AI long-form drama airs on satellite TV for the first time, share price sees consecutive daily limit gains, Mango TV's AIGC reverse experiment of "building AI with content" | Exclusive Interview

By Li Xiaoxia, Lan Jie

Edited by Qiao Qian

On August 31, the mythological seasonal drama *Post Journey to the West* Season 1 "Huaguo Mountain Chapter" was officially launched on both Hunan TV and Mango TV.

Produced by Mango TV, created by the AIGC Innovative Content Center, and co-produced by Bojing Culture, this drama is the first series that adopts the "simultaneous review and broadcast" mode after the release of the "21 Broadcasting and Television Policies", and it is also the first AI long drama to be broadcast on satellite TV channels.

Driven by this positive news, the share price of Mango Super Media hit the 10% daily limit for two consecutive trading days.

To further sustain this momentum, at the Mango Co-creation Ecological Conference held on September 17, Ren Xu, General Manager of Mango TV AIGC Innovative Content Center, announced that the "Heavenly Palace Battle Chapter" (Episodes 6 to 10) of *Post Journey to the West* will be launched during this year's National Day holiday.

Around the *Post Journey to the West* project, Mango's AIGC layout has taken shape, which can be roughly summarized into three parts: production, content label, and creation platform, covering the entire AIGC production process.

At the aforementioned conference, Mango Super Media launched Lingchuang Theater, positioned as Mango's AIGC high-quality content label, with three key planned tracks: high-quality mid-to-long dramas for female audiences, classic Chinese-style IPs, and children's animations.

The company announced that it will co-create high-quality content with outstanding creators and leading film and television companies.

Meanwhile, Mango Lingchuang, the AIGC platform built on Mango's large model, was once again unveiled. According to Lu Haibo, Vice President and Chief Technology Officer of Mango TV, Mango Lingchuang has reached the international leading level in the professional film and television creation field, and the presentation of *Post Journey to the West* is a verification of this capability. The company will open up Mango's technical product ecosystem later to allow more participants to join in.

In terms of theme selection, Ren Xu explained the reason for choosing *Post Journey to the West*: "Most current AIGC content is male-oriented, but our platform has accumulated a large number of female users, so we will develop female-oriented content." For example, the platform will create *The Return of Female Generals*, bringing together characters including Hua Mulan, Mu Guiying and Fan Lihua to build a female hero universe IP, and form differentiated market competitiveness through serialized development modes.

In Ren Xu's view, the starting logic of Mango's AIGC business is not technology, but content itself. "We are essentially a platform deeply rooted in the content industry." She pointed out that "we should use content to empower AI development."

The following is the edited conversation between 36Kr and Ren Xu:

Part 1: "Empowering AI Development with Content"

36Kr: When did Mango start to lay out AIGC-related businesses?

Our company began to make layouts in the field of artificial intelligence algorithms back in 2018, and developed multi-modal AI models. It is fair to say that among domestic long-video platforms, we started the layout and R&D of vertical models very early. We recruited first-class technical talents in the algorithm field through the "Malan Mountain Cup" International Audio and Video Algorithm Competition. In 2024, we built the "Mango Large Model", and on the basis of this model, we built the "Mango Lingchuang AIGC Creation Platform" in 2025.

In 2024, we had a pioneer team called "Prism Studio" that started to develop AIGC-related content. At the beginning of this year, we established a first-level AIGC center. It can be seen that we are developing step by step by combining technology foundation and content creation.

36Kr: What is the core difference between Mango's AIGC business and that of other platforms?

If other platforms use AI to produce content, we are "empowering AI development with content". Our differentiation lies in that we are a platform deeply rooted in the content industry. Hunan Radio and Television has always focused on content creation, and we have abundant content resources and scenarios that need to be upgraded, so the platform will naturally grow and accumulate relevant capabilities.

36Kr: What specific directions are you focusing on at the content level at present?

In terms of content track planning, we will target three differentiated directions, covering children's animations, classic IP innovation, and high-quality mid-to-long dramas for female audiences, taking into account the development of long dramas, mid-length dramas and serialized content.

On the one hand, we will deeply cultivate the AI children's animation track, launch original IP *Gurubi Agency*, co-creation project *GG Bond: Out of Control Era*, build our own AI music children's song label "Mango Miaomiao Children's Songs", improve the children's AIGC product matrix, and accumulate full-link animation production experience.

On the other hand, we will tap the value of classic Chinese-style IPs to build a serialized oriental aesthetic content matrix. Starting from *Post Journey to the West*, the platform will continue to promote the development of Chinese-style projects such as *Soushen Ji* and *The Life of Xin Zhui*; we will create *The Return of Female Generals* to bring together characters including Hua Mulan, Mu Guiying and Fan Lihua to build a female hero universe IP.

At the same time, focusing on the preferences of female users, we will systematically polish key projects such as *Fragrance of Flowers in Ten Miles* and *Jade Color* in high-popularity tracks including sweet romance, emotional stories, suspense and ancient-style romance. Among them, *Feng Qi Ling Long Gu* and *Becoming a Phoenix*, as our first strategic works laid out in the hyper-realistic long drama track, directly benchmark against the production standards of S-level live-action long dramas, and will fully adopt hyper-realistic generation technology in visual effects, narrative rhythm and character logic.

In addition, in the future we will further connect with the PUGC mode, cooperate with outstanding independent creators emerging in the market, open up Mango's IP resources, and invite them to participate in empowerment and cooperation.

36Kr: What does "AI-enabled secondary creation" mean?

Ren Xu: For example, the derivative short drama *Daytime Reverse* derived from the already launched variety show *Busy Looking for Treasures*. Before the show was launched, we had carried out integrated creation with the program team, and made a derived interactive comic drama in advance through AIGC. When the program was broadcast, the interactive content of the comic drama was directly connected with the variety show, and there were clues that could circulate between the two types of content.

36Kr: Is that a comic drama or a hyper-realistic live-action drama?

That is a comic drama. But we are also developing hyper-realistic live-action projects for other works. For example, we may produce hyper-realistic derivative content for some previous TV series. If the main plot is already complete, we can develop the side lines, such as enriching the stories of the second male lead and the second female lead, to produce derivative dramas or character-focused content.

Part 2: How to Solve the Commercialization Dilemma

36Kr: How do you achieve a viable business model? Do you also need to invest money without immediate returns in the early stage of this business?

We don't call it "investing money without returns". Other platforms develop content after acquiring AI technology, but we have abundant content resources first, and the technology is used for content upgrading and integration. We adopt the strategic positioning of "content + technology as dual forward players", so our development foundation is completely different.

Commercialization is definitely a necessary goal. On the one hand, the content produced will bring value to member payment and derivative payment. We hope to bring better experience to members through content upgrading and interactivity, which will be implemented in the member business. On the other hand, revenue can also be generated through content revenue sharing. There are also cooperations on the advertising side — advertising clients are also looking for new creative forms and lower costs, and the automotive and pharmaceutical industries have already started to test related cooperation.

More importantly, we hope to promote the reform of production mechanism and production structure through these content cooperations, so that more participants can use AI to realize content innovation and reduce costs, which is the goal of the integrated development of "content + technology". At the same time, we also hope to finally bring Mango Lingchuang to the market and realize its commercialization.

36Kr: So Mango Lingchuang may be open to external creators for use in the future?

Mango Lingchuang is an aggregated platform, which includes our self-developed models, as well as the latest external models you can think of. Since Mango has long been deeply engaged in the film and television industry, we can access the new versions of many first-class models faster and earlier, which is very friendly for vertical users in the film and television industry. After the platform was developed in 2025, it has been open to the whole market.

36Kr: What are the differences between your large model and other similar models?

First of all, the training base of the model is very important. Mango has rich content resources, and all our technical products are "AI generated from content". We started to lay out relevant technologies in 2018, and we have accumulated more than 20 years of content resources including variety shows, dramas and scripts. We have dozens of PB of copyrighted content, as well as the creative ideas and methods of our own directors. All these are our barriers, and will not be open source to the public.

Secondly, we have made a lot of optimizations on the base model. For example, in the aspect of image fusion, the patterns on ancient costumes are prone to poor light and shadow texture and pattern deformation when processed by general models. After our fine-tuning and optimization, the output effect is much better, which is especially suitable for themes including ancient costume, immortal hero and fantasy.

In addition, we have developed many customized toolkits. For example, the canvas function: if you draw a line, an arrow, or adjust the angle of the "3D camera", you can generate images from a specific perspective. A large number of these seemingly small but very practical functions for film and television creators are integrated into our platform.

36Kr: Which AI functions of Mango Lingchuang have higher usage frequency?

We focus on more refined and high-quality film and television creation, so we hope to provide users with the best models as soon as possible. We hold regular customized exchange meetings with model developers, feed back demands directly to them, and they will also send developers to provide customized training.

In terms of self-developed platforms, the canvas, 3D camera and image fusion functions are highly praised by our creation colleagues for their high usage rate. In addition, we are an integrated film and television creation platform that supports team collaboration, so users do not need to generate images on one platform and then go to another platform to generate videos. You can import the script, and the AI will help you sort out the storyboard scripts, then the director can select and assign the shot production tasks to team members, and the whole process is very smooth. Our team collaboration function is at the leading level in the whole industry, and it is iterated at a very high frequency.

*Post Journey to the West* further combines the above single tools into a project-level production flow: computing power, models and cloud pipelines jointly support character setting, storyboard making, shot production and post-production of finished films, and the assessment standard of the platform's capability has shifted from "whether it can generate a good shot" to "whether it can support nearly a hundred people to collaborate stably under the same aesthetic standard and schedule".

Part 3: AIGC is a New Opportunity for Long-video Platforms

36Kr: You just mentioned that AIGC can improve the efficiency of content production. Do you have specific examples in terms of production efficiency or cost?

We can make a comparison. The cost of traditional 3D animation is very high. To produce a 5-minute animation, a traditional team of 10 to 15 people or even dozens of people needs to spend 500,000 to 600,000 RMB. With the AIGC method, the labor cost is greatly reduced. It used to take several years to train a 3D animator, but now with AI, as long as you have animation thinking and director thinking, you can realize optimization through engineering methods. Now a super team can produce one episode per week. The cost can even be reduced to several thousand RMB per minute, which is a very significant improvement compared with the traditional mode.

In the past, to produce a long animation, buy an IP and make 26 episodes of 10 minutes each, it took a year and a half from planning to launch. If you cooperate with a relatively good production company, the cost is about 80,000 to 90,000 RMB per minute, with a total investment of more than 10 million RMB. It took a year and a half after the launch to know whether the work can make a profit.

Now for the long animation of Cluebie, we started planning in April and launched it in summer vacation, which only took two or three months. The cost is several thousand RMB per minute, which is one tenth of the traditional cost or even lower. The efficiency has been improved by more than 5 to 10 times. A traditional animation company that used to have 20 employees now only needs 5. The same is true for TV dramas: from IP development to drama presentation, it used to take at least one and a half years, but now it only takes 3 to 4 months.

36Kr: You cooperate with top S-level variety shows such as *Who's the Murderer* and *Singer*. What are your criteria for selecting the first batch of cooperative IPs? Are there any difficulties in the cooperation process?

We prefer top shows that can better integrate with AIGC. For example, *Who's the Murderer* is more suitable than *Viva La Romance*. The program teams themselves also know which IPs can be better combined with technology to achieve innovation. In terms of TV dramas, we give priority to ancient costume, immortal hero, fantasy and sci-fi themes, instead of realistic themes.

There are definitely difficulties at the beginning of the cooperation. Everyone is used to the traditional working mode, and the boundary of cooperation needs to be explored. We choose small breakthroughs, do not intervene in the main storyline, and start with derivative content. For example, for the "suspect's self-narration" part, we do not produce the main content first, but make derivative content. After that, the two sides jointly establish a standardized process: discuss at the script stage, implement a double review mechanism, and clarify how to present a picture without conflicting with the main storyline. At present, the derivative short drama *Daytime Night Walk* linked with the variety show *Busy Looking for Treasures* has been launched on Mango TV, and has received a lot of praise for its unique art style and story.

36Kr: Do you have AI anxiety inside your company? Many media companies and large tech enterprises are promoting all-staff AI application.

We don't have such anxiety at all. We are very excited every day, because we clearly realize that creativity is the core, and AI is just a tool that empowers us. Our variety shows and dramas attach the utmost importance to creativity itself, and now AI can help reduce costs and improve efficiency, which is very popular among our employees.

Colleagues from the group have been hoping that we will hold the third phase of training courses, everyone wants to learn how to apply AI to their own work. Our situation is a little different from some teams in Beijing — we interact with AI every day, and we even spend an hour playing with AI after work. It is a feeling that "human + AI" is fully activated and inspired.

In the training camp, all the participants are very enthusiastic. We do not force everyone to produce a certain number of videos or use AI every month, instead, the use of AI is completely spontaneous.

36Kr: AI has also greatly empowered your program editing work, right?

Yes. Since 2018, we have embedded many AI capabilities into our internal workflow, such as intelligent erasing, intelligent editing, AI dubbing, and multilingual translation. These capabilities are invisible to users, but they have helped us save tens of millions of RMB in costs. In the past, we needed to purchase various technologies and hire a large number of employees to complete these tasks, but now the embedded technologies can improve efficiency or reduce costs.

36Kr: In the past two years, long dramas have been greatly impacted by short dramas, which can be seen in the financial reports of all platforms. Now that AI tools are becoming mature, many participants are shifting from short dramas to long dramas, and you also said that you will develop AI long dramas. Why did this evolution happen?

From the content perspective, our exploration of long dramas is not a simple extension from short to long, but a pursuit of quality upgrading. A good story needs a longer narrative space to carry emotions and scenarios, which is the inherent demand of content itself. Many movies and hyper-realistic dramas need long time to build the narrative system to tell stories well.

From the technical perspective, AI was not mature at