Roundtable: Speed and Human Resonance: AI Enables Infinite Content Generation, Who Takes Charge of the "Heart-touching" Experience | 36Kr 2026 East Forward Globalization Summit

In 2026, Chinese enterprises' overseas expansion is shifting from "pursuing incremental growth" to long-term global operation. On September 17, the 2026 East Forward 2nd 36Kr Globalization Summit with the theme "Rising Tide, Shoreward Reunion" was held in Shanghai. Overseas-oriented enterprises, investment institutions from the consumer, technology, manufacturing and other sectors, as well as globalization service providers gathered at the scene to discuss practical issues including localized operation, user insight, supply chain, compliance, channels and global organization. When "going global" is no longer a niche choice, enterprises have to face the real challenge: how to survive and take root in overseas markets, turn a one-off market breakthrough into continuous operation, and gradually build their own global competitiveness.

The following conversation is edited and organized by 36Kr:

Host: Zhou Xuechun, Head of Global Business, 41 Finance

Guests:

Guo Rui, General Manager of Commercialization of Tian Gong AI Large Model, Kunlun Wanwei

Lu Yu, General Manager of Linmon International

Zhou Xuechun (Host): Good afternoon to all friends present at the scene and partners watching the live broadcast online!

The theme of our sharing today is "Speed and Human Heart", but we hope this roundtable discussion can slow down the overall pace, share more insights with the audience on site, and discuss how AI changes content production, as well as who should take charge of the part that truly "touches people's hearts".

Let me introduce myself briefly first. I am Zhou Xuechun from 41 Finance, and I am mainly engaged in media communication and overseas marketing. Based on this opportunity and the overall industry changes in recent days, I can clearly feel that AI has made overall content production much easier. In the past, when we created a set of brand content, including overall creativity, scheme output, graphic design and video production, the video production process used to involve on-site shooting, lighting, costume and props arrangement, actor communication, post-production and editing special effects, which was a complete large-scale project. But now AI has greatly compressed the entire production chain, making content produced faster, in larger quantity and at lower cost.

However, from the perspective of marketing, I now also have a concern: now every brand and every company can generate dozens or hundreds of pieces of content a day, are users really willing to spend more time watching them? From my own experience, I think AI only improves production efficiency. Users' attention and fragmented time have not increased as a result. What has become more scarce instead is whether users are willing to stop and pay continuous attention to your content after seeing it, whether they can trust your content more, and whether they are willing to remember you after finishing watching.

Today we specially invited two guests in different positions of the industrial chain. One is Mr. Guo Rui, General Manager of Commercialization of Tian Gong AI Large Model of Kunlun Wanwei, who stands at the front line of AI and product commercialization. The other is Ms. Lu Yu, General Manager of Linmon International, who is deeply engaged in the front line of the content industry. Please give a brief self-introduction respectively.

Guo Rui: Hello everyone! I'm Guo Rui. Thank you very much for 36Kr's invitation to this overseas expansion summit today, and I'm very glad to see you all again. The Tian Gong AI business of Kunlun Wanwei covers large models, tools and short drama distribution platforms, and we are exploring how to expand overseas under the AIGC wave.

Lu Yu: Thank you all for your time this afternoon. I'm Lu Yu, the person in charge of Linmon International. Linmon International is mainly engaged in the overseas distribution of Chinese film and television content. In the past, we mainly focused on the overseas distribution of long dramas, and in recent years we have also expanded more content forms, including the overseas distribution of micro short dramas. As AI has been directly applied to our work, especially this year we have begun to make extensive use of AI in content creation, marketing and all other work links. I am very glad to have more discussions and sharing with you on this occasion today.

Zhou Xuechun (Host): It can be heard that Mr. Guo, Ms. Lu and I, the three of us, have all deeply rooted our businesses in overseas markets. Now AI can realize infinite content generation, but who on earth should be responsible for the part of touching people's hearts? Today we will hold a roundtable discussion around this theme.

Let me start with a simple question. The overall development of AI is very fast. I may hold a certain mindset today, but in two or three days, there will be great changes in model iteration and AI applications. Based on the current form of AI, what do you think is the most important? Is it "product", "content" or "marketing"? Which one has the biggest change? If you have to choose one from the three dimensions, which one will you pick and why?

Guo Rui: The question is which one has the biggest change after the arrival of the AI era. Instead, I would like to first think about what has not changed. Human time is limited, and the total amount of content that people can consume in a day or even a lifetime has not changed. Now it is obvious that the production supply has changed greatly due to technological innovation, which is visible to all, and we don't need to elaborate on that.

The more important thing has become how to realize the matching between limited demand and nearly infinite supply. Good content and the most matching content are still the most important. From the perspective of unchanged factors to see what has changed, the change of marketing is very huge, because the situation of both demand and supply has undergone absolute changes, so marketing is bound to change. The change of marketing lies in how to make good content find the right people who truly need it.

Because of the changes in magnitude and technology level, it inevitably requires the change of products. The change of products not only includes how the production end of the workflow works, but also includes how the marketing end distributes and releases the content. The discussion of content distribution now is completely different from that in the past, and all these need to be changed.

The question just now is also very interesting, asking me to choose one from the three, which is very difficult. Because we ultimately have to return to business. If the essence of business has not changed, the essence of business is still that users need good content, and enterprises need to deliver good content to them. None of these three links can be skipped. Skipping any link from content to marketing to product will reduce the possibility of the final success of the whole thing.

From our point of view, we still need to see what the overall situation is like, and what corresponding changes each link should make.

Zhou Xuechun (Host): It can be heard that what Mr. Guo wants to express is to promote marketing through content, and then drive product iteration in reverse. What do you think, Ms. Lu?

Lu Yu: We have always been content providers, and our business perspective is different from that of Kunlun Wanwei. Even our content products are also supplied to Kunlun Wanwei. What Mr. Guo said just now is very good. Let's first see what has not changed. From my point of view, I'd like to talk about what has changed the most.

My feeling is that the biggest change is on the product side, mainly in product type, category, scale and output. At the current stage of AI, AI is not human yet, and it has not reached the state of being completely like human. From our daily front-line work experience, AI provides us with very easy-to-use tools. Whether in the dubbing and translation of daily long dramas/short dramas, localization, or even some content creation, the original production capacity was limited. After AI provides us with the basis of cost reduction and efficiency improvement, we can achieve higher production capacity.

Markets that were previously unreachable can now be accessed more easily and effectively with the help of AI tools in dubbing, translation, localization and cultural integration. These aspects are much more convenient, so the production capacity and efficiency have been greatly improved. The biggest feeling is the product itself.

There is actually little change in content. In fact, we rarely use AI to write scripts now, because our annual output is very small, and we are still focusing on high-quality dramas. Even for vertical-screen micro short dramas, we still adhere to the standard of high-quality dramas. The so-called high-quality dramas mean that we still hope to maintain a higher hit rate of popular works.

For this reason, the polishing of scripts and the requirements for localization are very high. At this stage, AI is still difficult to replace human beings, which is also difficult to be replaced. At least for now, AI is not that human-like.

The creation and guidance of content still rely heavily on human beings. AI has helped us a lot in marketing. Now many overseas marketing clips are completed or tested with AI, which has really helped us a lot. Overall, the most effective and biggest change is still on the product side itself.

Zhou Xuechun (Host): Ms. Lu, from what you just said, AI can promote the increase of script output. According to probability calculation, if the number of scripts reaches a certain large scale, can AI replace manual work, and will the probability of attracting users be higher?

Lu Yu: The biggest help AI brings to us at the script level is cognitive help. Because the markets we face daily actually cover more than 190 countries and regions, and each country and region has its own unique market. Our original understanding of these markets is very shallow. AI can help us quickly understand these markets and provide us with basic data and information to support our basic judgment. For example, we may not know very well what types of dramas and categories can be developed in the Indonesian market, and what the taboos are. In fact, there are many taboos we are not aware of, and some content cannot be touched from the very beginning.

To give a simple example, AI can provide us with a lot of cognition in category selection, scripts, stories and narrative methods. But in practical operation, human beings still need to lead the work. Our practice is very simple. After the story outline is established, we will invite local screenwriters to participate directly. There are many local so-called "junior editors", who are experienced practitioners that can cooperate in the creation of micro short dramas and scripts to help us with the work.

This is the level where we use AI now.

Zhou Xuechun (Host): AI promotes the generation of the whole content, but when the content is implemented in specific projects, it still needs manual work, including optimization and adjustment according to the actual situation. At this stage, content is no longer scarce. From the product perspective, Ms. Guo, when content is completely non-scarce, what kind of dimension do you think will be scarce in the next stage?

Guo Rui: Good content is still scarce. The reason is that human time is limited. The more limited time is, the richer the supply is, and good content is still scarce. From the product perspective, for example, our Sky Production (Tian Gong) is a short drama workflow/workbench, and we need to consider what is scarce. Behind the scarcity of good content is the scarcity of judgment, the ability to judge what is good and what is bad. Just like when doing localization, we need local screenwriters to participate, which is essentially the demand for cultural understanding and judgment.

So when we develop products, we will separate them appropriately. There is a concept called "director sovereignty". How to define the product boundary? AI can do a lot of auxiliary work, but the work that relies on judgment will be retained for human beings, so the product boundary will be planned in this way. Let people with judgment amplify their correct judgment with higher efficiency. This is our consideration on products.

Zhou Xuechun (Host): It can be heard that the two of you have reached a certain consensus. Good content is still scarce, but now content is no longer particularly scarce with the help of AI. High-quality content is still in shortage, and we need to iterate the output of good content through products, content and marketing to serve the audience.

Mr. Guo, you are in charge of the commercialization of Tian Gong AI large model. I particularly want to ask a question from the product side. Now the large models of all parties are getting better at writing, painting, and generating various types of videos. Where do you think the differentiation of AI lies? In the past, the competition was about model capability, generation speed and output quantity, but now the gap in basic capability is getting smaller and smaller. Why will users choose your product instead of others'? This question may be a bit sharp.

Guo Rui: It's normal. The large model determines what can be done and what cannot be done, while the product determines whether the thing can be done successfully. If a product is just a "display cabinet" of model capabilities, it does not mean that you have truly mastered these intelligence. Just like walking into a library, all the books are placed there, which does not mean you have mastered the knowledge in them. What a good product really needs to solve is: in order to complete one thing, what kind of intelligent capabilities should be called at what time. The so-called model capability is intelligence and cognitive capability, so cross-disciplinary talents are very scarce now, which need to meet the following three points:

First, have a very good understanding of the short drama industry, understand what "director sovereignty" means, or what the parts determined by human beings are. In each workflow, what creators want is more control, what they want is scale, and what they want is testing. So we need to have this understanding when developing products.

Second, have the understanding of technology, including the boundary capability of the model, what parts are handled by the model, and what parts are handled by engineering.

Third, this point may be rarely heard in other tools, but we attach great importance to it, which is the understanding of business or the understanding of the final closed loop. Because we also have our own distribution platform, we need to understand that the final goal is not to finish producing the video. The final closed loop is to achieve success and get feedback in the target country. The combination of these three understandings can create a handy product and a handy tool for the new era of production methods.

Zhou Xuechun (Host): I understand that the product is the superficial presentation, but the underlying model is the moat, which is deeply rooted in the market, user experience and usage scenarios.

Let's focus on it. Kunlun Wanwei itself develops AI products for overseas markets. When AI products face different countries and different users, besides outputting in the local language (localized language), what else do we need to understand to make local users willing to stay and use our products and pay attention to us for a long time?

Guo Rui: After years of overseas expansion, the core is ultimately cultural understanding. In the past, it was difficult for us to directly reach a large number of users. In the past, when we expanded overseas, we had many partners. When partners helped us with distribution and script creation, we actually obtained local insights and cultural understanding. Now overseas expansion is also a matter of technical capability. We can directly reach consumers and users, so cultural understanding must be internalized by enterprises: either accumulated through the team, or accumulated with the help of AI capabilities. Because content needs to deal with emotions, the understanding of human inner needs and emotions is actually a kind of competitiveness. This is also the part that we just mentioned that "director sovereignty" should not be separated from, and it is also a very important part.

Zhou Xuechun (Host): What Mr. Guo said is very good. The insight into local users, and how to create local content and deliver it properly. The same problem also exists on the content side. For example, the landing of Mr. Guo's product is to make users willing to stay and use it. But as for content, especially the overseas distribution of short dramas that you are engaged in, the goal is to make users keep paying to watch the following episodes continuously.

The content industry is one of the industries that have been most adapted to generative AI, so you must have deeper feelings. Now AI can write scripts, create various types of materials, generate videos, and even complete the whole production process of short dramas directly through Agent or several AI combinations. I particularly want to ask: when AI can generate stories infinitely, will there be more good stories as a result?

Lu Yu: The question you just asked is whether AI can generate good stories.

Zhou Xuechun (Host): For the overseas distribution of short dramas, good stories are of course very important. The characters, emotions, rhythm and resonance in short dramas and TV dramas that drive users to keep watching, how to touch the inner heart of users, and trigger the sensitivity and inner resonance of the audience. In the past, all these were realized by writing scripts manually. Now with the mode of AI or AI+, the number of script content has increased, and we need to filter out more good stories. With the continuous iteration of AI and the deepening of user understanding, will there be more good stories?

Lu Yu: It is a general trend that AI is getting closer and closer to human capabilities. Now AI is not human yet, and there is still a big gap, but it