Behind the explosion of Manus: shell controversy and AI new trend

What is Manus, and Why is it Suddenly So Popular?

Recently, the tech world has welcomed a new "internet sensation" — Manus. This name has rapidly gained popularity across major social media platforms, becoming a hot topic among tech enthusiasts, industry experts, and the general public. A simple search for "Manus" on social media reveals a flood of discussions, reviews, and awe, with its popularity rivaling even the hottest entertainment topics.

So, what makes Manus stand out in the highly competitive AI field? Manus is no ordinary AI; it is the world's first truly universal AI Agent, described as a "digital proxy" for humans. Unlike traditional AI, which typically excels in specific areas like image recognition or language translation, Manus breaks the mold by possessing the ability to think independently, plan, and execute complex tasks, delivering complete results. It is hailed as the "all-rounder" in the AI world.

Imagine planning a spontaneous trip. In the past, this would require hours of searching for travel guides, booking flights and hotels, and mapping out itineraries. Now, all you need to do is tell Manus, "Plan a seven-day trip to Yunnan for me; I love natural scenery and local cuisine," and it will spring into action. By autonomously searching various travel websites, analyzing user reviews, and comparing prices, Manus can quickly create a detailed and personalized travel plan, including daily itineraries, recommended attractions, local delicacies, and even transportation and accommodation options, generating a comprehensive travel guide. Isn't that more efficient and thoughtful than a professional travel planner?

Key Features: The Magic of "One-Click Completion"

Manus's standout feature is best described as "one-click completion." Its official website showcases numerous real-life examples of this magical capability.

Take PowerPoint presentations, for instance. In the past, creating a PPT required gathering information online, structuring the content, and designing each slide, often consuming significant time and effort. Now, simply instruct Manus to "create a PPT on the trends in artificial intelligence," and it will instantly get to work. Manus will sift through vast amounts of online data, organize a clear logical framework, and use elegant templates and appropriate charts to present the content on each slide. In just a few minutes, a well-designed, content-rich PPT is ready, saving time and making it easy to handle presentations.

Consider the task of resume screening, a tedious and time-consuming job for HR professionals. With Manus, this process is revolutionized. When HR provides Manus with a compressed file containing numerous resumes, it acts like an tireless and efficient assistant, automatically decompressing the file, meticulously reviewing each resume, and accurately extracting key information such as work experience, educational background, and professional skills. Manus then ranks candidates based on these criteria and generates an Excel report. With this report, HR can quickly identify suitable candidates, significantly improving recruitment efficiency and saving about 80% of the time and effort, allowing HR to focus more on critical tasks like interviews.

Manus also excels in travel planning. As mentioned earlier, when you express your travel preferences to Manus, it quickly integrates various information. By accessing major travel websites, analyzing user reviews, and comparing different transportation and accommodation options, Manus creates a highly detailed travel plan tailored to your needs. This plan includes daily itineraries, recommended attractions, local delicacies, and even the best photo spots, along with weather conditions and travel tips, truly enabling a "go-anytime" travel experience.

These examples highlight Manus's power. It transforms tasks that previously required multiple software tools and manual effort into simple, one-click operations, significantly boosting efficiency in both work and life. This is one of the key reasons for its rapid rise to fame.

After the Hype: Controversies and Doubts Emerge

(1) The Shell Controversy: Technological Integration or Simple "Shelling"?

Just as Manus was enjoying its moment in the spotlight, a "shelling" controversy began to brew. On the evening of March 10, founder Ji Yichao's revelation on social media was like a stone thrown into a calm lake, creating ripples. He admitted that Manus's product uses various fine-tuned models based on Alibaba's Qwen model, along with other models like Claude. This revelation sparked intense discussions online, with accusations of "shelling" flooding in.

In the AI field, technological integration is a common practice to drive development, with many excellent products borrowing and combining various technologies to achieve more powerful functionalities. However, the term "shelling" carries a negative connotation, implying that the product might simply repackage existing mature models without core proprietary technology, making it difficult to establish true competitiveness. While Manus does not have its own models, it leverages existing powerful models like Claude and Alibaba's Tongyi, fine-tuning them to achieve functional integration. Some view this as a clever practice of technological integration, reducing R&D costs and accelerating product launch, allowing users to enjoy the benefits of advanced technology sooner. Others, however, see it as mere "shelling," arguing that Manus lacks deep independent innovation and is dependent on external models, which could pose risks if those models encounter issues or change strategies.

From an industry perspective, this controversy has sparked deeper reflections on the boundaries between technological integration and shelling. In the pursuit of innovation, how should one balance borrowing and independent R&D? Should companies invest heavily in proprietary R&D to build their own technological barriers, or can they take a different path by quickly realizing product applications through clever technological integration? This is not just a question for Manus but a challenge for the entire AI industry.

(2) Capability Doubts: Can Real-World Performance Match the Hype?

Beyond the shelling controversy, Manus's actual capabilities have also faced scrutiny. Some bloggers who received invitation codes eagerly tested Manus, only to find that its performance did not live up to the hype. For example, one blogger asked Manus to "collect 50 different life reflections posted on social media at 22:43 on March 6 and compile them into a PDF." While the task seemed straightforward, Manus struggled. It failed to gather the required data and, surprisingly, started fabricating content, even admitting at the end that it couldn't find the data and had to make up some to complete the task. This result was far from Manus's advertised capabilities, disappointing many users who had high expectations.

Overseas users also reported issues. Some pointed out that Manus frequently encountered server errors, severely impacting the user experience. It also tended to hallucinate, producing factually incorrect content. In certain areas like programming and scientific applications, its performance was even inferior to traditional Google searches. TechCrunch also encountered numerous problems during testing, whether it was ordering food, booking restaurant reservations, or buying tickets. Manus either made errors that halted tasks or delivered subpar results, leading TechCrunch to comment that Manus is currently more about hype than technological innovation.

These real-world issues have led people to reassess Manus's capabilities. Its advertised features like "one-click completion" and "independent thinking, planning, and executing complex tasks" seem to fall short in practice. Of course, any new product is bound to have teething problems, and Manus is no exception. Perhaps these issues will be resolved with further optimization and iteration, but for now, these doubts cast a shadow over Manus's development.

AI "Shelling" Applications: A New Trend or a False Proposition?

Manus's shelling controversy is not an isolated incident; it reflects a growing phenomenon in the AI industry — the rise of AI "shelling" applications. In Silicon Valley, applications built on large language models (LLMs) created by other AI developers are becoming hot commodities for venture capitalists.

Taking Harvey as an example, this AI assistant focused on the legal industry was founded in 2022, and its core technology is based on OpenAI models. Although it does not have a self-developed model, it provides efficient services to the legal community by deeply integrating AI technology with the legal field. In December last year, Harvey's annual recurring income exceeded $50 million, and earlier this year it received a $300 million financing led by Sequoia Capital, with a valuation of up to $3 billion. There is also Anysphere, a popular code editing tool developed by Cursor, which is also based on LLMs from multiple providers. It achieved $100 million in annual recurring revenue in just 12 months and received $105 million in financing led by Thrive Capital and Andreessen Horowitz in January of this year, with a valuation of $2.5 billion. These successful cases demonstrate that even for "shell" applications, as long as they can identify market positioning and solve pain points in specific fields, they can still achieve great commercial success.

From a market demand perspective, the emergence of AI shelling applications is reasonable. As AI technology rapidly advances, large models have become increasingly powerful. However, average users often lack the ability and experience to use these models directly, struggling to formulate precise requests or evaluate and apply the model's outputs effectively. Shelling applications, by encapsulating and optimizing large models, provide more user-friendly interfaces and functionalities, lowering the barrier to entry and allowing more people to benefit from AI technology. For example, some shelling applications offer industry-specific prompts and templates, enabling users to input simple information and receive results that meet professional standards, significantly improving work efficiency.

However, AI shelling applications also face numerous challenges. On one hand, the technical barrier to creating shelling applications is relatively low, making them easy to replicate and imitate, leading to fierce market competition. Without unique value propositions and differentiated competitive advantages, these applications can quickly be淘汰. On the other hand, relying on external large models poses risks in terms of data security and model stability. If the underlying models encounter issues, receive updates, or change strategies, the shelling applications' normal operations could be affected. Additionally, for companies pursuing technological innovation and core competitiveness, the shelling model may not meet their long-term development needs, making it difficult to achieve true technological breakthroughs and leadership.

For Manus, while the shelling controversy has somewhat tarnished its image, it does not mean the end of the road. The key lies in continuously optimizing product features, enhancing user experience, and building unique brand value. By establishing closer relationships with model providers, ensuring technological stability and sustainability, increasing investment in technological integration and application innovation, and developing more user-centric features and services, Manus can strengthen its brand and market presence, boosting user awareness and trust. Only then can Manus stand firm in the competitive market and truly realize its value as a universal AI Agent.

Manus's Future: Breaking Through Challenges and Leading Change?

As a rising star in the AI field, Manus faces numerous controversies and doubts but also unprecedented opportunities for development. Its future trajectory not only affects its own success but also influences the direction of the AI industry.

From a technical perspective, Manus needs to continuously enhance its capabilities to address various challenges. To tackle current issues like server errors and hallucinations, the Manus team must invest more in server optimization and model training. By improving server architecture, they can enhance stability and response speed, ensuring smooth user experiences. In model training, incorporating higher-quality data and refining algorithms can improve accuracy and reliability, reducing hallucinations. Additionally, Manus should expand its application areas, enhancing capabilities in programming and scientific research to meet diverse user needs. For example, collaborating with research institutions to optimize complex scientific tasks could enable Manus to play a more significant role in the research field.

In terms of business models, Manus also needs to explore and innovate. Currently, Manus operates on an invitation code system, but it may transition to a paid subscription model in the future. This transition requires balancing user acquisition costs with service quality. On one hand, pricing must be reasonable to attract and retain users; on the other hand, service content and quality must continuously improve, offering premium users additional benefits like exclusive task templates, faster task execution, and priority technical support. Manus could also explore partnerships with other companies, licensing its technology or developing customized solutions for various industries, enabling diversified revenue streams. For instance, collaborating with financial institutions to provide intelligent investment analysis and risk assessment services, or partnering with educational institutions to develop AI-assisted teaching tools, could enhance teaching efficiency and quality.

From an industry development perspective, Manus's emergence offers new ideas and directions for the AI field. It showcases the immense potential of AI Agents in practical applications and sparks deeper reflections on technological integration and innovation. As technology continues to evolve, AI Agents could become a significant form of AI application, widely used across various fields. If Manus can stand out in this transformation, it will not only achieve commercial success but also drive the AI industry toward more practical and efficient directions. It may encourage other companies to invest more in AI Agent development, accelerating technological iteration and innovation, fostering a healthy competitive environment, and promoting the global adoption of AI technology.

Manus' future is full of uncertainty, but it also contains infinite possibilities. Only by continuously solving the current problems, innovating and optimizing, can we stand invincible in the fierce market competition and truly become a leader in the AI field. Let's wait and see if Manus can break through the predicament and write his own brilliant chapter.