GuShiio.com 
Image source: Generated by AI
If we look at the headlines of AI media, no one in the AI circle around the world should have survived yesterday because they were “bombed” again.
The bomb dropped this time was a product called “Manus”. It is an AI Agent product developed by a China team and released only through an all-English promotional video. It cannot be used without an invitation code. Officials call it “the world’s first universal Agent.”
It was as if OpenAI’s Operator and Anthropic’s Claude Coders were blown out of existence overnight.
After its release, it quickly spread in a small circle, exploding constantly, and quickly moved towards another familiar flavor:
Another DeepSeek is here in China.
China definitely needs more than one DeepSeek, but what kind of existence it should be is a question.
1、As if Open AI and Anthropic don’t exist
Without looking at these titles first, let’s take a look at what product Manus is.
According to its videos and information on its official website, it is an agent that can “solve various complex and changeable tasks, and can flexibly call various tools in its own virtual environment to directly deliver complete tasks for you through independent thinking and system planning.” Task results, not just providing suggestions or answers.”
Today, all the obscure AI Agents will basically introduce themselves in this way.
It is not currently publicly available and can only be used through invitation codes, while its release is mainly through a demonstration video.
In the demonstration video, it shows three scenarios: uploading a compressed package containing multiple resumes, and based on your prompt words, Manus automatically helps you decompress and analyze your resumes, and then gives suggestions; proposing your real estate needs, Manus automatically helps you organize house prices and information; and generates analytical reports on Nvidia and Tesla stock prices.
You should know that OpenAI’s Operator and Smart AI’s Phone Use have previously demonstrated similar functions on real computers, and Google Gemini’s last year demonstrated that they were already multi-modal agents that can play games with you through a camera based on 10 minutes of memory.
If you are familiar with the iteration direction of mainstream model capabilities and the current status of AI products, you will quickly realize that these scenarios and functional displays have been implemented in products with various Agent concepts, and Manus’s interfaces are basically Devin and Artifact look.
There is no official report on the technology. Someone has disassembled the method behind it, which is basically a combination of Claude’s basic capabilities and the post-training planning capabilities of the open source model. At the same time, various Agents are built in to pass a certain set workflow to complete tasks.
What’s interesting about it is that, unlike most previous Agent products, Manus “encapsulates” these capabilities into a cloud virtual environment, and it is also a new product in the trend of “asynchronous” interaction.
However, these innovations in development engineering are far from the “universal” approach to end-to-end implementation that the industry is already pursuing today.
Compared with using a long string of words to describe an Agent, Anthropic recently gave a clearer definition of an Agent in a sharing: Agents are models using tools in a loop-it must be a powerful model first and foremost. As currently the most powerful Agent model, Claude’s implementation of “universal” is the powerful model itself, plus a universal protocol (MCP).
This also points to a problem faced by products like Manus, which is that these capabilities can be achieved by essentially any model with stronger Agent capabilities. And Manus did not give a solution.
2. If you want to work side by side with DeepSeek, you cannot rely on self-hi and invitation codes
But none of this prevents Manus from becoming a “swivelling” product in a small circle.
Judging from the design of this product, it is a casing Devin that targets “ordinary users”, turning Devin into a part of the product process that users do not need to perceive, and then providing users with more direct use. Such products are actually popular in Perplexity. It has also proved to have a market.
Its approach is also very flexible. Marketing is an important part. It will design the sharing mechanism. For example, users can share their actions recorded on Manus on social media, thus creating communication.
The more classic one is the mechanism of the invitation code.
The invitation code is a promotional design that always appears during the self-improvement stage of a small circle. It is also the basic operation when a product is created in the mobile Internet era, a kind of muscular memory from the previous era.
Manus’s communication can be called a replica of the classic template of the mobile Internet: invitation code hunger marketing (“I used my connections to get it”)+KOL scream evaluation (“Human beings are defeated” and “numb all over”)+ leverage other celebrity products (this time we are bound to DeepSeek). However, when people open overseas technology forums and social media, the discussion about Manus is almost vacuum.
Are you bullying people for not understanding English? Where is the imaginary “Sleepless Silicon Valley”? Even if someone is not asleep, they may be discussing the open source version of QwQ by Tongyi Qianwen.
Obviously, what Manus himself wants to pursue is the popularity of DeepSeek. Based on the spread and expectations it creates, it also wants to compete with DeepSeek. But the path seems to be wrong. Deepseek has never produced invitation codes. Before it became popular in the industry, it has been a product that has been deeply involved in the developer community for a long time and has formed a reputation through sincere open source.
But these early work became unimportant to those who wanted to be on a par with DeepSeek. Later, the “domestic light” style of going out of the circle became a technical operation rather than a continuous work.
Except for your random search today, the title you came up with is:
“China’s AI Double Star: DeepSeek Technological Breakthrough, Manus Commercial Implementation” and something like that.
Manus himself is still emphasizing its results on GAIA and saying it has defeated OpenAI.
Yes, the Manus team has a good background. It produced an AI product that was a little innovative in engineering and front-end interactions, and then became a social asset for a small circle. Coupled with a round of “burst” communication, it took advantage of people’s deep impression of the DeepSeek phenomenon and created a sentiment that “another China team has made Silicon Valley sleepless” that does not exist. It is indeed effective, but it is not serious.
China needs the next DeepSeek too much, but the “next DeepSeek” cannot be self-proclaimed and self-satisfied. The situation is good now. If many people in Silicon Valley really noticed Manus, were lucky enough to get the invitation code, and then were disappointed and commented after the test, it would really consume the positive value DeepSeek brings to China AI, which should be something that everyone would not want to see anyway.
I hope not.
