ByteDance is building a complete ecosystem of AI video tools.

Meeting creation demands across different creation scenarios with differentiated products, ByteDance is laying a solid foundation for its fast-growing AI content portfolio across multiple fields. This is inherently aligned with the development law of AI tools, which evolves from a single general-purpose tool to vertical Agents tailored for different production scenarios and different service targets.

Meanwhile, although the tool upgrades in entertainment sectors such as film, television and animation are most noticeable to ordinary users at present, the continuous improvement of large model capabilities is further penetrating into more real industries.

In the past two months, ByteDance has made continuous moves in the layout of AI video tools, and has made arrangements in multiple tracks.

In mid-August, ByteDance launched the internal test of "Manju Creation Tool", an Agent targeting the manhua drama industry; on September 21, its overseas brand BytePlus launched Dramagic, an overseas-oriented short drama production Agent; during the Douyin Creator Conference, CapCut released CapCut Hub, an AI video creation entrance that connects AI generation and editing.

From the perspective of products at the application layer, ByteDance has currently built an AI video tool matrix covering Jimeng, Skylark, Manju Creation Tool, Dramagic and CapCut Hub within its ecosystem, which realizes systematic segmentation for different content fields and production scenarios.

Jimeng AI, launched in May 2024, is the earliest AI video generation platform launched by ByteDance. At present, the core functions of Jimeng AI are divided into generation and canvas. The generation function covers image, video, audio, music, digital human, motion imitation and other aspects, while the canvas, as the director's workbench for professional creators, provides extremely high space for manual adjustment, enabling precise control over details such as character costumes, spatial positions and camera movements.

It can be seen that what Jimeng pursues and demonstrates is the upper limit that AI technology can achieve, targeting "film-level creation". In Jimeng AI, the skills displayed in the front positions are also a series of essential functions in high-quality AI content creation, such as film-level long shot camera movement, creation of storyboards, and micro-expression control.

Since its launch, Jimeng AI has always been ByteDance's core platform for connecting top high-quality creators. Since 2025, through a series of activities such as the Future Image Plan and the Young Director Cooperation Plan, Jimeng AI has discovered a group of AI super creators, and sent their works to international film festivals including Cannes, Shanghai International Film Festival, and Busan International Film Festival.

Since the beginning of this year, Jimeng's incubation and support for high-quality creators and works has been upgraded from technology and computing power to a more systematic comprehensive ecosystem construction.

The most representative example is Jimeng Film Studio, a film and television content brand launched in August this year. Continuing Jimeng AI's long-standing support for film-level creation, Jimeng Film Studio solicits works for film and series projects. In addition to technology and computing power, in the distribution and commercialization links, Jimeng Film Studio will also assist selected projects to connect with theaters, provide revenue mechanisms such as box office revenue sharing, joint distribution or guaranteed minimum sharing, and support brand investment promotion, IP derivative development and cross-media cooperation. It will also assist in submitting works to important domestic and foreign film festivals such as Cannes, Berlin and FIRST, and plan offline premieres, themed screenings and customized promotion plans. Jimeng Film Studio's first batch of partners include traditional leading companies in the animation and film and television fields such as Shanghai Animation Film Studio, YHKT Entertainment, Chen Liaoyu Studio, Huanyu Film, and Wanzhong Hexing Pictures.

In February, the launch of Seedance 2.0 not only improved the quality of AI video generation, but also further lowered the creation threshold for ordinary users. Jimeng APP, the mobile end of Jimeng AI, began to show stronger independent product attributes, building a mass-oriented UGC AI content community. To put it simply, Jimeng APP tries to create an AI version of Douyin.

Through the AI avatar function, ordinary users without any AI creation experience can also get a creative video of their own. During the Spring Festival, Jimeng AI cooperated with *La La Land* to launch the 10th Anniversary Re-screening AI Creative Challenge, and launched officially authorized materials. At the same time, it also invited many celebrities such as Li Dan, Li Xiaolu and Huang Yi to settle in the community.

In the Jimeng APP interface, in addition to the "Follow" and "Recommend" columns, the special "Inspiration" column allows users to directly extract other creators' prompts and content to shoot the same style of works, which is very convenient for ordinary users to get started. When you open Jimeng APP now, the mainstream content is all highly interesting and entertaining AI UGC: AI costume change, AI travel shooting, AI song and dance, or playing as a domineering president to break through hail of bullets, cultivating immortals and flying with swords to travel around the world.

The recruitment details of the Jimeng AI Creative Ad Transformation Contest held in July can show the direction difference between the Web end and the mobile end. The Web contest is for manually polished creative works, recruiting high-quality creative clips longer than 12 seconds with plot twists, creative ideas and cool visual effects; the APP contest is for performance transformation works, with a duration of less than 12 seconds, generating creative short videos with the help of avatar performance, supporting one-click follow shot, brainwashing performance and creative secondary creation.

Also incubated by the CapCut team, Skylark's product transformation is also closely related to the launch of Seedance 2.0. In February, due to the availability of the latest Seedance 2.0 model, the number of Skylark users surged.

In March, Skylark launched the short drama Agent, which is the first short drama Agent in the industry equipped with the Seedance 2.0 model. At that time, it was at the node of the all-round explosion of AI short dramas. Skylark attracted the attention of the whole industry with its practical batch delivery capability: no reference characters, storyboard tables and editing are needed, upload the script and story settings, and the system can automatically complete the analysis and generate in batches, supporting up to 100,000-word scripts to be made into a finished film with one click.

In the past period of time, Skylark has also made many technical explorations for emerging film and television products such as "real person + AI" and interactive content. In June, Skylark cooperated with Huanyu Film to co-produce *Beauty*, exploring the mode of "real person shooting + AI", in which costumes, props, scenes, and the integration of real people and environments were all produced by Skylark; in August, Skylark and the Seedance 2.0 interactive model were applied to the interactive short drama *Global Freeze: I Built a Doomsday Safe House*.

In addition to the short drama Agent, another core function of Skylark is the marketing Agent. Skylark marketing Agent supports uploading reference images of characters and products, selecting appropriate creativity, Hook (video hook) and style for generation; the popular creative styles are interconnected with the data on the Douyin platform.

In the two directions of short drama and marketing, most of the customer groups targeted by Skylark are institutions with batch content production demands. The setting of the Agent's core functions points to the simplification of work links and the improvement of overall efficiency.

Skylark's positioning is undergoing new changes. On August 31, Skylark AI announced its brand upgrade, with "Create good stories together" as its new positioning; the relevant person in charge of Skylark AI said: "Skylark AI hopes not only to help creators generate a clip, but to accompany creators to complete a story with characters, plot and world view." This change did not happen overnight. In fact, since April, Skylark has launched multiple rounds of incentive plans for high-quality OPC of story type.

At present, Skylark has signed many super creators known for their world views and characters, such as @Xuan Jiu the creator of *Guixu*, @Mx-shell the creator of *Zombie Cleaner*, @Director Li Cen the creator of *After Ember*, and @Fei Feifei the creator of *The Laid-off Girl*.

On the other hand, in August, ByteDance conducted an internal test of a more vertical short drama Agent called "Manju Creation Tool". It is reported that it is equipped with multiple underlying models of Doubao, with the slogan "Make creation purer".

This Agent is a tool built by ByteDance for the manhua drama track, targeting creators and content companies, especially for batch, project-level manhua drama generation. The interface is very simple, and there is only one script upload function on the home page; after uploading the script, after four one-click selection steps of project planning, script outline, asset design and episode planning, the episodes can be finally generated. The originally built-in different large model options have also been replaced with more clear generation requirements that are more targeted to specific plots.

The biggest innovation of the Manju Creation Tool is reflected in its mass production capability: the maximum supported script length is 200,000 words, and team collaboration supports up to 30 people. At the same time, the ecosystem is interconnected and bound with ByteDance's short drama ecosystem. As soon as you log in to the system, creators will be prompted to bind the account of the Short Drama Creator Center.

A senior manhua drama practitioner shared her experience of using it with *Thirty Generations*: "The Manju Creation Tool is mainly for Hongguo short drama producers; it supports one-click film production, can produce character assets in batches, and even helps you write the prompts after parsing the script, and can generate 100 episodes at a time." She also revealed that in Q3 of this year, some friends in the industry have achieved millions of yuan in monthly revenue sharing with the help of the Manju Creation Tool.

Another AI video tool advancing synchronously is Dramagic, which targets the overseas manhua drama market and was launched by ByteDance's overseas business brand BytePlus on September 21 this year.

Its positioning and strategy are similar to the Manju Creation Tool: through four stages from script, assets, storyboard to video generation, it can "go from script to preview", emphasizing its industrial assembly line capability, which is more suitable for teams and enterprises; it is an Agent for ByteDance to lay out the domestic and overseas manhua drama markets respectively.

In addition, at the Douyin Creator Conference in September, ByteDance also brought AI capabilities to its editing tool CapCut, and launched a number of new AI functions such as "CapCut Hub" and "CapCut Assistant".

CapCut Hub interface

CapCut is strongly bound to the Douyin ecosystem. As an editing tool, it has a very high penetration rate and usage rate among the public, and is one of the core tools to link and empower individual creators to produce UGC content on Douyin.

The biggest breakthrough of this product update is the connection between AI generation and AI editing. CapCut Hub combines infinite canvas and Agent, with the generation and editing processes of text, image, audio and video all built in. The process is similar to creating in an Agent, including splitting the script, planning storyboards, saving assets, and generating previews. The convenient part is that by clicking "More Editing", you can directly enter the usual multi-track editing timeline of CapCut without importing and exporting again.

CapCut Assistant plays a more prominent role in improving efficiency, which can largely replace manual rough cutting, and automatically deal with problems such as slips of the tongue, repeated sentences, and overly long blanks. After the rough cutting is completed, the intelligent packaging function provides applicable video templates that can be directly applied to the editing timeline. In addition, there are many built-in functions such as cutting voiceover, commentary, and AI text-to-video.

In the relevant discussion posts on Xiaohongshu, some netizens commented that "the points are not consumed quickly" and "newbies like me already think it is very powerful". For most users, "AI + CapCut" can solve the problems of video generation and editing in one-stop to a large extent. It is expected to attract more users to participate in UGC content creation in the future, and expand the content scale of the community.

On the whole, ByteDance's product layout in the AI video track is very intensive. AI video content is developing towards high quality as a whole, and AIGC creators are also flowing in different content themes. Objectively speaking, there are inevitably overlaps in functions and ecosystems among ByteDance's various AI video tools at present. Recently, there are news that some businesses of Jimeng and Skylark are about to merge.

However, we believe that ByteDance's current layout in the field of AI video tools is not entirely the result of internal competition. The underlying reason is that ByteDance's content ecosystem is large enough and has high enough diversity. Different platforms such as Douyin, Hongguo and PineDrama cover different content types such as short videos, short dramas, medium and long videos