To be honest, I still don't have an invitation code for Manus, so if you insist that without experience you have no right to speak, you can just exit and save yourself ten minutes.
I want to discuss two main points: one is about the marketing controversy surrounding Manus; the other is Manus's clever product approach.
Product: No Breakthrough, Yet a Breakthrough
Manus has not achieved a technological breakthrough, which may be the biggest consensus after the controversy. The core validation case comes from the MetaGPT team (the programming agent team) that replicated OpenManus in three hours.
But there is no doubt that the product Manus has brought to everyone is shocking. It uses AI as the "human hand," automating a series of processes that include a lot of information collection, browser interaction, and more. These will ultimately be encapsulated, and users only need to tell Manus what they want to do, then they can sit back and watch the show.
If you, like me, regularly follow developments in the AI field, you will find that Manus, compared to the innovations in deepseek model training, resembles more of a patchwork creation:
Task organization and knowledge base invocation. Most models have this capability, and from the perspective of prompt engineering, breaking down and organizing tasks in advance will improve the final output of AI. In my daily interactions with AI, I either list items directly or ask AI to organize based on my needs.
Information collection, organization, and analysis. This is also known as Deep Research, which is now basically supported by many.
External tool function calls. Whether it's MCP or a large number of open-source tools like browser-use, they have already been incorporated into daily usage scenarios.
Multi-agent collaboration. This has been around since Devin, probably for a year or two (I don't remember the exact time). The MetaGPT team that replicated it in three hours is doing multi-agent collaboration in the programming field.
Many people cling to the point that Manus has no innovation and start to mock it when they see praise.
One must know that arrogance is the greatest enemy of progress.
It might be worth asking yourself: since these are all existing things, why did integrating them suddenly explode? If they are all existing things, why don't you patch them together?
People's perspective on innovation is still too narrow, always fixating on technology, unaware that innovation in product concepts and business models has a far more profound impact.
I have included an article at the end that provides a very good understanding of the Manus product (because it is a paraphrase of the team's own words):
This statement is also the main reason I wrote this article. I believe the points mentioned here are worth savoring for everyone involved in product development.
Let me share what I see as valuable lessons from Manus:
1. Interference Resistance in Agent Workflows
In gaming, a similar concept is often mentioned: "flow."
The user experience of a product is composed of a series of behavioral processes. For example, when I use Alipay to pay a bill, I need to open Alipay, click "Recharge" on the homepage, select a phone number, choose an amount, confirm, pay, and receive feedback that the recharge was successful. If an external factor interferes with this process, it will disrupt the overall flow and degrade the product experience.
Existing AI agents, while not completely excluding user workflows as described above, do have issues with occupying workspace. For instance, in the commonly used AI web reader, the AI provides understanding based on the current webpage, and you cannot modify the current webpage; otherwise, it becomes ineffective. Otherwise, you have to manually copy the link and provide it to the AI in a separate window. Another product I recently tried, same.dev, which focuses on directly copying front-end source code, has the issue of interrupting operations by bypassing the webpage, effectively occupying a workspace and resulting in a very poor experience.
Of course, this is not to say one is better than the other. But the changes at the product level of Manus are meaningful for our thinking about the evolution of AI agent forms.
From embedding AI in browsers to embedding browsers in AI interaction pages, the former meets users' daily needs (AI is just a part of work and life processes), while the latter meets the needs of intelligent agents (reducing interference from unrelated factors).
AI Agent products only need to expose an interactive page to users, receiving input, displaying processes, and outputting results. The process is only allowed to be displayed, and users will not be disturbed by the process.
I am reminded of the issues I mentioned earlier regarding the embedded browser in the OKX wallet. You cannot let users actively interrupt the process to achieve their goals.
The experience of interrupting "flow" is very poor.
2. Rethinking "External Tools"
In the eyes of most people, only standard protocols like MCP or pre-packaged libraries count as "external tools."
In fact, tools are defined as having stable input and output, allowing users to have clear expectations of a black box.
Programming, as one of the most deterministic application scenarios for AI, can package countless scripts and modules as tools.
It's fine to compare solving math problems from the perspective of model training; but if you directly apply a math problem from the perspective of model application, that would be foolish.
Why not write code to solve it instead of using some vector mapping?
Manus has actually pointed out a key point in the design of general-purpose agents: do not attempt to use AI to solve all problems directly.
AI is just a hand.
The hand uses tools to solve problems. Tools can be predefined or temporarily written.
Human - Hand - Tool - Task.
What does it matter how many layers of tools are in between?
3. Non-destructive Use Scenarios
Those who frequently program must deeply understand that different projects have messy dependency library versions; if one key library version does not match, the project may run into errors.
This is also the necessity of virtual environments in programming. npm installs based on project dependencies, Python creates virtual environments, Docker containers, etc. In my understanding, these are all to ensure the independent customization of environments.
This may be the consensus among all directly consumer-facing intelligent agents at this stage: do not invade the user's local environment, use the cloud.
Previous products like bolt.new and mgx.dev have chosen to run and debug directly in the cloud. However, these are all programming-type agents, and there is a lack of comparison for general-purpose agents.
In stark contrast to Manus's approach is another product, Highlight. After you download and install it, it creates a floating window on your desktop, which has some AI-integrated operations based on the current workspace.
At first glance, it seems eye-catching?
For example, I don't know how to crawl data; I switch to the browser page and let Highlight crawl for me?
From my personal experience, Highlight's approach has already interfered with the original workflow because I must keep the process running for Highlight to operate. In reality, anyone working will switch pages back and forth; it is impossible to wait specifically for AI to complete its operations. Additionally, if AI uses my browser to crawl, is it using my IP? Will it affect my future access?
The local scene has been disrupted.
These points may seem easy to think of when stated. But how to frame a series of designs is still worth experiencing.
Finally, let me say something about the limits of Manus, which is my personal expectation.
My expectation is to control expectations—AI cannot do everything for us. Even if future smart homes introduce AI modules that become "tools" usable by the "human hand"; even if more and more "tools" appear on the desktop that allow us to control production-level software with just natural language; many times, humans must still perform process control and verification. This is because AI's understanding of the world is built on a massive black box, and they can exhibit "hallucinations."
In any case, since we believe that human reality is greater than AI reality, humans need to verify the production of AI. The more we delegate to AI, the more content needs to be verified. This will ultimately balance out to a boundary.
This boundary is the limit of general-purpose agent products.
Marketing: Not Afraid of Controversy, Afraid of No Controversy
Looking at this wave of Manus's breakout, it is more accurate to say that there are two "hands of God" guiding it behind the scenes rather than spontaneous controversy within the circle.
These two hands of God likely come from the official side. Hand in hand, isn't that wonderful?
First, I had Grok summarize the major marketing events of Manus during this time:
It is clear that the core promotional slogan from the official side is "the world's first general-purpose AI agent."
This is a very controversial statement.
A. For outsiders, this statement is very eye-catching;
B. For someone like me who follows the field but is not an insider, it is immediately apparent that it is playing with words: as I mentioned earlier, the product is a patchwork and cannot be called the first in any case; but if you add the word "daily," there indeed hasn't been a consumer-grade product claiming to handle general tasks that has caused significant public opinion before. Moreover, promotional slogans are inherently exaggerated, so saying this is not unreasonable;
C. For insiders, I think most may feel quite indignant, as their hard-earned research results have been stitched into this, or they may feel that this thing has no technical content in their eyes, yet it has stolen the spotlight.
—Opposing positions have emerged.
Where there are opposing positions, there is debate, and public opinion ultimately continues to ferment, bringing a leverage effect to Manus's promotion.
Do not look at the surface; look at the results.
The result is that Manus has gained global attention. The cost-effectiveness of this marketing is maximized.
And the invitation code mechanism.
While attention is at its peak, the invitation code is strictly limited, partly due to cost considerations and partly to cover up product shortcomings. After all, having claimed to be "general-purpose," once opened up, it could immediately be bombarded with various bugs and feedback from Class A users. That would not create opposing positions but rather a one-sided situation, which could likely lead to failure. From this perspective, the invitation code mechanism is actually similar to an early testing phase; the more devs fix bugs, the more codes are released, allowing seed users to help improve the product first.
There is also the strategy of hunger marketing. Hunger marketing is essentially about competing for attention; those who used the invitation code will inevitably be scolded by those who "love but cannot have," which is very normal.
Subsequent events, such as the freezing of X accounts, technical jailbreaks, and being open-sourced, have been seen by many as the consequences of excessive marketing.
Regarding this, my personal view is that the AI circle is still too pedantic. I suggest taking a stroll in the crypto circle to learn a bit about the spirit of shamelessness.
The Manus team is actually doing fine, as their previous product, Monica, has already been profitably stable. However, many small teams are still struggling on the edge of survival. At this time, if you tell me that marketing should not be excessive? If high marketing effectiveness can be achieved at low cost, I would like to ask why not?
Can face value support the R&D team to continue? Can face value provide enough funds for innovation?
Face value is not worth much. In this age of entertainment to death and information explosion, attention is what is valuable.
The most brutal and also the most training ground in the crypto circle is that it is too close to money itself, so what you see are the most real human natures and the most bloody tricks (not that the tricks are bloody, but you are bloody).
One must have principles, but you cannot fail to understand and learn to accept some realities that are not in line with your principles.
Otherwise, it will end very badly.
References
Complete Review: How Did Manus Come to Be? | Geek Park
Behind the Explosion of Manus, How Do Agentic AI Products Build Lasting Competitive Advantages?
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。
