Groundbreaking or Overhyped? Manus AI Sparks Debate in the Tech World
Image Source: Manus | Youtube
In a development that has sparked widespread debate in the artificial intelligence community, Chinese researchers have unveiled what they claim is the world’s first fully autonomous AI agent, Manus. Launched last week by Monica, an AI startup linked to the Singapore-registered Butterfly Effect PTE. LTD., the company is reported to primarily operate from China, with teams based in Beijing and Wuhan.
While some experts praise Manus as a breakthrough in AI autonomy, others question its true capabilities and warn of potential privacy risks.
[Read More: Why Did China Ban Western AI Chatbots? The Rise of Its Own AI Models]
A New Breed of AI: Autonomy in Action
Unlike conventional chatbots that rely on a series of human prompts to function, Manus is designed to take a single instruction and run with it, completing intricate tasks independently. According to Monica, the AI can sift through résumés, rank job candidates, and even produce formatted spreadsheets in mere seconds—an efficiency showcased in its launch video. Beyond clerical work, the company asserts that Manus can analyze stock market trends, pull data from the internet, and construct fully functional websites from scratch. Its cloud-based operation allows it to continue processing tasks even after users disconnect, a feature that sets it apart from many existing tools.
Yichao “Peak” Ji, co-founder of Manus, emphasized this distinction in the unveiling video last Wednesday. “This isn't just another chatbot or workflow tool”, Ji stated. “It's a completely autonomous agent that bridges the gap between conception and execution”. The startup also claims that Manus surpasses OpenAI’s Deep Research model on the GAIA benchmark, a widely recognized metric for evaluating AI assistants. However, access remains restricted to an invite-only testing phase, leaving many eager to verify these bold assertions.
[Read More: ChatGPT Deep Research vs Grok 3 DeepSearch: Which AI Wins?]
Under the Hood: Built on Existing Foundations
Speculation about Manus’s origins surfaced almost immediately after its debut. Some early testers, including Pierre-Carl Langlais, co-founder of AI startup Pleias, suggested that it might not be an entirely original creation. In a LinkedIn post on Sunday, Langlais pointed out that Manus employs token conventions unique to Anthropic’s Claude 3.5 Sonnet, hinting that it could be more of a sophisticated overlay than a standalone innovation.
Ji addressed these claims in a Monday post on X, confirming that Manus integrates Claude 3.5 Sonnet v1 alongside fine-tuned versions of Alibaba’s Qwen models.
“We use Claude and different Qwen-finetunes. Back when we started building Manus, we only got Claude 3.5 Sonnet v1 …… so we need a lot of auxiliary models”, he explained. “Now Claude 3.7 looks really promising, we are testing internally, will post updates!”
This multi-model approach, Ji suggests, is central to Manus’s ability to juggle complex tasks, though it has fuelled debate over how much of its technology is truly novel.
[Read More: Agentic AI in 2025: The Rise of Autonomous AI Agents by OpenAI, Microsoft and Nvidia]
Echoes of DeepSeek: A New Milestone for China?
The launch of Manus has inevitably drawn parallels to DeepSeek, another Chinese AI breakthrough that stunned the industry in January. DeepSeek’s model, touted for matching Western competitors at a fraction of the cost, triggered a US$600 billion drop in Nvidia’s stock and earned the label of China’s “Sputnik moment” in AI. Manus, with its promise of full autonomy, is now being positioned as a potential successor in this narrative of Chinese innovation outpacing U.S. restrictions on advanced chip access.
AI policy researcher Dean Ball weighed in on X, arguing that Manus might exceed DeepSeek’s significance.
“DeepSeek was about replicating capabilities already achieved by American firms”, Ball wrote last Sunday. “Manus is actually advancing the frontier. The most sophisticated computer using AI now comes from a Chinese startup, full stop”.
Yet, Manus’s dependence on existing models tempers claims of a similar seismic shift.
[Read More: Signal President Warns of Agentic AI Privacy Risks at SXSW 2025]
Mixed Reviews: Breakthrough or Bust?
The AI community’s response to Manus has been a study in contrasts. Victor Mustar, head of product at Hugging Face, offered glowing praise, calling it “the most impressive AI tool I've ever tried” in an X post. He lauded its agentic capabilities, suggesting they could render traditional coding obsolete. “This could kill vibe coding”, Mustar mused, envisioning a future where ideas alone drive development.
On the flip side, skepticism abounds. TechCrunch reporter Kyle Wiggers and Pleias co-founder Alexander Doria reported disappointing results from their tests, citing factual inaccuracies, execution hiccups, and instances where Manus became trapped in endless loops. Langlais, despite acknowledging its user-friendly interface, criticized what he called “deceptive communication” and “hunger marketing”—tactics that artificially amplify hype by limiting access to a select few.
“What the AI market really needs now is better standards of openness and transparency. At every level: model, data, business”, he urged on LinkedIn.
[Read More: DeepSeek AI Among the Least Reliable Chatbots in Fact-Checking, Audit Reveals]
Privacy Shadows Loom Large
Beyond its technical merits, Manus has stirred unease over data privacy—a concern it shares with DeepSeek. AI researcher Luiza Jarovsky questioned the security of user information in a Substack post, asking,
“Where are the servers located? Is user data transferred to China?”
With no response from Monica and Butterfly Effect’s ownership opaque, speculation persists about Chinese jurisdiction over user data. Corpora.ai CEO Mel Morris noted this could overshadow Manus’s promise, echoing concerns tied to DeepSeek.
[Read More: EU Blocks Chinese AI App DeepSeek Over GDPR Compliance Concerns]
A Cultural Coincidence? The "Monica" Connection
The startup’s name, Monica, has sparked curiosity beyond its tech claims, echoing Leslie Cheung’s 1984 Cantopop anthem “Monica”—a defining hit in Hong Kong’s music history. While no evidence ties the AI firm to Cheung’s legacy, the shared name with a cultural icon from the region could invite speculation, especially given Butterfly Effect’s Chinese ties. This overlap might confuse some viewers or hint at an intentional nod.
[Read More: Does DeepSeek Track Your Keyboard Input? A Serious Privacy Concern]
Source: Business Insider
