Closest to the user, creating the best AI Agent.
Authors: shiyun, Zhang Yongyi
2025 is the inaugural year of AI Agents—this statement was validated in the early hours of March 6, Beijing time.
"After DeepSeek, another sleepless night in the tech circle."
Many users commented this way on social media.
Everyone stayed up all night just for a usage invitation code for a product—the world's first AI Agent product "Manus" developed by Monica.im.
According to the team, "Manus" is a truly autonomous AI agent capable of solving various complex and changing tasks. Unlike traditional AI assistants, Manus can not only provide suggestions or answers but also directly deliver complete task results.
The introduction video of Manus is only 4 minutes long but incredibly powerful.
Image source: Monica.im
As the name "Manus" suggests, it symbolizes "hand" in Latin. This means that knowledge should not only reside in the mind but also be executable by hand. This is the essential evolution of Agent and AI Bot (chatbot) products.
What makes Manus impressive? The most intuitive way is to look at the official website's showcase and user-generated use cases, which Geek Park has summarized as follows:
Travel Planning: Not only integrates travel information but also creates customized travel manuals for users. For example, planning a trip to Japan in April, providing personalized travel suggestions and detailed manuals.
Stock Analysis: Conducts in-depth stock analysis and designs visually appealing dashboards to present comprehensive stock insights. For example, performing deep analysis on Tesla stocks and creating a visual dashboard.
Educational Content Creation: Creates video presentation materials for middle school teachers, explaining complex concepts like the momentum theorem, helping teachers teach more effectively.
Insurance Policy Comparison: Creates clear insurance policy comparison tables, providing the best decision-making advice to help users choose the most suitable insurance products.
Vendor Procurement: Conducts in-depth research across the network to find the most suitable vendors for user needs, serving users as a truly fair agent.
Financial Report Analysis: Captures market sentiment changes towards specific companies (like Amazon) through research and data analysis, providing market sentiment analysis for the past four quarters.
Startup Company Listing: Visits relevant websites to identify eligible companies and organizes them into tables. For example, compiling a list of all B2B companies from the YC W25 batch.
Online Store Operation Analysis: Analyzes sales data from Amazon stores, providing actionable insights, detailed visualizations, and customized strategies to help improve sales performance.
When the Agent outputs an incredibly complete and professional result through a long chain of thought processes and tool calls, users begin to exclaim, "It really can help humans get things done."
According to information from the official website, Manus achieved new state-of-the-art (SOTA) performance across all three difficulty levels in the GAIA benchmark test (assessing the ability of general AI assistants to solve real-world problems).
In summary—what Manus aims to be is your literal "agent" in the digital world. And it has succeeded.
Just as you might expect, the launch of Manus in the early morning instantly woke up everyone in the AI circle!
01 Manus, Your "Digital Agent"
First, Manus differs from previous LLMs in its experience:
It emphasizes the ability to directly deliver final results rather than just providing a simple "answer."
Manus currently adopts a Multiple Agent architecture, operating similarly to the Computer Use released by Anthropic, running entirely in independent virtual machines. It can also call various tools in the virtual environment—writing and executing code, browsing the web, operating applications, etc., to directly deliver complete results.
In the official release video, three work cases completed by Manus in real-world scenarios were introduced:
The first task is resume screening.
From 15 resumes, it recommends suitable candidates for the reinforcement learning algorithm engineer position and ranks them based on their expertise in reinforcement learning.
In this demonstration, you don't even need to unzip the compressed file or manually upload each resume. Manus already shows a "intern" side, manually unzipping the file and browsing each resume page by page while recording important information.
Manus automatically understands the hidden instruction "unzip the compressed file thrown by the boss" like an intern.
Image source: Geek Park
In the results provided by Manus, there are not only automatically generated ranking suggestions, but it also categorizes candidates into different levels based on important dimensions like work experience. When the user prefers to present the results in an Excel format, Manus can automatically generate the corresponding table by writing Python scripts on the spot.
Manus can even remember user preferences during the content generation process, such as "the user prefers to receive results in table format," and will prioritize presenting results in table form for similar tasks in the future.
Manus can remember user preferences in the content generation process.
Image source: Geek Park
The second case is tailored for the Chinese audience, which is selecting real estate.
In the case, the user wishes to purchase property in New York, with requirements for a safe community environment, low crime rate, and quality primary and secondary education resources—of course, including the most important budget that is affordable within their fixed monthly income.
In this demand, Manus AI breaks down the complex task into a to-do list, including researching safe communities, identifying quality schools, calculating budgets, and searching for properties. It carefully reads articles about the safest communities in New York and collects relevant information through web searches.
Next, Manus writes a Python program to calculate the affordable property budget based on the user's income. It filters the property list according to the budget range using relevant housing price information from real estate websites.
Manus can automatically search and filter out properties that do not meet user requirements.
Image source: Geek Park
Finally, Manus integrates all collected information and writes a detailed report, including community safety analysis, school quality assessment, budget analysis, recommended property list, and relevant resource links—just like a professional real estate agent. Moreover, due to Manus's inherent attribute of "fully considering user interests," its use and experience are even better.
In the last case, Manus demonstrated its ability to analyze stock prices.
The task given was to analyze the correlation between the stock prices of Nvidia, Marvell Technology, and TSMC over the past three years: it is well-known that these three stocks are closely related, but for novice users, it is challenging to quickly sort out the causal relationships.
Manus's operation is very similar to that of a real stockbroker; it first accesses historical stock data through APIs from information websites like Yahoo Finance, while also cross-verifying the accuracy of the data to avoid being misled by a single information source, which could significantly impact the final results.
In this case, Manus also utilized its ability to write Python code, perform data analysis and visualization, and introduced professional financial tools for analysis, ultimately providing users with causal relationships through data visualization charts and detailed comprehensive analysis reports—just like the daily work of a "intern" in the finance field.
Moreover, the Manus official website showcases over a dozen scenarios where Manus can be used: directly using Manus to organize itineraries, personalize travel routes, and even allowing it to learn to use various complex tools to streamline daily work.
What truly sets Manus apart from traditional tools is its autonomous planning to ensure task execution capability.
The ability to learn independently also enhances Manus's work capability logic to resemble that of a real human—although at this stage, it may not yet achieve expert-level proficiency in a specific field, the potential is already evident.
With the addition of autonomous learning capabilities, the versatility of AI Agents has significantly improved. In user tests of Manus, you can even describe relevant content from a video frame directly to it, and Manus can accurately find a specific Douyin short video link based on the corresponding information, overcoming the limitations of platform content for search engines.
Since the current version of Manus operates entirely in the cloud asynchronously, its capabilities are not limited by the end-user platform form or computing power—users can even temporarily shut down their computers after issuing commands to Manus, and it will automatically notify them of the results once the tasks are completed.
This operational logic is also very familiar—just like someone asking an intern on WeChat to "send me the organized files" after work. However, now, this intern can truly respond to you 24/7 without worrying about "office politics."
02 Multiple Agents + Self-Check, Running the AI Agent Flow
From the above cases, it is not difficult to see that Manus's real trump card is not the "AI Agent" concept that has appeared in Computer Use, but its ability to "simulate human working methods."
Compared to "running calculations," Manus's working logic resembles "thinking and executing commands." It does not accomplish tasks that humans currently cannot do; this is why some users who have experienced the current version of Manus describe it as "an intern."
On the Manus official website, numerous tasks that Manus can complete are showcased, including a case demonstrating how to use Manus in B2B business. It quickly and accurately matches your ordering needs with global suppliers.
In conventional products with similar demands, integrating global supply chain enterprise information within the platform to help users complete supplier/demand matching is a common logic in the industry. However, in the case of Manus, you can see a completely different implementation approach.
Manus AI uses a framework called "Multiple Agent," running in independent virtual machines. Through a collaborative mechanism of planning agents, execution agents, and verification agents, it significantly enhances the efficiency of handling complex tasks and shortens response times through parallel computing.
In this architecture, each agent may be based on independent language models or reinforcement learning models, communicating with each other via APIs or message queues. At the same time, each task runs in a sandbox to avoid interfering with other tasks while supporting cloud scalability. Each independent model can mimic the human process of handling tasks, such as first thinking and planning, understanding complex instructions, breaking them down into executable steps, and then calling appropriate tools.
In other words, through Manus's multi-agent architecture, it operates more like multiple assistants working collaboratively to complete tasks such as resource retrieval, connection, and verification of information validity, helping you complete the entire workflow—this is not just like hiring an "intern," but more like directly becoming a miniature "department head."
In the B2B business case, Manus automatically searches the vast ocean of the internet using web crawlers and coding capabilities, matching potential suppliers based on your own needs in terms of product quality, price, delivery capability, etc. It not only presents conclusions visually in charts but also provides more detailed operational suggestions based on this data.
Manus may be more useful in B2B scenarios than built-in tools from a single platform.
Image source: Geek Park
As for how the Monica team achieved the video effects, reports suggest that the team may reveal this on March 6, Beijing time.
03 The Ultimate "Sewing" is Explosive
What kind of company is Monica.im behind Manus?
Monica is an All-in-One AI assistant, with its product form evolving from a browser plugin to an app and web version. The mainstream usage scenario is that when users click its small icon in the browser, they can directly use the major mainstream models it connects to. By accurately understanding the segmented user needs, Monica has picked the "low-hanging fruit" of large models.
Its founder, Xiao Hong (nickname Xiao Hong, English name Red), is a young serial entrepreneur born in 1992 and graduated from Huazhong University of Science and Technology. After graduating in 2015, he started his first venture, which was not very successful in the early stages (such as campus social networking and second-hand markets). In 2016, he launched a tool for WeChat public account operators that provided editing and data analysis, gaining a million users and achieving profitability, ultimately selling the product to a unicorn company in 2020.
By 2022, after the wave of large models, he officially founded Monica, focusing on overseas markets. Through the independently developed product ChatGPT for Google, the product quickly completed its cold start.
In 2024, at the first opportunity of the launch of GPT-4o, Claude 3.5, and OpenAI o1 series, Monica allowed users to access the latest SOTA models. With new developments in model integration, features like professional search, DIY Bot, Artifacts for writing mini-programs, and memory have also been well-received by users. Monica presents different interactive forms and functions across various web pages like YouTube, Twitter, Gmail, and The Information, updating hundreds of web pages with personalized AI experiences to meet specific user needs.
In 2024, the number of Monica users doubled to 10 million. At the same time, it maintained considerable profitability, ranking among the top in similar overseas products.
Monica's strong performance validates one thing:
Maximizing the shell leads to both TPF and PMF, ultimately all pointing to user value.
Monica Homepage
Image source: Monica
Manus may continue this line of thinking from the Monica team—Xiao Hong stated in an interview with journalist Zhang Xiaojun that products cannot only have the chatbot form; Agents will be a new form that requires new products to support.
He drew inspiration from AI programming products Cursor and Devin. According to Geek Park, the former mainly operates in a copilot mode, while the latter is more aligned with human needs in an autopilot mode. Agents should also be like Devin, aimed at the general public, truly executed by AI. However, the past issue was that the models were not smart enough.
But based on the existing capabilities of the models to provide scene encapsulation services, this may be the advantage of the Monica team. Xiao Hong mentioned that there are currently not many teams working on Agent products because it requires many composite abilities, such as experience in chatbots, AI programming, and browser-related work (since everything runs in the browser), and a good perception of the model's boundaries—what level it has developed to today and what level it will develop to next.
"There aren't many companies that possess all these capabilities, and those that do may be focused on a very specific business, but we happen to have team members who have the time to work together to make this happen," he said.
Why was it Monica that made this happen? He summarized, "First, I think we are quite lucky. Second, to some extent, if everyone is doing reasoning today, could it be that some time has been freed up for startups? How far can the model's expected capabilities overflow?"
He believes that Agents are still in the early stages. First, they are still in the planning phase and have not yet reached execution in the physical world; second, the capabilities of large models are still developing, and everything remains unpredictable.
"I certainly didn't know that Agents could be developed in this way; it is an unknown matter," he said.
Interestingly, Monica, which "doesn't know how to make Agents," has now created a product that has shocked the entire AI circle.
Manus may not be the final AI Agent, but it undoubtedly raised people's expectations for AI by another order of magnitude after the explosive success of DeepSeek.
免责声明：本文章仅代表作者个人观点，不代表本平台的立场和观点。本文章仅供信息分享，不构成对任何人的任何投资建议。用户与作者之间的任何争议，与本平台无关。如网页中刊载的文章或图片涉及侵权，请提供相关的权利证明和身份证明发送邮件到support@aicoin.com，本平台相关工作人员将会进行核查。
