Manus: The 300 Days Before Meta
Here is our debut story.
By Team elsewhere
Liu Yuan woke up to a standard March morning in 2025. To the ZhenFund partner, nothing seemed out of the ordinary until he unlocked his phone. The bottom of his WeChat interface was a solid blur of red notifications.
The company he backed had just become the center of the internet.
The night before, the founding team—Xiao Hong (”Red”), Ji Yichao (”Peak”), and Zhang Tao (”hidecloud”)—had released Manus, a general AI agent. The demo was already flooding WeChat Channels. But unlike the polished, high-production launches typical of the industry, the Manus debut was starkly lo-fi. It was just Ji Yichao sitting alone in front of a camera, a nondescript sofa and a potted plant in the background, smiling as he walked through the code.
Rumor placed Mark Zuckerberg in front of a screen in Menlo Park shortly after, logging on as one of Manus’s first users.
Nine months later, on December 30, Meta announced it was acquiring Manus for more than $2B. The Chinese internet exploded once again. For China’s venture capital and AI sectors, 2025 was the year defined by Manus.
From launch to exit, the entire lifecycle took nine months.
But the atmosphere surrounding this exit felt distinctly different from 2018, when Meituan acquired Mobike for $2.7 billion. Back then, the mood was heavy with pessimism. It marked the end of a specific narrative for the frantic elite: burn cash, flood the market with volume, and when the unit economics fail to balance, seek shelter inside a tech giant. It was a capitulation.
Xiao Hong and Manus represented a new logic. The story wasn’t just about the multi-billion dollar price tag. It was about a team rooted elsewhere—once dismissed by critics as mere “wrappers” repackaging American APIs—who squeezed every ounce of utility out of the tools at hand to force a different outcome.
This is a trajectory unique to the AI era.
elsewhere spoke with multiple sources close to Manus and their investors to reconstruct the anatomy of this company’s rise.
The young guy
January 2015.
Xiao Hong was still a student at the Huazhong University of Science and Technology when he registered his first company. He was green — barely out of the classroom — when he entered a hackathon to pitch a demo for an editor plugin. He had three minutes on stage.
By pure chance, Liu Yuan, a young investor at ZhenFund, was sitting in the audience.
Xiao Hong happened to be wearing a T-shirt with a graphic resembling a Pink Floyd album cover. For Liu Yuan, an investor known for his bohemian streak, the signal was irresistible.
Xiao Hong had planned to return to Wuhan the next day. Liu Yuan practically held him hostage: “Stay one more day. I’ll book the hotel. We are investing in you. If the other partners disagree, I’ll use my ‘silver bullet’.” In China’s VC landscape, ZhenFund built its reputation on betting on people—specifically, the very young.
That is how the story began.
For a long time after that, Xiao Hong’s “Nightingale Tech” was nothing more than a widget shop. In the early days of WeChat Official Accounts (like Substack), their product, “YiBan,” was a formatting tool for newsletter editors. The company name came from Oscar Wilde’s The Nightingale and the Rose, implying the role of a silent guardian.
But in practical terms, it was a mediocre business. There was no grand vision. They were operating in China’s enterprise software (To B) market—a notoriously difficult sector where investors rarely looked twice.
2021, Wu Minghui, founder of Mininglamp, entered the picture. Wu had deployed 1,000 engineers internally to build a similar product, burning tens of millions in monthly wages. Looking at Nightingale’s growth curve, Wu decided to stop building and start buying.
This was Xiao Hong’s first pot of gold. It was also his first lesson in structural survival: inside the seams of a tech giant’s ecosystem, a plugin often possesses a sharper survival instinct than the native system itself.
The young guy was now a serial entrepreneur.
The First General Agent
Late 2022. Xiao Hong exited Mininglamp, looking for the next variable.
He set his sights on the world’s largest plugin ecosystem: Chrome. While most founders were busy training fundamental models from the ground up, Xiao Hong made a different opening move. He paid out of pocket to acquire an existing plugin named “ChatGPT for Google.”
Instead of renting a glass office in mega cities, the team stayed entrenched in Wuhan. Throughout 2023, while China’s model companies burned cash on compute to rival GPT-4, Xiao Hong focused on a different problem: how to use a browser plugin to intercept AI conversation traffic on a global scale.
Top-tier VCs began to circle. The consensus was polarized. Some found Monica indispensable; others scoffed: “It’s just a wrapper company.” One due diligence report was particularly blunt: “Strong execution, but the founder lacks vision—he refuses to leave Wuhan.”
Fundraising was a grind. Then, in early 2024, ByteDance’s strategic investment team made contact.
By then, Monica had broken into the top tier of the global plugin market. ByteDance put an offer on the table: $30 million for a full acquisition and acqui-hire.
The irony was sharp. An investor who had previously rejected Xiao Hong once advised him, “Have you tried Doubao? You should learn from them.” Doubao was ByteDance’s own AI product, a household name in China.
Xiao Hong rejected the ByteDance’s offer.
His fundamental belief was simple: Monica should not be a “skin” for any large model. Through an introduction by ZhenFund, he began meeting frequently with Ji Yichao. Ji was a legend in his own right—a wunderkind who dropped out at 17 to build Mammoth Browser. He had just wrapped up a tenure at 4Paradigm.
Dai Yusen, a partner at ZhenFund, brokered the critical merger. He also brought in Zhang Tao, a veteran of Lark, and LightYears.AI Beyond. The viral product demo that would later define their marketing style was originally posted on Zhang Tao’s personal blog.
It was a convergence of instincts. Xiao Hong brought the commercial killer instinct for “intercepting” traffic; Ji Yichao brought the technical depth regarding Agents. They stopped discussing how to make AI talk. They started discussing how to make AI control a browser
Shortly after, a project codenamed Manus was born.
Zero to $100M ARR
March 6, 2025. Manus goes live. The name is derived from the Ancient Greek word for “hand.”
Unlike Monica’s strategy of silent interception, Manus was built to bypass the search bar entirely, using agents to execute tasks directly.
Months later, during a meeting with Manus, a partner from the legendary Silicon Valley firm Benchmark Capital posed a question: “You have the potential to become larger than the foundation model companies themselves. But today you rely on them. How do you plan to break your chains?”
For Manus team, the question itself was the ultimate compliment.
By December 2025, Zhang Tao announced on social media that Manus had hit $100 million in Annual Recurring Revenue (ARR). It was a milestone.
There is no doubt: Xiao Hong has become the figurehead for a new generation of Chinese global founders. And yet, for most of the world, he has only existed for nine short months.
The Hard Way
2021, Wujiaochang, Shanghai.
Xiao Hong sat on the curb, crying.
He had just walked out of an investment committee meeting with Hongshan. He didn’t know how to wrap his product — WeiBan —in the glossy narratives that top-tier VCs crave. The meeting was over. The deal was dead.
The weight wasn’t financial; it was the sting of unrecognized identity. Xiao Hong was a pragmatist, a builder who had been monetizing software since his university days. But to him, the evaluation criteria of the elite capital class felt like a wall he could not scale.
At that exact moment, a colleague ran up to him with news: two major clients had just wired payments directly to the bank account.
Years later, Xiao Hong would tell this story as a joke over drinks. But at the time, that fundraising awkwardness was a cage he couldn’t escape.
Liu Yuan, his most critical backer, told elsewhere that he exhausted every contact he had. He created over 130 groups, pitching Xiao Hong to essentially every venture capitalist in China.
The process revealed a panorama of absurdity.
One firm slashed the valuation from $50 million to $30 million, citing the “low ceiling of the plugin market.” When Xiao Hong used raw growth data to prove them wrong, they grudgingly nudged the offer up to $40 million. Xiao Hong conceded to $45 million. The firm dug in their heels at $42 million.
Another investor, after meeting Xiao Hong, pulled Liu Yuan aside: “What’s the catch with this company?” Liu Yuan was helpless. It was nitpicking for the sake of nitpicking.
This gridlock lasted until late 2024. On the eve of Monica’s transformation into Manus, Hongshan and Tencent finally wired the funds.
It is a peculiar historical irony: the two most prestigious funds in China eventually boarded the ship, but only just before it left the harbor.
A Decade
December 30, just before 7:00 AM there.
When a Chinese business outlet broke the news that Meta had acquired Manus, social media detonated.
It felt like the definitive annotation of the era: an obscure company that began the year in silence had ended it by rewriting the headline.
Rewind to 2018. When Meituan acquired Mobike, the $2.7 billion price tag concealed a grim reality: mountains of abandoned bicycles clogging the streets and maintenance costs that defied calculation. The ecosystem back then was addicted to “scale effects.” Founders were essentially just employees of their capital backers. Their exits didn’t feel like victories; they felt like exhaustion — the byproduct of massive resource incineration.
Xiao Hong and Manus offered a different narrative possibility.
From Nightingale to Monica, and finally to Manus, Xiao Hong kept his asset structure radically light. With a skeleton crew, he leveraged the APIs of mature platforms—WeChat, Chrome, LLMs—to accumulate users and understand their intent.
The euphoria sweeping the industry stems from a fundamental shift: the script for the new generation of Chinese global founders has flipped.
They are building products with a truly global aesthetic, using smaller teams and moving with greater velocity. Their definition of “entrepreneurship” has evolved, too. They are no longer the “lonely warriors” obsessed with total domination or death. They are pragmatists, willing to make rational, strategic choices when the window of opportunity opens.
“Logic leads to despair,” Xiao Hong once said.
If he had followed conventional logic, he would have folded in 2015. He would have pivoted when Sequoia rejected him in 2020. He would have sold it when ByteDance made their offer in 2024. Every logical indicator screamed that a plugin developer rooted elsewhere could not possibly carve out the largest slice of the AI era.
But he did.