GuShiio.com Article source: Boundaries Unknown
Image source: Generated by AI
AI has changed again.
On the morning of March 6, a product video called Manus was posted on the entire network. In the video, Peak, co-founder and chief scientist of Manus AI, said:Manus is the first universal AI agent.
Some netizens lamented after watching the video,“Tonight is as historic as the night when ChatGPT was released.”
Since the morning of March 6, Manus’s official website has been in a state of collapse. The circle of friends and WeChat groups are full of people who ask for invitation codes but cannot get them.
▲Screenshot of the crashed Manus official website
That afternoon,The price of one of Manus’s internal beta codes was called to 60,000 on idle fish.
▲The “outrageous” selling price of Manus invitation code on idle fish
Many people lamented that Manus’s voice was louder than DeepSeek. So, as the “first universal AI agent”, what are the differences between Manus and previous large model products?
Manus who is not “bullshit”
In official terms,”This is not just a chatbot or workflow, it is a truly autonomous agent that bridges the gap between concept and execution, and Manus delivers results while other AI knowledge is producing ideas.”
Speaking in human language means,Other AI talks nonsense to you, Manus helps you work.
From an example on the official website, when a user asked Manus to help write a “7-day trip to Japan and proposal for marriage.” Manus will think like DeepSeek, break down tasks and make a to-do list. Then it will start to browse the web, collect data, and organize information like humans do.
We can see that a small window will appear on the right side of the screen, which simulates the screen status and operation process seen by Manus.
▲Screenshot of Manus task processing
Even Manus encountered problems with Captcha when visiting a web page and was displayed.
▲Manus encountered a Captcha when visiting a web page
After the relevant work is completed, Manus will output six documents, including a seven-day itinerary, a summary of romantic marriage proposal locations, a Japanese tourism research, an HTML file, and a to-do list (this is not something users need to complete when going to Japan, but what Manus needs to complete during the planning process, such as understanding the basics of traveling to Japan, studying cities and attractions, creating itineraries, etc.).
▲Work result document output by Manus
Taking the core “7-day itinerary” as an example, Manus detailed the daily itinerary and where coupons can be redeemed.
▲Screenshots of Manus’s travel plans
For comparison, after we translated the same prompt words into Chinese, we fed them to DeepSeek, Tencent version of DeepSeek, and Baidu version of DeepSeek respectively.
Among them, the official version of DeepSeek output a seven-day itinerary, specially recommended marriage proposal locations, HTML travel manual templates, budget allocations and other suggestions after in-depth thinking. The content is relatively simple, but the information is fully available.
But DeepSeek can’t output files like Manus does, requiring users to paste the content themselves.
▲DeepSeek official version of task output results
For comparison, the Baidu version of DeepSeek outputs HTML travel manual templates and budget allocations, which are relatively crude and not aesthetically pleasing.
▲DeepSeek Baidu version task output results
After thinking about it, Tencent’s version of DeepSeek only output HTML travel manual templates, and the typesetting was messy.
In comparison, we will find that Manus can perform multiple tasks at the same time and output documents that can be used directly.
▲DeepSeek Tencent version task output results
But if you want to say how much ahead Manus is than the currently widely used large models, it also needs to be divided into comparison objects.
For example, if compared with Tencent and Baidu versions of DeepSeek, Manus ‘results in performing tasks are obviously far ahead. But if compared with the DeepSeek official, Manus is actually better than the unified delivery of documents. The official version of DeepSeek can also complete the same task well.
Of course, what is shown above is just one of Manus cases. On Manus’s official website, there are many similar cases, such as in-depth analysis of Tesla stocks, sorting out a list of YC companies, online store operation analysis, and exporting a website SEO.
Judging from these cases, Manus has been able to help migrant workers in many work scenarios。
Manus’s three “controversies”
But just as Manus was surfing the entire network, some controversy always accompanied Manus.
For example: shell?
After Manus came out, some netizens commented:Putting the shell to the extreme is innovation?
This comment obviously has some negative emotional overtones. In theory, Manus really only combines a large model and multiple small models on cursor products, that is, it really doesn’t have much new on the technical level.
However, it is obviously wrong to just roughly summarize the work they do as a “shell”. In fact, many people underestimate the amount of work involved in integrating various large models and various agent + compute use capabilities.
▲Manus product launch video
The second controversial point is the issue of barriers.
As mentioned earlier, Manus has not made much innovation at the technical level. And what Manus can do, DeepSeek and Claude can do, and even OpenAI has developed similar products before.
There is a question here. When Manus’s products are released from the circle, will they be quickly followed up by major manufacturers and seize the market? Will Manus’s capabilities be internalized by basic models such as OpenAI and Claude in the future?
“I think Manus’s barriers are limited.”
Fang Yuan is a popular AI application entrepreneur who mentioned it to us after seeing Manus’s products. “Compared with the underlying R & D, product functions are easy to copy, such as changing buttons in bytes.” Fang Yuan said: “Because Manus is finally tuning various applications to help you do things, compared with startups, Byte and many large manufacturers have many application matrices themselves, which have more advantages in this regard.”
“I feel that after Manus, a large number of similar applications may come out soon.”Fang Yuan said.
Therefore, although Manus is another popular application after DeepSeek, unlike DeepSeek, which focuses on the research and development of basic large models, Manus relies on building applications based on the underlying large model to acquire users and achieve a business closed-loop.Therefore, how to open up a living space between large factories and model manufacturers may be a relatively practical problem.
But that’s not to say Manus t have a chance,“If it can pick up this part of the traffic well, Manus may become a traffic distribution place similar to the Agent application store, doing things that Baidu, Microsoft and other big companies have always wanted to do but have not done.”Fang Yuan said.
▲OpenAI announced agents in January
The third point of contention is how much value the work Manus has accomplished can create.
Judging from the 50 cases on Manus’s official website, it seems that it can replace the work of “migrant workers” in many scenarios.
For example, in the resume screening scene, He Lin is a senior HR. After watching Manus’s functional demonstration, she told “Boundless Unknown”,This function is more effective during large-scale recruitment, especially the ability to automatically organize all resume information. It is very practical in their daily work and can save a lot of time.
But this does not mean that Manus can replace HR.
“Manus should be able to complete two tasks. One is preliminary resume screening, which means screening out some resumes through some fixed conditions, such as whether they graduated from 985/211, whether they are under 35 years old, and whether they have three years of work experience.”
“After completing this part of the work, HR needs to manually screen the remaining resumes to determine those non-standard conditions, such as project experience, matching with the position, etc. For an experienced HR, it will be more advantageous in this regard.” He Lin said.
After completing this step, the Agent will score all resumes and screen out the last few resumes for the interview stage to improve work efficiency.
▲Screenshot of Manus resume screening workflow
In addition to resume screening, another Manus application scenario that “Boundless Unknown” understands is SEO optimization.
As a senior SEO, Wang Qiang registered two domain names about Manus early in the morning of March 6. After reading Manus’s SEO demonstration case, WangQiang said that it can basically reach the level of junior and middle school SEO.
But judging from the SEO optimization report on personal blogs output by Manus, what he has completed is only the preliminary work. “The SEO optimization report it outputs is very detailed on many levels, but the value of this report only lies in the diagnosis made when the website was just established or when SEO optimization was done for the first time.”
Wang Qiang said: “Because its analysis framework is mechanized, what affects website SEO is not only fixed factors such as website structure, content quality, and keyword layout, but also related to the macro public opinion environment on the Internet and the development background of the industry.”
▲Manus outputs a screenshot of the website’s SEO optimization plan
Therefore, Wang Qiang’s judgment is that Manus can currently complete basic work, but if he wants to rely on Manus to do SEO well and refined, judging from the current Demo, he cannot make a judgment.
Of course, the above judgments, whether it is HR’s resume screening or website SEO optimization report, are just judgments made by the two people based on their years of experience in their respective positions after watching Manus’s demonstration, or after in-depth experience of the product. It is unknown that Manus can play more value in these tasks.
After communicating with multiple HRs,”UnKnown” learned that many large state-owned enterprises have begun to explore combining DeepSeek with internal recruitment systems to explore application applications in interview assistance, resume screening, performance management and other aspects.。
The changes Manus brings
Although judging from the current judgment, Manus has not brought as much disruption to the industry as DeepSeek, the impact still exists.
“After seeing Manus, my first reaction was that I wanted to open up my software and make an interface that supports AI tool calls.”
Fang Yuan believes that many applications and websites after Manus may have open interfaces for calls by software like Manus. The sooner he does this, the easier it will be for the product to be promoted.“In order to facilitate calls, a data standard similar to the current communication protocol may be formed between various products in the future.”
In the future, every large model application may be a very vertical workflow, and software like Manus will become a hub that calls various applications to complete tasks. The value of application entrepreneurs will evolve into doing their applications well enough in segmented scenarios so that software that calls the workflow can choose.
▲OpenAI founder Sam Altman introduces the Agent application platform
In addition, Fang Yuan also felt that Agents have transformed today’s corporate workflow.“The company’s previous product development process is no longer suitable today.”
In the past, when developing a software, the product manager first drew a sketch and then handed it to the interaction designer. After the interaction designer decided on the interaction, he handed it to the visual designer. After this series of work was completed, the product manager then conducted R & D review, discuss the development cycle, collaboration methods, etc.
Nowadays, this work process is being completely subverted in the AI era. With the help of AI, many links have been simplified and abandoned. Many development processes that originally required the collaboration of multiple roles can work independently, such as front-end development and visual design. etc.
AI is indeed transforming our work and lives more and more extensively.
This phenomenon is gradually confirming the judgment mentioned by OpenAI founder Sam Altman in his year-end summary at the end of 2024:“We believe 2025 will see the first artificial intelligence agents join the workforce and fundamentally change the company’s output.”
Since 2025, major model companies have also bet on Agents.
In early January, OpenAI announced the launch of a beta feature called Tasks that allows users to automate processing by simply explaining requirements and timing to ChatGPT. In early February, Microsoft CEO Nadella announced that GitHub Copilot will all-in agents and launch Microsoft’s first independent SWE agent. In China, Step Star’s latest press conference also announced the creation of an Agent cooperation ecosystem. It divides agents into two categories: vertical agents and intelligent terminal agents, and cooperates with different ecological partners.
Therefore, although Manus became popular suddenly, it was definitely not an accident.
It is almost foreseeable that after Manus, more and more Agent applications will continue to grow, bringing one surprise after another. We can wait and see.
