6 posts tagged with "AI game"

This article is republished with permission from "Game Artist Game Art".

​Played Meow Island — it’s hard to imagine that the team behind this AI-powered game consists of just six people, and none of them come from a traditional game development background.

This is a simulation game developed by Miao Jituo Studio, combining AI cat emotional companionship with island management gameplay. Compared to traditional games like Animal Crossing, the biggest difference is the addition of “AI cats”—cats that can talk to players, understand the player’s intentions, respond to their needs, and feel truly “alive”.

Producer Li Chi previously explored the AI + games direction at ByteDance’s AI Lab. Back then, he encountered many games and became deeply interested in how AI could genuinely change the gaming experience and surprise players.

After founding Miao Jituo Studio, Li Chi and his team developed several games. For them, the most meaningful outcome was not so much an increase in development capability, but a more professional and deeper understanding of what “game experience” really means.

This is crucial for finding a truly valuable way to combine AI with games. And the simulation management genre that Meow Island has chosen is one of the most mainstream directions for AI Native to reshape traditional game mechanics.

This summer, Meow Island was nominated for the second “Shulong Cup” Best AI Game Award and won the gold prize in the AI Game Track of the 2026 Photon Game Competition.

In this game and the team behind it, we can see a progressive path from AI technology to game experience. The singularity moment for AI games has not yet arrived. Besides the limitations of AI itself, the harder part is how to turn the technology into an experience that players can feel, immerse in, and stay with long term.

​Game Artist spoke with Li Chi about how Meow Island has become a valuable exploration in the current era of AI reshaping the game industry, using simulation management as a starting point to find more possibilities for combining AI and games.

In June this year, Meow Island launched its first playable demo. With a cute Q‑style art, light simulation management gameplay, and story simulation experience, players could already enjoy a relaxed and fun island life: adopt AI cats with rich personalities and emotions, chat with them, play games, fish, farm, and freely decorate their own home.

Compared to traditional simulation management games like Stardew Valley or Animal Crossing, Meow Island aims to use AI to break the boundaries of pre‑designed content, so that characters, stories, and even the entire island can change with the player’s choices, delivering a more personalized experience.

The core change comes from the AI cats, each with distinct personalities, emotions, and memories. Driven by agents, the AI cats can act autonomously, interact with players, remember what players like, and turn the player’s ideas into tangible in‑game objects.

Li Chi says one of the positive feedbacks for him in making games is that the AI cats really surprise players.

Three years ago, when Li Chi was still working at ByteDance’s AI Lab exploring the AI + game direction, he encountered the field of AI NPCs. He quickly realized that using AI just for NPC chat was boring and not the kind of experience game players actually enjoy.

Just that year, the “Stanford Town” paper caused a sensation. Inspired by it, Li Chi’s team also ran a similar AI town experiment, placing more than 30 AI NPCs modeled after colleagues into the same virtual town, letting them move around and socialize autonomously.

During the experiment, the AI kept generating unexpected stories: colleagues who were close in real life often argued in the game, while a male and female colleague who barely spoke in reality developed a romantic relationship in the game.

This became the team’s source of joy and showed Li Chi a new possibility for combining AI and games—players could act as observers, watching the natural interactions between NPCs.

Li Chi told Game Artist that the team even reported the idea to Zhang Yiming, and ByteDance’s founder also found it interesting.

But at that time in 2023, AI technology wasn’t ready for a formal project: on one hand, the cost was high—running an experiment could consume hundreds of US dollars in tokens per day; on the other hand, they hadn’t found a gameplay that perfectly matched the core framework. Eventually, the experiment didn’t turn into a real project.

The same year, ByteDance adjusted its game business. Li Chi, who still wanted to pursue game development, gave up the opportunity to transfer to Douyin and instead chose to start a company in the AI games direction he had been researching, founding his own studio, “Miao Jituo”.

Today, Miao Jituo has eight members: six people and two cats acting as mascots. Starting from the AI town, the team wanted to preserve the autonomous interaction of AI characters while making the experience differentiated. So they chose an “island” as the game space—it is naturally isolated and separated, suitable for both single‑player experiences and easy addition of multiplayer features.

​Breaking Pre‑written Narratives，AI Games Can Be “a Thousand People, a Thousand Faces”

Before Meow Island, Miao Jituo had already explored many directions for combining AI and games. Li Chi admitted that the team mainly came from a technical background and initially lacked game design experience. Over the past three years, they made many mistakes, and several projects were abandoned because the combination of AI and gameplay wasn’t ideal.

How should AI truly enter games to bring new gameplay and experiences that are hard to achieve with traditional methods? This is a key question for the industry today, and the Miao Jituo team’s understanding has gone through several iterations.

Li Chi initially had a framework: AI could create content that is difficult to produce through traditional means, such as generating game assets and storylines in real time. The team’s earlier games—Mengzhua Pai Dui, Mengzhua Da Luan Dou (originally titled Yanling Jihua), and Bao Wai Jiu Yi—all started from this direction.

Through practice, they found that AI generating large amounts of content in real time could indeed bring a different game experience, but without optimization, the content could become redundant and variable in quality, making players feel bored.

So in the second stage, the team started to “subtract”: they would rather delete low‑quality content to ensure that the AI‑generated game experience was not inferior to manually crafted games.

This process initially relied on human judgment. The team accumulated a large amount of game experience, structured and tagged cases, and further recorded “what was good and why it wasn’t good”. As data accumulated, they used it to train models, letting agents assist in judgment to improve efficiency.

However, game experience is hard to fully quantify. The team has a set of basic principles: AI‑generated content should be concise enough to avoid information overload, leaving room for players to think and imagine; more importantly, it should bring a sense of surprise.

This surprise could come from AI’s connection to the player’s personal experiences. For example, if a player owns a cat in real life, they might encounter a very similar cat in the game. Such content is not pre‑written but generated by AI based on the player’s background.

In traditional simulation management games, the storylines and NPC interactions are mostly pre‑designed, so when players replay, the dialogue and events are largely fixed. Meow Island aims to use AI’s “emergent design” to break this preset, making the game a “living” world: characters, events, and stories all change based on the player’s choices, preferences, and experiences.

What the player has done before, what they like or dislike, all become part of the content generated later. Each player and each island can develop different stories, and they can extend infinitely.

In the team’s view, this might be the greatest value of combining AI with games.

“Dawei from miHoYo once said he wanted to make a truly ‘thousand people, thousand faces’ game, where every player has their own exclusive experience,” Li Chi said. “We are doing something similar.”

In Li Chi’s view, Meow Island cannot be just a “pure research” game that demonstrates AI capabilities; it must be genuinely fun, so that players will appreciate it and pay for it.

And to make players stay, the first problem is whether the AI NPCs can “come to life”.

In Meow Island, the team hopes that the cats not only have unique “cat personalities”, but also change appearance, state, and personality as they interact with players, growing through shared experiences, becoming “digital cats” that accompany players long term.

To ensure these cats are smart enough and say interesting lines, Meow Island has built a “Model Mesh” on top of a multi‑agent framework with mixed models, matching different models to different tasks: stronger models for complex tasks, lighter models for simple role‑playing and chat, to solve performance, cost, and latency issues when running 30–50 agents simultaneously.

The choice of “cat‑form” NPCs rather than “human‑form” was also deliberate.

Li Chi believes that pet images are more suitable for AI companionship at this stage. Human emotions and cognition are more complex, making it difficult for AI to simulate a real person stably over time. In contrast, players’ expectations for pets are lower; a cat that is occasionally forgetful or looks a bit “dazed” might even be endearing.

Of course, from the studio name to earlier titles, it’s clear that the team really loves cats.

Li Chi thinks that besides attracting cat lovers, Meow Island targets two other types of users: players who enjoy simulation management, building, and social interaction; and younger users who are interested in AI itself and already use various AI products. Especially young people growing up with AI may want a “second world” where they can express themselves freely, beyond real life.

Based on this, Meow Island wants to further expand the freedom of AI + UGC in the game. Players can use AI to generate buildings, flowers, and grass on the island; they can also create social mini‑games like “Truth or Dare” or “Draw and Guess” to play with others.

Previously, this kind of play style was limited by token costs. But now, Li Chi says the cost of running AI in games has dropped very low. The technical model is somewhat like ChatGPT and DeepSeek without calling APIs; a player might play for twenty hours and consume only a few cents. So the team’s focus is less on cost and more on how to retain players long term.

From the feedback after the demo launch, players gave many positive comments. Besides finding it “fun”, they also developed an emotional attachment to the island and the cats, feeling that they gained companionship and spiritual comfort.

Li Chi himself enjoys casual simulation games, using them to escape the hustle and bustle of city life. “Just hearing those cat sounds, farming and fishing—it recharges you.”

In the future, the team will gradually improve the multiplayer social system, exploring moving from “AI companions” to an “AI society”.

In Li Chi’s view, the “AI society” is already taking shape. In Meow Island, this “AI society” will take a more concrete form: AI pets and NPCs gather on the same island, chat, play games, become friends, and generate new stories through interaction. This is the “AI life” that AI games can create.

Miao Jituo Studio was founded in 2023, precisely when large language models became the trend, and discussions about AI restructuring the game industry swept the field. Three years later, Li Chi still finds it hard to judge when the real turning point for AI games will come.

“Maybe we’ve only explored the tip of the iceberg; 90% of the domain is still in complete fog, and the full map hasn’t been opened yet.” It’s too early to determine when the entire AI game track will take off.

But some possibilities have already emerged. For example, History Simulator: Chongzhen — Li Chi thinks it captured a core element: the “emergent design” that Meow Island is also working on.

Although high freedom comes with many problems—such as how to maintain AI consistency, NPCs forgetting what they said before, disconnected story contexts, logical fallacies, etc.—that doesn’t stop it from being loved by some players, and the direction for optimization is already clear.

Li Chi believes that AI’s value will differ across game genres. For example, highly mature genres like MOBA already have well‑defined core gameplay and art systems, and don’t rely heavily on AI to generate new mechanics or content, so the incremental value AI can bring may be limited.

Furthermore, because AI‑native games have not yet formed a clear user base or mature community, the distribution logic of traditional genres is hard to apply directly. New ways may be needed to reach potential players who are interested in AI‑native experiences.

For now, Miao Jituo’s most important goal is to finish Meow Island completely, and then move on to a truly free open world. In terms of business model, Meow Island leans toward a long‑term operation game that continuously supplies content, referencing monetization methods of mature products like Heartbeat Town.

In fact, over the past three years, what truly brought the team to break‑even was not a game, but a self‑developed AI game creation tool—Meowa.

While developing Meow Island and other games, the team built tools and technical pipelines for different needs and gradually integrated these capabilities into Meowa. As more users wanted to use this tool, the team decided to open it externally, and it now has over 30,000 users.

Recently, the team also plans to start a new round of financing—several million US dollars—which will be used mainly for two things: advancing Meow Island’s development, and pushing Meowa from a development tool toward a platform.

Li Chi uses the relationship between Valve, Steam, and Half‑Life as an example; in the future, he wants to build Miao Jituo’s own game platform. He envisions that if large‑scale, high‑quality game generation becomes feasible, the distribution logic of the platform might also change.

Unlike the “shelf‑style” display of traditional game platforms, players might prefer to discover and experience games through recommendation feeds, forming a model similar to “game version of Douyin”. This is an area many game creators are interested in, and investors are more willing to pay for it.

“Everyone is very curious about what the next ‘Douyin‑style’ game content platform will look like.”

However, all of this depends on whether these AI games themselves are high‑quality and fun enough to truly attract and retain players.

Li Chi says Meow Island wants to explore more cutting‑edge possibilities of combining AI and games. It may still need some polishing and optimization time, “but we will definitely do our best to make it good and serve everyone a truly delicious ‘dish’.”

This article is reproduced with permission from "Game Art".

​In recent years, AI has already demonstrated powerful content generation capabilities in text and video domains. Many small teams and indie game developers have widely applied AI to NPC dialogue, art generation, and other areas. However, in the generation of interactive 3D spaces, the industry is still relatively early—modeling, layout, materials, lighting, and rendering still rely on professional software and trained creators.

If AI could generate game scenes, spaces, and even go a step further to build an interactive virtual world where the game world is no longer entirely pre-built by developers but instead allows players to participate through AI, co-creating, sharing, and enriching virtual spaces—this sounds a bit like the "metaverse" from a few years ago, but the underlying technological implementation is already completely different.

Recently, SenBOX won the Best AI Game Nomination Award and the "Most Commercial Value Award" at the second "Shu Long Cup" Global AI Innovation Awards hosted by Century Huatong. The product's developer, Shengjing Technology, chose to start from "home": users input a sentence, a reference image, or even just a vague inspiration, and the system can generate a 3D home that can be further edited.

Since entering the public eye last year, SenBOX has accumulated nearly 2 million fans across the entire network, with a total exposure exceeding 300 million. Reservations on TapTap have reached 240,000, and about 4,000 users participated in the beta test, generating 1.2 million room states. Currently, the product is about to open a new round of testing and is expected to officially launch in October this year.

Taking this opportunity, we had a conversation with Zhuang Ziyang, co-founder of Shengjing Technology, and found that what they want to do is far from just an "AI home-building game." The team is more concerned about whether spaces, like text and video, can be created, distributed, and traded by ordinary people with a low barrier.

​Why go from a B2B space design tool, to a consumer-oriented game?

In 2023, Shengjing Technology was founded in Nanshan, Shenzhen. "Shengjing" means "generating virtual environments," which has been the company's focus from the start—a 3D space generation system. This young team, cultivated by Chinese Academy of Engineering academician Meng Jianmin and Professor Li Zexiang, now has a scale of nearly 100 people, with R&D technical staff accounting for 90%, and has accumulated nearly 100 million yuan in investment from institutions including Nanshan State-owned Assets and Sequoia China.

Zhuang Ziyang told Game Art that the team initially had no plans to make a game; they were developing a set of AI space design tools aimed at the B2B market.

In the traditional interior design and architecture industry, a space from design to realization typically requires a complex process including modeling, placement, modification, and rendering—long cycle times, high costs, and heavy reliance on professional skills. What the team wanted to solve was this very practical pain point: using AI to industrialize "one-click space design."

As a result, early on, Shengjing Technology mainly served home furnishing brand clients such as Panasonic and Nature Home. The real motivation for the team to shift to the consumer side came from user feedback.

The team estimates that about 70 to 80 million young people rent apartments each year and have the need to decorate their rooms. However, professional software like 3ds Max has a high barrier, making it difficult for ordinary users to complete designs on their own. On platforms like Xiaohongshu, a large amount of content has emerged where users share room designs and seek space renovations. There are also many young users who don't necessarily have real renovation needs but enjoy the fun of designing and creating ideal spaces.

The team thus realized that a space creation tool that is simple enough and gamified enough might have a larger user base than professional tools.

During this process, more user feedback validated the feasibility of the product direction: when the product was still positioned as a tool for the home furnishing industry, the related account had only a few thousand followers; after shifting to a consumer-oriented design tool, followers grew to about 50,000; after further turning it into a game, the follower count quickly exceeded one million.

"We spent zero on traffic acquisition—it's still because young people genuinely love it and want to play this kind of AI simulation home game," Zhuang Ziyang said.

Currently, SenBOX's target user profile is predominantly young women under 25, a conclusion drawn from the team's analysis of tag data on platforms like Xiaohongshu and Douyin.

In Zhuang Ziyang's view, a good digital home must first evoke emotional resonance, which is why the product chose the 3D realistic style of "home" as its entry point.

"Home is the easiest space for users to form an emotional connection. If you arrange an office or a forest, users' desire for expression might not be as strong. With the current popularity of 'homebody culture,' people spend a third of their time at home. Starting from the needs and experiences of simulating real life has a lower barrier, and users can understand it more easily."

SenBOX's current user comment section is like a small cultural sample. Some are maximalists who like to fill their rooms with stuff, while others pursue minimalism; some are obsessed with Bauhaus style, others prefer cream style or cyberpunk style... Different MBTI types and zodiac signs also have their preferred room layouts, and some pay attention to Feng Shui.

Shengjing Technology is continuously drawing inspiration for product iteration from these personalized feedbacks.

​Spatial model, letting AI understand the order between objects

SenBOX's current core gameplay is "one-sentence generation." After the space is generated, users can drag, drop, edit, and fine-tune; they can also take photos to convert real objects—like figurines or water bottles—into 3D assets and place them into the virtual space, creating a connection between virtual and real.

When AI handles the tedious basic placement work, users can focus their energy on aesthetics and content creation. For example, a beautiful house needs 300 pieces of furniture; AI can first help users place 200 basic items, and then users add 100 more and DIY the details.

According to the team's plan, generated spaces will expand from rooms to buildings, streets, and cities. Users will also be able to control and upgrade more components, experiencing gameplay such as visitor socializing, quest bounties, and item collection.

Behind this product roadmap lies Shengjing Technology's ongoing investment in the "spatial model."

Zhuang Ziyang stated that text models solve the order of words, image models solve the order of pixels, and spatial models solve the order of 3D assets like tables, sofas, doors, and windows in space. Each is like a computable symbol containing coordinates, dimensions, properties, materials, and other information; different layouts and room types can combine into a wide variety of styles.

This structured spatial encoding path allows the system not just to "paint" a space at the pixel level but to understand the positional, scale, and functional relationships between objects. According to Shengjing Technology's data, generating a 3D space currently takes only 15 to 20 seconds, with relatively low token consumption.

Regarding practical requirements like circulation and aesthetics in home design, Zhuang Ziyang said the model can already basically understand them. On one hand, comprehensive layout rules refer to tacit knowledge from many furniture brands and designers; on the other hand, it also learns color matching and visual rules in different styles through image data.

It is understood that during the beta phase, SenBOX has generated 1.2 million room states. Every user edit, drag, drop, and rearrangement becomes training data for the model to understand spatial rules, forming a positive data flywheel.

On this basis, Shengjing Technology has also accumulated a million-level 3D asset library, tens of millions of image-text data, and spatial coding experience accumulated through partnerships with home furnishing brands. Zhuang Ziyang expects that in the future, there will be hundreds of millions of space-state data generated by millions of users, and scarcity will gradually increase.

The team's next-stage goal is to achieve deeper interaction between humans and spaces, such as allowing virtual characters to understand "how to pick up a cup and drink water." He estimates this capability will still take one to two years to mature.

Zhuang Ziyang believes that compared to traditional home-building games like Animal Crossing: New Horizons and Xindong Town, SenBOX tries to lower the barrier to creation. Traditional home-building games rely on manual construction, requiring users to invest time in learning operations and accumulating resources, while SenBOX uses AI to complete the basic generation, allowing users to quickly enter the expression and adjustment phase.

This also aligns with the current R&D trend of AI games. AI brings not only content production efficiency but also changes in content diversity and gameplay structure. For example, traditional games have limited levels, but AI can continuously generate 3D content, giving the game world potential for constant expansion.

Moreover, the driving force of traditional games mainly comes from completing tasks and unlocking achievements, while SenBOX hopes to form a loop based on content creation and aesthetic selection among users. When users design rooms around the same theme, comparisons, competitions, and natural selection occur, and high-quality creators gain attention.

Therefore, Zhuang Ziyang prefers to take Roblox as a long-term reference. The similarity is that both aim to create low-barrier, diverse, and high-freedom worlds; the difference lies in the content production foundation. Roblox relies on a developer ecosystem and toolchains, while SenBOX hopes to use AI to turn space generation into an action that ordinary users can complete.

"The world needs people, goods, and scenes to run. People can be users themselves, scripts can be driven by AI and agents to interact with NPCs, and scenes are the problem that has never been well solved before," Zhuang Ziyang said.

SenBOX aims to solve exactly this core problem. Just as in the short video industry, the emergence of CapCut reduced the cost of video creation, allowing ordinary people to become content producers and triggering a huge content explosion, SenBOX hopes to use the same logic to lower the barrier to 3D space creation and boost the supply of 3D content.

​More than just a game, also aiming for space distribution, e-commerce, and data closed loop

Therefore, SenBOX's ultimate direction is not limited to a home-building game; Shengjing Technology wants to build an AI space creation UGC platform.

The team does not shy away from the word "metaverse," but gives a more concrete definition: enough content, low enough barriers to entry and editing, and users can socialize and trade within it.

In Zhuang Ziyang's view, the metaverse concept didn't really take off in the past, and the core issue is not just the concept itself but insufficient content supply and high creation costs. If AI can reduce the threshold and cost sufficiently, the boundary between game worlds and virtual living spaces will blur.

On this basis, SenBOX also plans to establish its own content distribution system. Users can browse 3D space-related content or enter spaces to interact; creators can gain attention through their design abilities, forming an ecosystem similar to that of home furnishing bloggers and space design bloggers on various platforms.

For example, a musician's themed room might evolve into a reusable theme template, and a basketball-themed space could expand into an interest community.

"There will be trending lists. Just as Douyin recommends high-quality content, we will recommend high-quality spaces. In a sense, it's like a space version of Douyin, but carrying the next generation of digital life content," Zhuang Ziyang said.

He further explained that the "space version of Douyin" is more suitable as an analogy for distribution logic rather than copying the short video product form. The platform needs to solve how 3D space content is recommended, entered, reused, and how creators get sustained returns.

In terms of monetization, SenBOX is not limited to in-game purchases. The team stated that as the UGC ecosystem forms, the platform may establish a trading system through virtual asset transactions, brand partnerships with home furnishing companies, advertising, value-added services, etc. Real-life home furnishing brands can enter the virtual space; users can pay for virtual furniture and may also connect to real-world consumption; if creators can drive consumption through space design, they should also receive corresponding incentives.

Currently, among Shengjing Technology's three existing business lines, SenBOX serves as the C-end space creation entry, SenHOME handles space e-commerce and brand connections, and SimHub provides spatial synthetic data for embodied intelligence. The three product lines share a unified spatial model foundation, corresponding to content creation, industrial applications, and robot training data scenarios.

Zhuang Ziyang stated that the longer-term value lies in spatial data. Compared to text and images, real, usable, and structured 3D spatial data has been accumulated for a shorter time and is more scarce. Every placement, adjustment, and edit by users, under authorization and compliance, can feed back into the model, helping it understand spatial rules, and further support B2B businesses like SenHOME and SimHub.

Whether this closed loop can truly succeed still needs validation after the product launches, including whether C-end users are willing to continuously create, whether the creator ecosystem can form stable supply, and whether B2B clients are willing to pay for spatial data and generation capabilities. More than exposure volume, retention rate, creator activity, and transaction conversion will more directly determine SenBOX's ceiling.

When asked about future vision, Zhuang Ziyang said: "We hope that in the future, SenBOX can truly bring people a better life. In the short term, it can solve the needs of home renovation and room decoration; in the long term, it may become part of people's spiritual world."

This article is reprinted with permission from “Game Artist.”

Not long ago, ByteDance officially announced Seedance 2.5 at the 2026 Volcano Engine FORCE conference, expected to launch in July. In addition to revealing upgrades such as native 30-second video output, joint generation of 50 full-modality assets, and consistency-preserving local editing, ByteDance also put forward judgments like “video generation is one of the paths to a world model.”

Every technological iteration naturally sparks discussion about the boundaries of its application. Whether it’s the expansion of AI video technology’s real-world use cases or the future innovation of AI-driven interactive entertainment, interactive film and gaming (interactive film-games) is one of the most watched directions right now.

Not long ago, ByteDance Games’ Jiangnan Studio publicly disclosed the Bu Wen Fan Chen project. This cultivation-themed product, developed based on AI video models like SeeDance, has its characters and scenes entirely generated by AI to create photorealistic images.

The sheer volume of video performances makes it look like a well-produced interactive film-game. Yet after talking with the team, you realize it’s a complete-loop cultivation interactive product.

This product uses turn-based combat, features multiple main storylines and dozens of endings, and includes complete cultivation loop mechanics such as sect development, pill refining, equipment crafting, auctions, and world bosses. It’s understood that the full content is expected to last around 30 hours and will be released on Steam within this year.

Currently, interactive film-games have attracted participants from many quarters. Film and TV teams take interactive storytelling as a breakthrough; AI-native teams start from technology application; platform players try embedding lightweight interactions into information feeds… The backgrounds of entrants vary, and the product form is not yet settled.

What makes Bu Wen Fan Chen worth noting is that its production team provides a way of thinking that starts from gameplay and industrial pipeline construction and works backward.

Before Bu Wen Fan Chen, producer Xin Jie had already led his team to develop three AI interactive products. But it was only after they “imagined the form of a world model after its implementation and built a content engine accordingly,” deeply integrating AI video technologies like SeeDance with the product’s original gameplay pipeline, that they truly discovered the feasibility and necessity of AI interactive products.

In Xin Jie’s view, what this direction may bring in the future is a content engine based on real-time task calculation, one that better understands context and can better present content based on interaction, eventually having the chance to become the content carrier of the next era. And the boundary between film and games will grow increasingly blurred in this process, eventually converging at certain points.

Recently, Game Artist / Things Entertainment had a detailed conversation with Xin Jie, producer of Bu Wen Fan Chen, around topics such as the product boundaries of AI interactive content, the construction of AI technology pipelines and product landing paths, and the potential differentiation and future opportunities of AI + interactive film-games.

“Bu Wen Fan Chen” is your fourth AI interactive product, and it already has a strong “AI interactive film-game” feel. How would you define this product?

Xin Jie: My definition of an AI interactive product is: could this thing have been created without the age of AI? Bu Wen Fan Chen is precisely a product that can only be made using AI. We often joke internally, calling it a “live-action cultivation” interactive product. It doesn’t have the common anime or 3D art style; from the first glance it sets itself apart in terms of experience.

Secondly, it’s a cultivation-themed title. Cultivation is very much like China’s native version of anime culture – it is rooted in the core of Chinese culture and has formed a sense of community gathering and emotional resonance. There are many popular cultivation anime and TV series now, but not many interactive products, especially those offering rich gameplay content. We hope its ultimate experience is “free cultivation,” like a “playable cultivation novel,” without being too large in scope.

In essence, we are building a content engine based on AI video. Bu Wen Fan Chen is the first work produced by this engine at the current stage of technology that can satisfy both us and the players.

How do you distinguish the content boundaries of AI interactive film-games, AI RPGs, AI visual novels, etc.?

Xin Jie: I think the core lies in what the player’s consumption target and actual experience are.

For most interactive film-games, the main focus is still “watching.” They are rooted in series experiences, like early works such as Black Mirror or The Invisible Guardian, where you accompany the protagonist through different choices and see the outcomes, but the choices are limited – you mainly “watch” a great story.

But Bu Wen Fan Chen emphasizes “playing.” Every moment presents dense choices – where to go, which sect to join, whether to fight someone – these choices accumulate in non-narrative ways, affecting your stats and how the entire world reacts to you.

For example, if you kill an NPC, your alignment changes, good people in the city may refuse to trade with you, but other villains might recruit you. You decide most things; you are cultivating “your” path to immortality, not watching someone else’s story.

So the boundary lies in the mode of participation and interaction. For instance, playing means experiencing “making sufficient choices in this world” and “having to bear the consequences of my decisions.”

Does that mean the cultivation theme itself is more suited to this kind of rich, free-form gameplay, giving it a better chance in the AI era?

Xin Jie: Yes. The cultivation theme emphasizes free exploration; it is more sandbox-like. The player imagines being thrown into this world, choosing sects, learning techniques, making friends or enemies with various people. It’s unlike some anime themes where the main storyline is a linear journey with companions – cultivation is more like a “sandbox.”

More importantly, cultivation can essentially be seen as China’s sci-fi genre. Unlike wuxia’s flesh-and-blood combat, it is full of high-concept, grand-scale fantasies. These conceptual designs are extremely difficult to realize in traditional film or games – expensive and hard to deliver. As a result, a large portion of the user base accumulates in literature. And for a good genre, people naturally want to see its movies, TV shows, and games.

AI happens to solve this problem, visualizing the cultivation scenes and combat skills in people’s imagination cheaply and effectively – not just making them, but making them to a degree that satisfies players.

Rebuilding the content engine and industrial pipeline, using AI video models

Traditional live-action cultivation interactive film-games have huge costs and production difficulties. How about the cost of Bu Wen Fan Chen using AI?

Xin Jie: Let’s look at two aspects. First, what we are building long-term is a content engine; the infrastructure and technology investment themselves are much higher than simply setting up a single project. Moreover, as a complete interactive product, besides video interaction, we have a dozen additional systems including combat, pill refining, equipment crafting, sects, auctions, world bosses, etc. This is heavy industry in the content domain – system, progression, stats, and story are all indispensable – so the cost naturally rises.

Second, in terms of content production, we have set a very high quality standard, definitely at the leading level in the industry. Part of that is hiring people who are very good at AI content, and part is our own strict internal requirements, resulting in a high waste rate and investment.

What is the difference between using an AI video engine and using traditional engines like Unity or Unreal Engine?

Xin Jie: From a technical logic perspective, a traditional engine is a “spatial engine.” You place objects in 3D space and then render them back into a 2D image for the player via the graphics card.

What we do is more like a “temporal engine.” It replaces the core physics and rendering issues in parallel on a technical architecture. It renders based on AI, slicing the expressive content of all possibilities along a timeline, forking, overlapping time, or fusing video, then presenting it to the player. The difference in technical logic is that the latter combines AI video gameplay with interaction.

We chose this path because its visual expression can surpass most interactive products, offering stronger immersion, while serving more players with better content presentation. However, there are some short-term limitations, such as action-oriented gameplay with strong interaction, which may have to wait until the world model is implemented before gradually migrating.

AI video technology like SeeDance is also widely used in AI micro-dramas, AI movies, etc. How is it different when used in interactive products?

Xin Jie: The biggest difference is that you have to build an entire product pipeline from scratch. How does the video-based story play? How is the logic implemented? How is content managed? Most interactions are video-based – how do you design gameplay around video? This is very different from traditional development.

Currently, most of our video is still pre-generated. But as costs decrease and quality improves in the future, there will be more online real-time generated content. How to create a good feedback loop between player choices and real-time generated video will bring completely new experiences. I predict that by next year, we might see some decent online-generated content. This will have a huge impact on film, games, interactive entertainment, and so on.

This touches on a concern many developers have. AI technology evolves so fast that a newly built workflow might be upended by new tech soon. What’s your take on that?

Xin Jie: I felt this deeply earlier this year. Those companies doing Agent work or AI engineering architecture for B2B instantly follow every new model – this has become the new normal. What matters more is which team has delivery capability at which stage, and whether they can quickly follow new tech, produce something, and commercialize it.

Back to the question, the key is whether your team builds the pipeline on top of AI, or merely uses AI as a tool. If your pipeline’s connecting points are all human, and you only use AI to improve efficiency within traditional software, then the bottleneck will always be people.

A true AI team integrates many steps. For example, AI-generated video merges scene, set design, lighting, cinematography, directing, acting, costumes, and props all into one person. Going further, let Agents participate in the mass production process – while you sleep, they are doing review and optimization.

How do you know if you’re doing it right? Just ask yourself: when a new model three times more powerful appears, are you happy or sad? If happy, it means you did it right – you can immediately use it to accelerate production. If sad, it means you’ve been building a “shell” on an unstable foundation.

Has the Bu Wen Fan Chen project itself ever had to overturn its original creative plan due to model updates?

Xin Jie: Rarely, because from the very beginning I imagined the form after the world model’s implementation and built the entire content engine accordingly.

But it’s not nonexistent. The most typical disruptive node was the emergence of SeeDance. Before that, we were more like making an animated film, connecting keyframes. After it appeared, our work mode completely changed – more like a director: you tell it the script and design, and it can give back something decent. Such disruptive technology nodes bring pipeline iterations.

When transitioning from traditional interactive content creation to AI-native content creation, have you or your team experienced conceptual disagreements? What was the biggest disagreement, and how was it resolved?

Xin Jie: Definitely. Most of the disagreements centered on judgments about the pipeline. I tend to be more aggressive in predicting AI development. For example, the idea for this project was proposed last September. At that time, SeeDance hadn’t come out, and people didn’t recognize Agent capabilities. Choosing that direction at that point was quite a gamble.

Because the window for AI content products is about half a year – you have to predict what the tech will be like when you mass-produce half a year later. If your prediction is too aggressive, you might not finish. If too conservative, what you produce might lack appeal. This is a real challenge for current AI teams.

The biggest disagreement was not about AI itself, but about product direction. I was still more determined to first make an AI-based interactive product – that was the entry point and gameplay approach I envisioned from the start. Within the team, this kind of issue is fairly normal.

The Bu Wen Fan Chen team has fewer than 20 people. How is the work divided? Compared with traditional development, are there any new roles or workflow changes?

Xin Jie: The core is still programmers, designers, and artists – you can’t skip them. But there are changes in implementation details. For example, since coding is not a bottleneck, we require designers to also output HTML interaction prototypes for validation.

Additionally, we have two very key roles: one is the engineer who builds internal AI tools and pipelines; the other is the director. Now that AI is combined with film and TV, I very much hope that directors who understand cinematography and performance can spark with AI, expressing the design of story and interactive content.

The development rhythm has also changed. Traditional AAA games follow a very strict pipeline: half the time in pre-production, then a full schedule of mass production for the next year or two. But for AI single-player products, it’s not like that. You need to set up the pipeline, but at the same time, every two or three months you have to adjust gameplay and pipeline according to changes in AI technology – more like an agile development model.

When a large amount of work is taken over by AI, what do you think is the core value of human creators?

Xin Jie: This is like after the camera was invented, painters no longer needed to perfectly record reality, which gave rise to Impressionism and Abstraction, liberating everyone’s way of creating content. AI is the same.

First, it lowers the threshold for creative implementation. Everyone has a desire to express themselves. AI helps people release their inner thoughts, making expression easier. Just like Douyin’s success owes a lot to the massive content provided by mid- and long-tail creators – people discovered that videos don’t have to be masterfully produced to be consumed.

Second, it will bring new content carriers. Interactive content has always been a basic human need – when I watch something, I also want to make choices. But in the past, due to costs, timelines, and traditional investment logic, it was difficult. AI makes it possible.

Third, looking further ahead, with the development of world models, it may give birth to the content carrier of the next era, completely overturning past experiences. Even being a bit more radical, maybe not all expression is exclusive to humans; AI itself can generate good content for consumption. Setting aside productivity and social structures, for consumption alone, this is something to look forward to.

It’s said that 95% of the AI-generated material for Bu Wen Fan Chen was scrapped, and only 5% was usable. What criteria were used to select that 5%? Has the team formed a reusable “aesthetic consensus”?

Xin Jie: A 95% waste rate roughly means we set a quality bar far above the current average capability of AI. To reach that quality, you still can’t rely entirely on AI automated production.

This “aesthetic consensus” was brutal – we only truly found it one month before SeeDance emerged. For four or five months before that, there was nothing that made the team feel “at least acceptable.” Once we had usable technology, the consensus was built based on those clips that the team collectively approved of.

But there’s another layer above that: design. What a character looks like, what clothes they wear, what scene they’re in – all of these need separate design. AI can make the design land quickly, but that doesn’t mean design is unimportant.

What your male lead wears, what background an NPC has, whether a fire-themed sect uses magma as its main body or has a volcano erupting behind it, how clothing reflects hierarchy from sect leader to underling – all these still require effort in design. AI cannot think these through for you.

If I lowered the quality to what current AI can output directly, I wouldn’t need to spend time on this. But essentially we want to make good content. Actually, it’s also a matter of product choice: do you make a good AI content now, or wait a bit? Many problems that bother you now, like character expressions not looking human enough or repetitive movements, may not be issues by next year.

Bu Wen Fan Chen has multiple main storylines and dozens of endings, and also supports dynamic NPC dialogue and personalized interactions. In AI-driven open-ended content generation, how does the team keep the worldbuilding and character settings from collapsing?

Xin Jie: This part is still firmly controlled by humans. Most of the text and copy is still designed and written by people. It’s difficult to let AI generate complete and rigorous content right now. For example, the several sects, their histories, hierarchical relationships, character connections – all have to be designed first. This not only affects the story but also impacts gameplay and plot.

The boundary between film and games will get increasingly blurred in the future

Are you worried about players complaining about the “AI feel”? Are there any specific designs to address this?

Xin Jie: We don’t shy away from the fact that it’s AI-made. When we released the first trailer and asked players, “Can you accept AI?”, the comments were roughly split 50-50. Then we invited players for internal experience testing and found that after actually playing, the acceptance of AI rose to around 90%.

This shows that when players are immersed in a complete, high-quality experience, as long as the content is good and doesn’t feel out of character (OOC), they accept this style. An OOC experience can be jarring for players. Even though it’s AI-made, we try to minimize that.

I believe “AI style” in the future may become like anime style, Ghibli style, or 3D animation – just a new visual style brought by a new production paradigm. Whether you shoot on a phone or on film, it’s still a movie. Ultimately it comes down to whether you offer good content. What people dislike is not AI, but shoddy production.

From a creator’s perspective, with the boost from AI, will interactive storytelling and film/TV storytelling converge or remain independent? What new creative possibilities do you hope to see from their collision?

Xin Jie: My judgment is that in the future, the boundary between film and interactive content like games will become increasingly blurred.

For example, a film/TV team that makes a first interactive film-game will likely want to add progression mechanics and different gameplay elements to it. Conversely, a game team that builds an interaction framework will want to fill in more NPC backstories and narrative expressions. Given enough time, these two will converge at some point.

But “interactive film-game” may not be a stable new category. Why does it exist? Partly because live-action art style differs from rendered graphics; partly because it has interaction, but not enough to be called a simulation management or RPG game, yet it’s released on Steam, so it gets the label “interactive film-game.”

Looking forward, if its interaction becomes richer, it might be a “story-rich simulation management” or “story-rich RPG.” I think everyone will gradually discover that games have such good gameplay models and monetization methods – it’s a shortcut for those making interactive content. And when you take that shortcut to a certain extent, the problems you encounter will be similar to what we face now.

But interactive content has always been a real human need. As products continue to be produced, it will certainly find its own audience.

“Bu Wen Fan Chen” counts as an in-depth attempt at AI interactive film-games. As a creator, how do you see the future development prospects of such products?

Xin Jie: I am very optimistic about the prospects. AI is still developing rapidly. Many of the difficulties we are discussing now may not be issues six months from now – though new problems may arise then. If the waste rate can be reduced, the entire production and delivery will be faster, and we will have more resources to make better products.

What plans does the team have for new gameplay, new products, and commercialization going forward?

Xin Jie: The most important thing now is to make Bu Wen Fan Chen a great product. We will release another trailer in July, showing gameplay that hasn’t been revealed before, including auctions, sect tournaments, world bosses, etc., to let everyone know that we don’t just have story – we have built a solid, complete loop experience for a cultivation interactive product. For the next product, we may continue with other themes and gameplay modes.

Regarding commercialization, for the first product we hope it is both well-received and commercially successful. But choosing the Steam platform is not purely a commercial decision – it’s for global distribution, to validate the market for a product going from 0 to 1.

After that, there are indeed many more commercial approaches, but that depends on subsequent products. For now, nailing the 0-to-1 step for this user base is the most important thing.

This article is reprinted with permission from “Game Artist Game Art”

​Every technological revolution will give rise to a new wave of application opportunities. The explosion of AI is rewriting the rules of content creation and distribution. It is not only reshaping the video track but also having a profound impact on interactive entertainment, especially games.

The ever-richer AI technologies and tool platforms are gradually lowering the barrier to game development, opening the door for many creators to build interactive experiences in a UGC manner.

Over the past year, products like Aippy, Loopit, Astrocade, Rezona, Sekai, and Yoroll have emerged, each more or less triggering discussions about “playable TikTok” or “the TikTok of the AI gaming world”. These products share a similar core logic: using AI to lower the barrier to entry for game creation so that ordinary people can become creators, and then distributing these interactive contents to users just like short videos.

The capital market has also taken notice. In May this year, Astrocade, co-founded by Fei-Fei Li, completed a $56 million Series A+B financing. Loopit secured three rounds of funding in the first three months of the year, totaling nearly $100 million. Behind Aippy is the listed company Chizicheng Technology, which has a natural accumulation in the overseas social market.

However, although the core logic is similar, the six representative platforms currently exhibit three different forms in terms of specific technical routes and product models.

​Aippy、Loopit：Vertical feed, lightweight interactive mini-games

When you open Aippy, you see a page similar to TikTok: a vertical screen with like, comment, share, and follow buttons. Swipe to the next video, and a parkour game with nearly 20,000 likes appears on screen. You try it for a few seconds, Game Over, then swipe away – the whole process feels very close to browsing short videos.

Besides playing others’ games, Aippy also has a “remix” feature. Users can see a game someone else made well and modify it with one tap – change a theme, tweak a rule, add a line of dialogue – and after editing, it becomes “your” new work. According to the company, about 40% of Aippy content comes from the Remix path, and the quality score of Remix works is higher than that of original creations.

After the content volume grew, a distribution mechanism began to take shape. Aippy’s recommendation system does not simply rely on popularity ranking; instead, it comprehensively considers the quality of the game itself, user preferences, behavior of similar users, and a certain proportion of new content exploration. This design not only prevents top content from monopolizing traffic but also retains exposure opportunities for long-tail creators. Many users, when recommending Aippy to friends, compare it to “TikTok but for games”.

Aippy was initially incubated by Chizicheng Technology and has since become independent. Its founder and CEO, Evan Ye, is also a co-founder of Chizicheng Technology. He started teaching himself programming in middle school and independently developed software and small games. During college, he embarked on the mobile internet overseas entrepreneurship journey together with the other two founders of Chizicheng Technology. In June this year, Aippy completed its first round of financing of tens of millions of dollars, invested by Glowill Capital, with a post-investment valuation of $250 million.

Loopit (Yongyue Intelligence) is the product on this track with the fiercest recent financing momentum. In February this year, Musk commented and retweeted a 9-second video: a cartoon Musk character stood in the center of the screen, with the question “Which is the best LLM in the world?” above, and two options below: “Grok” and “Others”. But the “Grok” button was so small it could barely be clicked, and it would dodge every tap.

This distinctly self-deprecating content quickly went viral, bringing the AI interactive platform behind it, Loopit, into the public eye overnight.

Loopit’s co-founder and CEO, Chen Weipeng, is a former co-founder and head of the large model team at Baichuan Intelligence, where he led the training of models like Baichuan 1-4. Previously, his career path revolved around distribution and the evolution of content forms. Since 2026, Loopit has completed three rounds of financing, totaling nearly $100 million. The latest round was $50 million, led by global top gaming company Garena.

In February this year, Loopit officially launched. Two months later, its global registered users approached 2 million. Similar to Aippy, Loopit also emphasizes an “instant-play” interactive experience. However, the growth logic of the two differs.

Aippy mainly relies on users’ secondary creation during content consumption and “imitation-driven creation” to stimulate viral spread, forming a community self-cycle. Loopit, on the other hand, focuses on producing “shareable interactive content”, achieving breakout via the widespread dissemination of single lightweight gameplay on external platforms like TikTok and X, and then maintaining user retention through frequent updates.

In simple terms, the former emphasizes community depth and creator relationships, while the latter values gameplay virality and content update efficiency.

However, the issue of content homogenization has begun to erode this growth story. In app store reviews, some users have complained, “This game is fun for about 5 minutes, then you realize it’s the same content over and over again.” This is not a predicament unique to Loopit; it is a common challenge across this entire track: the lower the generation barrier, the higher the risk of content homogenization. How to enable creators to continuously produce differentiated content will be a new problem all platforms face.

​Astrocade 、Rezona：AI-native creation engine, UGC content community platform

This category of platforms emphasizes the ability of AI to lower the barrier to creation itself. The platforms provide the ability to generate games using natural language. Users can create a variety of gameplay ranging from light to medium. It is also more like “YouTube for games”, with richer content and higher creative ceilings.

Astrocade is the representative of this direction and currently the most technically prominent product on the track. It is co-founded by “AI godmother” Fei-Fei Li, who also serves as the Chief Scientist. In May this year, Astrocade completed a combined $56 million Series A+B financing, roughly equivalent to 400 million RMB, jointly invested by Sequoia Capital, Google AI Futures Fund, NVIDIA, and others.

Astrocade has proposed a “wish to game” vision: users do not need to write code; they only need to describe their idea in everyday language, and the platform will generate a complete interactive game within minutes, complete with gameplay, interactive controls, and even music and sound effects.

On the Astrocade homepage, game content is categorized into multiple vertical sections: “Strategy”, “Puzzle”, “Horror”, “Simulation”, etc. In addition, the platform has a “Creativity” column, where the games listed are all works with over ten thousand clicks.

In terms of interface design, Astrocade is also very close to TikTok: only one game is displayed at a time in the center of the screen, and users can swipe up and down to switch. Each game is accompanied by buttons for sharing, commenting, liking, saving, and following the creator. This design clearly transplants the interactive logic of short videos into a UGC game community.

Public data shows that within six months of launch, Astrocade accumulated 20 million users, with over 20,000 new UGC game works added each month.

Rezona is currently one of the fastest players on the distribution side. Like Astrocade, it also tends to be a UGC content community platform. However, unlike Astrocade, which focuses on “generating any game with one sentence”, Rezona’s most distinctive difference lies in its content entry point: it specializes in turning internet memes and jokes into playable interactive games, and distributing them via a TikTok-style vertical feed.

Founder Zhang Jian was previously a product manager at TikTok. In his view, memes are “content compression packages”. For example, the once-popular “crazy husky meme” or “Trump meme” – users often only need to see the image or keywords to evoke the corresponding context and emotion. This content form, with high information density and low comprehension barrier, is naturally suited for early-stage propagation and fission in new media.

Rezona’s growth logic unfolds from here: when a meme erupts on social platforms like TikTok, a large number of related interactive games emerge in the Rezona community. After users enjoy a fun game, they actively share it back to social platforms, forming a new round of dissemination.

This tactic has yielded significant results. Public data shows that within less than four months of launch, Rezona’s weekly active users exceeded 1.5 million. By January this year, monthly downloads surpassed one million.

​Sekai、Yoroll：Deep immersion, long-form content / film-game fusion

If the first two categories follow a relatively lightweight interactive route, then the next category seeks more “immersion and sense of presence”. Here, most content consists of AI-driven role-playing, long-form narrative interactions, or branching storylines. Users can enter a story world generated by AI and experience interactive content with greater narrative depth and emotional involvement.

Sekai is the representative of this direction. Its official positioning is “fun interactive experiences”, characterized by immersive role-playing and world co-creation. The platform features a large amount of AI-driven character interactions and long-form storylines.

AI can turn users’ ideas into playable, shareable “mini-apps”. At the same time, the platform provides deep customization tools: from gender, appearance, and clothing to personality traits, backstory, and catchphrases, all customizable. Users can also save complete chat logs and plot developments, and export them as TXT or PDF files, allowing them to revisit key plot points at any time, which is convenient for long-term character development and complex story arc tracking.

Sekai’s founder, Lucky Zhang, is a seasoned serial entrepreneur. Sekai is his fourth company: the first two were acquired by Apple and TikTok respectively, and the third is Vietnam’s largest music streaming platform, still in operation today. This experience of going “from 0 to 1 and then being acquired by giants” has led investors to have high expectations for Sekai. In June this year, Sekai completed a combined $26 million seed and Series A financing.

However, compared to lightweight mini-games and memes with inherent buzz, long-form plots and immersive experiences are obviously less suitable for viral dissemination. The conversion rate of user acquisition and retention is an ongoing problem that immersive content platforms need to solve.

Yoroll, on the other hand, enters the space of AI-native interactive video game platforms. Creators can automatically generate a playable branching interactive film game from text, images, and short videos. According to the official description, the entire story, characters, gameplay, and storyboarding of the game are all generated by AI, making the creative process more like directing a movie.

Yoroll’s founder, Heath, was previously a core member of the social interactive direction at Douyin (TikTok) and later served as Technical Director at FunPlus. He explained the realistic judgment behind this choice: in the short term, it is still difficult for AI to completely replace large-scale 3D game pipelines, but the maturity of AI generating a “watchable” video is already significantly higher than generating a stable 3D game world.

Based on user feedback, Yoroll performs quite well in maintaining visual consistency and demonstrates strong capability across different game styles. The platform currently features themes such as space exploration, post-apocalyptic zombie survival, Eastern mystery puzzles, and all are complete products with full storylines, saveable progress, support for branching choices, and replayability.

A noteworthy case is the gender-bent version of the classic Chinese drama “Empresses in the Palace” made by a UGC creator – “The Legend of Hua Jun”. In this game, palace intrigues and emotional power struggles are replayed under a matriarchal world setting. More importantly, from concept to a shareable game clip, it took less than a week. Currently, this work has gained hundreds of thousands of likes and millions of views on Douyin.

The breakout of “The Legend of Hua Jun” verifies Yoroll’s practical value in lowering the barrier to creation. On this basis, Yoroll has begun recruiting AI creators in the short-video field, and has already signed dozens.

In the AI era, when production costs and complexity drop significantly, the industry is seeing not just faster versions of old games, but an entirely new content category. Such game products may be completed in just one weekend. They build interactive experiences around a single moment, a meme, or a cultural event, and the way they are created and shared is as casual as a TikTok short video.

Looking across these game platforms, the founding teams generally have backgrounds in the short-video track. This experience enables them to systematically transplant the proven traffic distribution logic and content operation methodology from the short-video field into AI interactive content platforms.

Of course, for these platforms to truly become “the TikTok of the AI gaming world”, there are still many hurdles to overcome. For example, most platforms can currently only produce games with simple mechanics; the generated works are generally rough in quality, and frequent failures occur when loading long narrative content. How to leap from “playable” to “fun”, and from “usable” to “reliable”, will determine whether they can truly replicate the TikTok moment.