Loading page…
The benchmark for Chinese teams going global: started as a browser extension, leveled up into a phenomenon-level AI agent
$75M round led by Benchmark
The single most relevant takeaway for you: a browser extension is the repeatedly proven opening move for Chinese teams going global — distribution rides the store, no ad spend required.
Fewer bars = easier, cheaper, or faster for an AI-assisted solo builder. Editorial judgments based on the case details.
Butterfly Effect (a Chinese team) started with Monica, an AI browser extension that built up a massive user base. In early 2026 they launched Manus — the world's first phenomenon-level AI agent product. Invite codes were nearly impossible to get, the waitlist topped 2.6 million people, and the company landed a $75M round led by famed Silicon Valley VC Benchmark at a $500M valuation.
The opening move was a browser extension, and that choice is the whole answer to distribution. Monica listed on the Chrome Web Store, which hands a team in China a global catalogue, store search, install counts and public ratings on day one — with no ad budget, no local sales office and no distribution partner, which are the three things that normally make a Western launch expensive. An extension also lands where the work already happens: it appears inside the page the user is already on, so there is no new habit to build and no destination to remember. That is why the extension keeps being the first step for teams selling outward from China rather than a fallback. By the time Manus launched in early 2026, Monica's install base meant the new product was announced to people who already had the company's software running in their browser. Manus then rationed invitation codes so tightly that a code was genuinely hard to get, and the waitlist reportedly passed 2.6 million people.
Two products, two different ways money arrives. An extension of Monica's kind earns as a consumer freemium subscription: the free tier is the acquisition channel, and the upgrade trigger is a usage ceiling that a daily user reaches within a week or two, which is why store distribution and revenue are one motion — the install is the trial. An agent is a different animal on the cost side. Each task runs a model for minutes or hours rather than seconds, so a flat monthly fee prices the heaviest users into a loss immediately. The structure that works at this layer is metered: credits bundled into a plan, consumed per task, topped up when someone wants more, so the bill follows the compute instead of the calendar. That is also why an invite-gated launch is more than marketing. Rationing seats holds compute spend inside a number the company chose, during the exact period when the queue is demonstrating that demand far exceeds what can be served.
Copy its path (extension → platform), not its endgame.
Paste into Claude Code / Codex and get a working version of this product end-to-end.
Two production-grade prompts per idea — a full build spec and an SEO growth plan — plus 10 tailored build specs a week. Members only.
Join the Unicorn Club — $5/moCancel anytime, in one clickEvery claim traces back to a listed source — open any of them and check the number yourself.
Data credibility: Public reporting / video roundup. All revenue figures are founder-reported or estimated by the video creator, unaudited.
The flywheel runs in two stages, and the first is the transferable one. An extension with a large install base is not only revenue; it is an owned audience that a second, bigger product can be launched into, which is precisely what the move from Monica to Manus is. The second stage is scarcity. Invite codes held back hard enough that a code was hard to obtain turn every new user into somebody holding a thing other people want, so codes get posted, traded and argued over, and that conversation is the advertising. A line reported at more than 2.6 million people is a demand signal no campaign buys, and it compounds: the longer the queue, the more worth joining it appears. That signal then funds the next lap. A $75 million round led by Benchmark at a $500 million valuation buys iteration speed, and at the agent layer iteration speed is the competition itself rather than a way of winning it.
The extension route is partly a way of never sitting in the dark. A Chrome Web Store listing reports installs, uninstalls, ratings and written reviews from the first week, all from users nobody paid for, so the question is never whether anyone cares — it is which slice of the install base still has the extension running seven days later. That is why building the extension first and levelling up into a platform afterwards is the safer order for a small team selling into a market it does not live in: the cheap product generates the evidence the expensive one needs. The harder silence sits at the agent layer, where the contest is funding and iteration speed. There the gap between shipped releases is the dangerous stretch, because a queue of 2.6 million people is attention borrowed against a product that still has to arrive, and a hype curve that peaks before the product is ready is the one failure a waitlist cannot repair.
Two things worth copying: (1) A browser extension is the repeatedly proven opening move for Chinese teams going global — distribution rides the Chrome Web Store, no ad budget needed. Your "extension-first" instinct is right. (2) Invite-code scarcity marketing: manufacture scarcity to fuel word of mouth — 2.6 million people in line is the best advertising you could buy. The endgame (an agent platform) is a war of funding and iteration speed. Copy the path, not the endgame.
That was one of the 50 breakdowns open to everyone. The Unicorn Club opens the other 1,285 at the same depth — plus pain-point filters and 10 build specs a week.
Join the Unicorn Club — $5/moCancel anytime, in one click