Manus has sparked a boom in AI agents in China

China saw a boom last year in foundation models – the large language models which do everything and are the basis of the AI revolution. This year the focus has shifted from AI agents to AI agents that can do things on their own, rather than just responding to user queries.

A number of Chinese startups are building these general-purpose digital devices, which can answer email, browse the web to plan vacations, or even design an interactive site. Many of these tools have been developed in the last two months. They follow in the footsteps Manus, a general AI agent which sparked a social media frenzy to get invite codes following its limited release launch in early march.

The emerging AI agents aren’t large language models. They’re built over them, with a workflow-based system designed to get things accomplished. Many of these systems introduce a new way to interact with AI. Instead of just chatting with users, these systems are optimized to manage and execute multi-step tasks – booking flights, managing schedules and conducting research – by using external tools and memorizing instructions.

China may be the first to build these types of agents. The tightly integrated app ecosystems in China, rapid product cycles and digitally literate users could create a favorable environment to embed AI into everyday life.

At the moment, China’s leading AI agent startups focus on the global market because the best Western models do not work inside China’s firewalls. But that could soon change: Tech giants such as ByteDance, Tencent and Alibaba are developing their own AI agents to integrate automation directly into native super-apps. They will pull data from their vast ecosystems of programs that dominate daily life in China.

As a race to define the ideal AI agent unfolds, a mixture of ambitious startups and established tech giants is now testing how these tools could actually work in practice — and for whom. Set the standard

The startup Butterfly Effect has had a busy few months with Manus. The company raised $75M in a funding round, led by US venture capital firm Benchmark. It also took the product on a global roadshow and hired dozens new employees.

Before registration opened to the general public in May, Manus was already a benchmark for what a consumer-oriented AI agent could accomplish. This “general” agent, rather than handling narrow tasks for businesses, is designed to help with everyday chores like trip planning or stock comparisons, or even your child’s school project.

Manus, unlike previous AI agents, uses a browser based sandbox to let users supervise the agent in real time, watching as it scrolls down web pages, reads an article, or codes actions. It also asks clarifying queries and supports long-term memories that would serve context for future tasks.

“Manus represents a promising product experience for AI agents,” Ang Li, CEO and cofounder of Simular in Palo Alto California, says that the startup is building AI agents which operate a real computer. “I think Chinese startups have an advantage when it comes designing consumer products. Thanks to fierce domestic competition, they are able to execute quickly and pay more attention to the details.”

The competition in the case of Manus is moving quickly. Genspark and Flowith are two of the most talked-about follow-ups. They have benchmark scores that are comparable to or even better than Manus.

Genspark, led by former Baidu executives Eric Jing, and Kay Zhu, connects many small “super-agents” through what they call multi-component prompting. The agent can switch between several large language models and accepts images and text. It can also perform tasks such as making slide decks and placing phone calls. Genspark integrates directly with a variety of tools and APIs, while Manus heavily relies on Browser Use, an open-source software that allows agents to operate a web page in a virtual window, just like a real person. The company, which was launched in April, claims to have over 5 million users with over $36 million a year in revenue.

Flowith, the work by a young team, first caught the public’s attention in April 2025, at a developer conference hosted by popular social media app Xiaohongshu. It is marketed as an “infinite assistant” and opens with a blank canvas, where each question is a node in a branching map. Users can go back, take new paths, and store the results in “knowledge gardens,” which are personal or shareable. The design is more like a project management tool (think Notion), than a typical chat interface. Each inquiry or task creates its own mind-map graph, encouraging a nonlinear and more creative interaction with AI. Flowith’s core agent, NEO, runs in the cloud and can perform scheduled tasks like sending emails and compiling files. The founders want the app to be a “knowledge marketbase”and aims to tap into the social aspect of AI with the aspiration of becoming “the OnlyFans of AI knowledge creators”.

What they also share with Manus is the global ambition. Both Genspark and Flowith have stated that their primary focus is the international market.

A global address

Startups like Manus, Genspark, and Flowith–though founded by Chinese entrepreneurs–could blend seamlessly into the global tech scene and compete effectively abroad. Founders, investors, and analysts that MIT Technology Review has spoken to believe Chinese companies are moving fast, executing well, and quickly coming up with new products.

Money reinforces the pull to launch overseas. Customers there pay more, and there are plenty to go around. “You can price in USD, and with the exchange rate that’s a sevenfold multiplier,” Manus cofounder Xiao Hong quipped on a podcast. “Even if we’re only operating at 10% power because of cultural differences overseas, we’ll still make more than in China.”

But creating the same functionality in China is a challenge. Major US AI companies including OpenAI and Anthropic have opted out of mainland China because of geopolitical risks and challenges with regulatory compliance. Their absence initially created a black marketas users resorted to VPNs and third-party mirrors to access tools like ChatGPT and Claude. That vacuum has since been filled by a new wave of Chinese chatbots–DeepSeek, Doubao, Kimi–but the appetite for foreign models hasn’t gone away.

Manus, for example, uses Anthropic’s Claude Sonnet–widely considered the top model for agentic tasks. Manus cofounder Zhang Tao has repeatedly praised Claude’s ability to juggle tools, remember contexts, and hold multi-round conversations–all crucial for turning chatty software into an effective executive assistant.

But the company’s use of Sonnet has made its agent functionally unusable inside China without a VPN. If you open Manus from a mainland IP address, you’ll see a notice explaining that the team is “working on integrating Qwen’s model,” a special local version that is built on top of Alibaba’s open-source model.

An engineer overseeing ByteDance’s work on developing an agent, who spoke to MIT Technology Review anonymously to avoid sanction, said that the absence of Claude Sonnet models “limits everything we do in China.” DeepSeek’s open models, he added, still hallucinate too often and lack training on real-world workflows. Developers we spoke with rank Alibaba’s Qwen series as the best domestic alternative, yet most say that switching to Qwen knocks performance down a notch.

Jiaxin Pei, a postdoctoral researcher at Stanford’s Institute for Human-Centered AI, thinks that gap will close: “Building agentic capabilities in base LLMs has become a key focus for many LLM builders, and once people realize the value of this, it will only be a matter of time.”

For now, Manus is doubling down on audiences it can already serve. In a written response, the company said its “primary focus is overseas expansion,” noting that new offices in San Francisco, Singapore, and Tokyo have opened in the past month.

A super-app approach

Although the concept of AI agents is still relatively new, the consumer-facing AI app market in China is already crowded with major tech players. DeepSeek remains the most widely used, while ByteDance’s Doubao and Moonshot’s Kimi have also become household names. However, most of these apps are still optimized for chat and entertainment rather than task execution. This gap in the local market has pushed China’s big tech firms to roll out their own user-facing agents, though early versions remain uneven in quality and rough around the edges.

ByteDance is testing Coze Space, an AI agent based on its own Doubao model family that lets users toggle between “plan” and “execute” modes, so they can either directly guide the agent’s actions or step back and watch it work autonomously. It connects up to 14 popular apps, including GitHub, Notion, and the company’s own Lark office suite. Early reviews saythe tool can feel clunky and has a high failure rate, but it clearly aims to match what Manus offers.

Meanwhile, Zhipu AI has released a free agent called AutoGLM Rumination, built on its proprietary ChatGLM models. Shanghai-based Minimax has launched Minimax Agent. Both products look almost identical to Manus and demo basic tasks such as building a simple website, planning a trip, making a small Flash game, or running quick data analysis.

Despite the limited usability of most general AI agents launched within China, big companies have plans to change that. During a May 15 earnings call, Tencent president Liu Zhiping teased an agent that would weave automation directly into China’s most ubiquitous app, WeChat.

Considered the original super-app, WeChat already handles messaging, mobile payments, news, and millions of mini-programs that act like embedded apps. These programs give Tencent, its developer, access to data from millions of services that pervade everyday life in China, an advantage most competitors can only envy.

Historically, China’s consumer internet has splintered into competing walled gardens–share a Taobao link in WeChat and it resolves as plaintext, not a preview card. Unlike the more interoperable Western internet, China’s tech giants have long resisted integration with one another, choosing to wage platform war at the expense of a seamless user experience.

But the use of mini-programs has given WeChat unprecedented reach across services that once resisted interoperability, from gym bookings to grocery orders. An agent able to roam that ecosystem could bypass the integration headaches dogging independent startups.

Alibaba, the e-commerce giant behind the Qwen model series, has been a front-runner in China’s AI race but has been slower to release consumer-facing products. Even though Qwen was the most downloaded open-source model on Hugging Face in 2024, it didn’t power a dedicated chatbot app until early 2025. In March, Alibaba rebranded its cloud storage and search app Quark into an all-in-one AI search tool. By June, Quark had introduced DeepResearch–a new mode that marks its most agent-like effort to date.

ByteDance and Alibaba did not reply to MIT Technology Review‘s request for comments.

“Historically, Chinese tech products tend to pursue the all-in-one, super-app approach, and the latest Chinese AI agents reflect just that,” says Li of Simular, who previously worked at Google DeepMind on lifelong learning agents. “In contrast, AI agents in the US are more focused on serving specific verticals.”

Pei, the researcher at Stanford, says that existing tech giants could have a huge advantage in bringing the vision of general AI agents to life–especially those with built-in integration across services. “The customer-facing AI agent market is still very early, with tons of problems like authentication and liability,” he says. “But companies that already operate across a wide range of services have a natural advantage in deploying agents at scale.”