As tech giants muscle in, companies like Lingzhu and Aippy are betting that social, remix-driven communities will give them the edge in the next generation of software.
By Da Cheung
One of the fastest-growing frontiers in artificial intelligence is software generation. A new crop of startups is betting that “vibe coding,” a process where users build software simply by describing what they want in natural language, will democratize app creation and fundamentally change how consumers interact with technology.
On July 2, Chinese startup Lingzhu AI announced it secured angel funding led by Wei Haijun, a prominent investor known for backing Musical.ly, the predecessor to ByteDance’s TikTok. Lingzhu is positioning itself as a “zero-threshold” platform, aiming to turn hundreds of millions of Chinese non-coders into software creators.
The company claims that since launching its closed beta in April, the platform has surpassed 5 billion daily token usages — a metric measuring the volume of data processed by the AI — and that 83.7% of users can build their first application within five minutes.
In a hands-on editorial test of Lingzhu’s platform by The Insight Asia, we asked Lingzhu to create a meme animation generator, which is a very complex application. It took four rounds of revisions over 20 minutes and the resulting tool offered marginal practical utility.
Yet, the platform’s “modding” feature revealed a much more compelling use case. By typing a single sentence, we successfully modified an existing, user-created mini-game into a customized, highly engaging version. Building on top of someone else’s successful foundation was significantly easier and more entertaining than starting from a blank prompt.
The remixing engine
This dynamic — where modifying existing content is more effective than original creation — is shaping the strategy of the entire sector. Startups are realizing that their true value lies not in the raw coding capabilities of AI, but in the communities they can build around shared, modifiable content.
Aippy, an AI game creation platform and social community, launched in mid-2025, has seen steep daily active user growth in core English-speaking markets like the U.S., U.K., and Canada. The platform, founded and incubated by Hong Kong-listed tech giant Newborn Town (9911.HK), operates much like a vertical video feed. But instead of watching videos, users swipe through playable, AI-generated games.
According to Aippy, roughly 40% of its content comes from users remixing existing games which consistently score higher in quality ratings than original creations, and users who engage in remixing show significantly higher retention rates.
Similarly, Lovable, a Swedish startup founded in 2023, has grown rapidly by allowing users to build websites and apps through chat interfaces. Valued at $6.6 billion after raising $540 million, Lovable is frequently cited as a major European AI success story.
Surviving the tech giants
Despite massive valuations and growing user bases, these application-layer startups face a critical vulnerability: they own relatively little of the core AI technology underpinning their products.
Companies like Lingzhu AI, Aippy, and Lovable do not build their own foundational large language models (LLMs) — the underlying AI engines that power their platforms. Instead, they rely on APIs — software interfaces that allow them to access LLMs built by major tech giants — rather than developing their own. Lovable, for instance, routes its prompts through models made by Anthropic and Google. And it recently announced a deepened partnership to integrate its products into Google Workspace. Lingzhu similarly relies on multiple domestic Chinese models to deliver its finished products.
This reliance carries what some investors see as an existential risk: if foundation-model providers continue moving aggressively into consumer products, they could eventually render parts of the application layer redundant. For the past few years, major AI developers have focused heavily on foundational research and model upgrades. Recently, however, companies including Google, Anthropic, and OpenAI have begun releasing highly refined, consumer-facing products.
In April, images surfaced on X suggesting Anthropic had created a new in-chat app builder that allows users to generate applications from simple prompts, effectively competing directly with Lovable’s core product. Tech Policy notes that this pattern is known in the tech industry as getting “sherlocked” — a term originating from Apple’s practice of incorporating features pioneered by third-party developers into its own software, often undermining the market for the original products.
Because the basic service of vibe coding is relatively easy for large tech companies to replicate, startups and their investors are racing against the clock. Their survival depends entirely on building a vibrant, sticky community before the tech giants fully pivot to consumer applications. They are not just trying to build coding tools; they are trying to build the TikTok of the AI era.
The end of the app store?
If these community-driven platforms succeed in changing consumer habits, the implications for the software industry are profound.
Many industry experts believe that the era of the fixed, one-size-fits-all application may eventually come to an end. In its place, users could rely on AI to generate and instantly modify hyper-personalized apps tailored to their exact needs in real time.
It’s too early to tell whether this model will become mainstream, but companies like Lingzhu have become a testing ground. And so far, their early success suggests that while AI can write the code, it is the human desire to play, share, and remix that will ultimately drive the next generation of software.
Sources
