, short drama, AI, and the market was also divided domestic and overseas. When all boundaries are broken, the opportunity is very large, especially now that we have better tools.

Now we can strengthen from three aspects:

First, grasp content. Grasp content with cultural heritage, especially Eastern cultural heritage and aesthetics.

Second, open up the broader vision. From the beginning of creation, find something with universally shared emotions and language, and lay out global content.

Third is process construction. We who do traditional film and television once felt aversion and anxiety about Hollywood industrial processes, but facts prove industrial processes are very important. Now AI workflows are in the process of being formed and tempered, and a good process will definitely give us great help.

As for what must be done now, as a content team, we think it is grabbing people, grabbing IP, and doing original creation.

Liu Meiling: I may be different from the three of you—I want to talk from the market perspective. Haiyi's market is very strong; we have always done paid traffic. When doing Xinghe, we had a 1000-person company in Chengdu doing market, R&D, and customer service, with the S-side in self-operated status; Haiyi now also has a team of nearly 500 people, with 30 million users overseas. Plainly speaking, our roots in the market are very deep.

AI equalization is certainly good—it solves many technical pain points. From a market perspective, we focus more on the European and American markets, currently with daily revenue of roughly $40,000. We extremely respect content localization and marketization—for example, our most famous blockbuster overseas drama has themes of werewolves, vampires, castles, maids, etc., all elements that European and American users like. Haiyi's characteristic is deep marketization, "give whatever food to whoever eats it."

Abner: Following up on Meiling's topic, Haiyi just launched its own short-drama platform, and in two to three months it has monthly revenue of $40,000 to $50,000, with werewolf and vampire themes being relatively popular. How do you all view content consumption differences in different overseas regions (such as the US, Middle East, South America)? When doing content global expansion, do you need to more respect local culture to cater to them, or can you do cultural export of China's relatively strong fantasy and other content with strong Chinese memories? Meiling can talk first.

Liu Meiling: I'll talk from the market perspective. Haiyi's approach and strategy are completely market-first, "the duck in the spring river knows the water is warm first." When doing Europe, the Middle East, the US, or Brazil, our strategies are completely different.

We will do numerous small cases with localized characteristics for paid-traffic testing, looking at clicks, payment rates, and conversions to decide whether to do this theme—I think this is also sharing practical insights with everyone. For example, Brazilians prefer passionate, uninhibited local stories accompanied by armed forces; Europeans inherently have a sense of nobility and prefer the castle, vampire, and lord set. We will also have personnel collect local preferences, then make corresponding customized dramas. The approach is "whichever piece of meat is fattier, we eat that piece first"—occupy the market first.

Abner: Meiling just talked about the demand differences between US and Brazilian users. MOOTION's users are in North America and the Middle East (such as Oman), with very large cultural differences. How does MOOTION view this issue?

Tong Chao: MOOTION's position is as a tool, relatively neutral, and doesn't carry too many attributes in content. But when operating the market, you must pay attention to what content users make with the tool. I only learned after going to the Middle East market that subtle differences in how Gulf countries wear headscarves, men's and women's clothing, etc., can cause users in different countries to have enormous differences or even rejection when accepting. You must do the local market well and pay attention to user habits.

As for the tool itself, one point I want to raise is: you must define a very Universal solution. You can't turn the product into that way just because you're going to a certain market. Use a universal solution to tell users how to make localized content. You can't set Portuguese as the product's core functional orientation just because you're going to Brazil—that way when you go to the US users will complain. The product must have a very consistent experience and core logic to hold on to.

Abner: I want to follow up with a question. Haiyi has performed very well in the Japanese market, long-term #1 in image and video generation. If MOOTION were to focus on the Japanese market, what strategy do you think would shake Haiyi's position?

Tong Chao: Then I'll first try to poach Meiling. Every single large market is a target everyone hopes to reach. We also tried going to Japan, but found that Japanese users overall are relatively conservative, so my choice was to withdraw first. Can't do five things at once. If it doesn't work for now, we'll wait and see.

Abner: Next, please answer, Mr. Wang. Jingying previously did very well with overseas short-drama platforms, and now also does high-quality overseas AI short dramas. How do you view content global expansion strategy?

Wang Jian: I think in the AI era, globalization is completely different from before, even reversed. Previously when doing globalization, constrained by cost, if you wanted to globalize a film, the content had to be the greatest common divisor that everyone could understand, then you'd add subtitles in different languages and send it out to harvest box office. But in the AI era there's completely no need to do this.

From the underlying logic, human nature is all universal, but the packaging settings are different. North America likes werewolf castles, China likes domineering CEOs, Southeast Asia likes wealthy-family feuds. They seem very different, but the core is similar—all are humiliation, revenge, reversal, comeuppance, etc., stimulating the same human feelings. With AI technology no longer constrained by the physical world, you can use the same core to do extreme localization for different local audiences. Previously it was one film that made 10 million people like it; now it's very likely making one content that makes 10,000 people "really, really like it." To summarize, it's using China's content production capability to tell each place's own stories well, each beauty its own beauty, and incidentally we make the money.

Abner: Mr. Wang is indeed from a technical background. I want to follow up with a technical question: what unique advantage does Jingying Technology's "narrative video generation technology" have compared to pure visual models like Sora, Kling, and Jimeng?

Wang Jian: I think we and model vendors solve problems at different levels. Whether Sora, Kling, or Jimeng, the model vendors' generation unit is a 5 to 10 second single shot. Their pursuit is to make shots realistic, emotionally powerful, close to prompts, approaching physical simulation.

Our tool doesn't solve this; we solve "how to generate an entire drama well." We mainly target directors and screenwriters, so the system needs longer-dimensional context understanding. For example, the first paragraph of the prompt says "I took out a phone to make a call," and after a while the next paragraph says "hearing shocking news, the phone fell to the ground"—the system has to recognize that both paragraphs are talking about the same phone. Or in the script, at the company everyone calls "Mr. Wang," at home the son calls "Dad," the wife calls "Husband"—the system has to recognize that this is the same person. This is the core problem we solve at a higher dimension.

Abner: Next, please share, Director Ding of Xinshijue. Xinshijue actually also has plans to go overseas, and has already signed some projects with TikTok. What kind of content are you preparing to go overseas with—mainly Chinese genetic culture or considering local culture? Including your previously domestically popular fantasy themes, how are you considering global expansion?

Ding Wei: The small boat going overseas has already reached shore. In our recent werewolf, vampire, and Western "dragon-king dramas," besides meeting customization requirements, we are also proposing original ideas. The core is finding commonality—providing emotional value. Whether Chinese or foreign, love-hate, growth, counterattack, rebirth, all have good empathy. For example, when doing Viking settings for the West, we will give a slightly different Viking according to our own aesthetic understanding. Persist in our own emotional understanding and aesthetics, while through AI helping us make a visual outer garment that overseas audiences can accept.

Abner: For example, themes like Taoism and cultivation—will foreigners be interested?

Ding Wei: I don't know if everyone has seen a news story— a European student found his takeout was often being taken, so he drew a Taoist talisman and stuck it on it, and as a result classmates didn't dare to touch it. Eastern mystical culture has unique charm in a Western context. The Taoist going overseas doesn't necessarily have to wear a Taoist robe, but the cultural core will definitely be there. With the help of AI, let overseas users understand this core and accept this image, so it's also possible for a white person to be using Taoist methods.

Abner: Next, I want to ask Meiling a question. Haiyi has been on the a16z list four consecutive times, maintaining a market-leading position. What exactly did Haiyi do right to enable it to consistently achieve relatively good results in the global expansion track?

Liu Meiling: Haiyi still has a lot to continue working on. We have maintained a heart of reverence for the product. When first going overseas in 2023, we made over a dozen language versions. At that time AI products were relatively English-oriented, and didn't do localization suitable for various regions.

The first market we exploded in was Brazil, "the countryside surrounding the city." At that time Brazilian bloggers discovered that Haiyi had come with Spanish and Portuguese—even if the translation wasn't that precise, but adapted to the local language environment, they felt the respect and understanding of the Chinese company. We reached our first 1 million MAU in Brazil. The second place to explode was Japan, where we did many pixel-style activities they like. Respect each region's cultural traditions, and genuinely do customized services with them.

Another point: many overseas products don't do enough details, don't genuinely stand from the user's perspective. For example, many beginners don't understand prompts, don't understand depth of field, neon colors or glossiness—Haiyi launched a blockbuster application: "one-click same-style." Even a beginner who understands nothing can generate a work immediately after one-click same-style, and the platform retains the user. Doing markets first of all requires being diligent, having a keen nose, and being willing to test.

Abner: Next, I'd like to ask Mr. Tong of MOOTION. MOOTION just launched and quickly surpassed 5 million registered users, and from day one rejected the free model. What successful experience and thoughts on future growth can you share with us?

Tong Chao: Pricing is a very fun topic. We tried a Freemium model for a while, but the problem was how to cover the inference costs of free users? The marginal cost of inference cannot be ignored, and if it can't be covered, you can't build a sustainable business model. In this situation, what suits us better is small-step experimentation, setting a relatively high threshold, and using pricing to purify the user group that truly has actual usage scenarios.

Of course there is an opportunity cost here: no one knows whether AI inference costs will drop to 1% of current levels within the next 18 months, and perhaps the current pricing choice misses the opportunity for large-scale user acquisition. Facing continuous technological iteration, MOOTION's choice is to have enough flexibility and space—perhaps every three months we have to make adjustments and experiments based on technological iteration.

Abner: Due to time constraints, our last question is for Mr. Wang and Director Ding. If the era of AI global blockbuster films has arrived, what will be the most scarce capability in this industry?

Wang Jian: I don't quite accept this premise. An idea contributed by one person can probably only strike a small group of people's hearts—it's not that easy to move everyone globally. If you really want to hit a large group of people, it would more likely require luck, or happening to ride a big hot trend—it's not easily designed.

Abner: Luck is indeed also very important. Director Ding, what do you think about this question?

Ding Wei: I also particularly agree that it may not be one person's thing, but AI has given new opportunities. In the AI era, who is the director? A director who understands AI is more likely to become the dominant one. I think the core capabilities are several points:

First, the capability to study human nature. Go take some supplementary literature classes—finding human nature from literature is very helpful.

Second, the capability to persist in aesthetics. Find confidence from Eastern aesthetics, and with the help of AI make very romantic, elegant global content.

Third, the capability to tell stories. In the future, a child only needs to give AI an instruction, tell a story he believes in, and the film may come out, so the capability to tell stories is most critical. With these points plus embracing AI tools, it is possible to complete good works.

---

Original publication: https://uniqueresearch.substack.com/p/src-20260422-02html
On-site reading page: https://ffcap.cn/en/research/src-20260422-02html