Be part of our day by day and weekly newsletters for the most recent updates and unique content material on industry-leading AI protection. Study Extra
Cease me when you’ve heard this one earlier than: A little bit-known Chinese language startup is making waves globally for a formidable new AI product.
No, we’re not speaking about DeepSeek R1, the AI reasoning mannequin that made waves amongst Western AI circles earlier this yr. As a substitute, the new new product du jour is named Manus, and it’s a brand new AI multipurpose agent — that’s, greater than an AI mannequin, it’s an interface for controlling a number of fashions that may go off and, by itself, autonomously, full difficult duties like producing experiences or working dozens of social media accounts on the person’s behalf.
If it sounds much like the Deep Analysis modes supplied by OpenAI, Google, and others, in addition to OpenAI’s Operator agent and Anthropic’s Pc Use mode (the latter two of which may, like Manus, take management of a person’s laptop or applications on it, shifting the cursors and typing to carry out actions inside software program), then congrats — you’ve understood what it goals to supply.
However for action-oriented leaders and resolution makers inside enterprises right here within the West and overseas — comparable to chief expertise officers, product managers, IT group leaders, and extra — what do they should find out about Manus and the capabilities it affords? Learn on to search out out.
What’s Manus and who’s behind it?
Manus AI was formally introduced on March 5, 2025 on the social community X with a submit from the corporate describing it as “the first general AI agent” that autonomously executes advanced duties slightly than simply producing concepts.
In response to a South China Morning Submit (SCMP) report, Manus was developed by a Chinese language firm known as Butterfly Impact, which has workplaces in Beijing and Wuhan. The corporate reportedly has just a few dozen staff however has quickly gained consideration in China’s AI panorama.
The founding group contains entrepreneurs and skilled product managers, led by Xiao Hong, a 33-year-old serial entrepreneur and 2015 graduate of Wuhan’s Huazhong College of Science and Expertise.
Xiao beforehand constructed WeChat-based functions that have been acquired by bigger firms and later launched Monica.ai, an AI assistant out there as a browser extension and cell app.
On its web site, Manus is claimed to be named for the Latin phrase for “hand,” a nod to the truth that customers can depend on it to carry out actions for them, or, in my phrases, to “lend them a hand.”
How does Manus AI work?
Manus AI is designed as a multi-agent system, which means it combines a number of AI fashions to deal with duties independently.
Not like AI chatbots that help customers by offering info, Manus can analysis, analyze knowledge, generate experiences, automate workflows, and even write and deploy code.
In response to X posts by Ji Yichao, co-founder and chief scientist of Manus AI, the system is constructed on Anthropic’s Claude 3.5 Sonnet — a nine-month-old AI mannequin at this level — and fine-tuned variations of Alibaba’s Qwen fashions.
The group is presently testing upgrading Manus to Anthropic’s latest and most performant mannequin, Claude 3.7, which is predicted to additional improve its reasoning and execution capabilities.
Manus AI operates asynchronously, which means customers can assign duties and stroll away whereas it completes them autonomously. It’s presently in personal beta, with entry granted by way of invitation codes.
How does Manus AI stack up in opposition to the U.S.-based competitors?
One of many greatest causes Manus AI has gained traction is its sturdy benchmark efficiency, beating U.S. agency OpenAI’s personal o3-powered Deep Analysis agent and the “previous state-of-the-art” in accordance with a graph posted on the official Manus web site.
This declare, together with real-world checks, has led some AI energy customers and early adopters to the conclusion that Manus could also be some of the succesful autonomous AI brokers out there in the present day.
Past benchmarks, Manus has already confirmed itself on freelance platforms like Upwork and Fiverr and in Kaggle machine studying challenges, efficiently executing advanced real-world duties.
AI influencers have a good time Manus’s arrival and spectacular efficiency
Dialog about Manus in media and AI circles took off late final week when customers on the social community X famous that some individuals have been utilizing it to automate the administration of as much as 50 social accounts at one time, in realtime, displaying off its capabilities to create fleets of engagement that companies might use for opinions.
As well as, although this hasn’t but been confirmed for Manus, the identical tech might presumably used for all types of selling and affect campaigns, even political propaganda or disinformation.
However for probably the most half, AI energy customers and influencers within the West have been largely impressed and celebrated Manus’s arrival — awed by their preliminary checks of it as soon as they obtained a number of the scarce beta invitations, or noticed the work of others with entry to the software.
Rowan Cheung, founding father of The Rundown AI e-newsletter, described Manus AI’s launch as a possible turning level for AI brokers and stated “China’s second DeepSeek moment is here,” in a submit on his LinkedIn account.
“This AI agent called ‘Manus’ is going crazy viral in China right now… It’s like Deep Research + Operator + Claude Computer combined, and it’s REALLY good.”
Cheung personally examined Manus and located that it:
- Created and deployed a biography web site about himself, with one hundred pc accuracy and real-time knowledge retrieval.
- Discovered prime rental spots in San Francisco based mostly on crime charges, AI {industry} presence, and entrepreneurship density.
- Developed a full AI course, producing eight chapters of content material, together with instruments, use instances, and prompts.
He obtained 500 invite codes from the Manus group and has been dolling them out to his subscribers and readers.
Former Googler and AI-focused YouTuber Bilawal Sidhu shared a hands-on video overview, calling Manus “the closest thing I have seen to an autonomous AI agent.”
“It’s like you’re standing over the shoulder of somebody using a computer… asking them what to do at the highest level, and it basically does it for you.”
Sidhu examined Manus on varied duties, together with:
- Researching places—scanning Google Maps and information sources to advocate one of the best locations based mostly on laws, accessibility, and security.
- Creating video functions—automating video results comparable to ASCII artwork and real-time filters.
- Extracting {industry} insights—pulling knowledge from Reddit, Twitter, and different sources to generate complete experiences.
Sidhu emphasised that Manus breaks down advanced workflows and executes them step-by-step, making it some of the autonomous AI brokers out there in the present day.
One other AI influencer, Chubby (@kimmonismus on X), claimed that Manus AI outperformed OpenAI’s Deep Analysis: “I do not get paid for promotion. I just tried preview access, and it outperformed OpenAI’s Deep Research by a lot! Mind-blowing.”
Andrew Wilkinson, co-founder of the startup exit supplier and acquirer Tiny, wrote on X that “It’s absolutely insane. I feel like I just time travelled six months into the future. I threw it a zip file of 20 applicants for a CEO job and it did a deep dive on each, one by one…”
Massive questions on Manus’s future stay
Regardless of the thrill, questions and issues have emerged about Manus AI’s true capabilities.
In response to the South China Morning Submit, only a few individuals have been in a position to take a look at Manus AI due to extreme server shortages following its sudden rise to fame.
Zhang Tao, Manus AI’s product companion, admitted in a social media submit: “The current invite-only mechanism is due to genuinely limited server capacity at this stage… The team underestimated the enthusiasm of the public response, and our server resources were only planned for a demonstration level.”
This lack of availability has fueled hypothesis that Manus AI is likely to be utilizing shortage advertising techniques to generate hype. Invitation codes for Manus have been reportedly being offered on China’s second-hand market Xianyu, resulting in frustration amongst potential customers.
As well as, customers comparable to @teortaxesTex on X (the self-described “stan #1”, or obsessive fan, of DeepSeek) opined that “it’s a product devilishly optimized for influencers, which is why it exploded so much. Generating threadboy content, trip plans and such general interest [mind blown emoji, hand pointing down emoji] stuff – yah. STEM assistance, coding – worse than googling. More LLM than agent.”
In different phrases, some consider it’s good for influencers trying to construct up a following by way of spammy lists and suggestions posts, however not so good as different instruments at work like improvement. Even former VentureBeat author and present TechCrunch-er Kyle Wiggers stated in a chunk over the weekend, “Manus probably isn’t China’s second ‘DeepSeek moment’.”
Is Manus AI the actual deal?
Not like DeepSeek R1, which developed its personal basis mannequin, Manus AI depends on current giant language fashions (LLMs), comparable to Claude and Qwen. This has led some critics to query how a lot of Manus’s expertise is actually authentic.
Yichao Ji acknowledged this within the official introduction video, stating: “Manus operates as a multi-agent system powered by several distinct models.”
Regardless of this, Ji emphasised that Manus AI is dedicated to open-source improvement and can launch a few of its fashions to the general public later this yr.
At the moment in personal beta, Manus AI is just out there through invitation codes. The corporate plans to develop entry quickly, however server limitations could delay a full public launch. An open-source element is deliberate, permitting builders to combine elements of its AI system.
Whereas some view Manus AI as a breakthrough in autonomous AI brokers, others stay skeptical about its true capabilities, enterprise mannequin, and scalability. Because the hype round Manus AI continues to develop, the query stays: Is that this the subsequent DeepSeek second, or simply one other overhyped AI product?
