3 weeks ago
28
China has made a new advancement in artificial intelligence with Manus AI Agent, an innovation that is attracting global attention, including from Silicon Valley. Recently launched with an invite-only preview, Manus represents China’s most ambitious foray into the emerging market of AI agents.
Unlike traditional chatbots, this AI agent does not just converse with users, but is capable of independently managing complex and articulated tasks with minimal human intervention. This characteristic distinguishes it from previous models, bringing it closer to a truly operational artificial intelligence.
Manus AI Agent: a revolutionary architecture
Developed by the Chinese startup Butterfly Effect, with financial support from Tencent Holdings, Manus AI leverages a multimodel architecture that combines the capabilities of multiple advanced language models. This approach allows Manus to use different AI skills depending on the needs, thereby increasing its reasoning and task execution capabilities.
According to CNN Business, Manus can handle complex tasks such as selecting resumes and creating websites. Additionally, it is capable of formulating concrete suggestions, such as producing detailed reports on real estate to purchase based on specific criteria.
Testing and performance in the real world of Manus AI Agent
MIT Technology Review conducted an in-depth analysis of the new AI agent, testing it on three categories of distinct tasks:
– Compilation of lists of journalists on specific topics
– Real estate search based on complex parameters
– Identification of candidates for the Innovators Under 35 program
The test results showed that Manus AI works as a highly efficient assistant, demonstrating remarkable autonomy but also some limitations. According to the analysis, the agent sometimes does not fully understand the task required, makes incorrect assumptions, or seeks shortcuts to speed up processes. However, it manages to significantly improve through detailed instructions and feedback.
One of the most innovative aspects of Manus is its “Manus’s Computer” interface, which offers unprecedented transparency in the AI’s decision-making process. This feature allows users to monitor every action taken by the AI agent in real-time and intervene if necessary.
Technical challenges and current limitations
Despite its advanced capabilities, Manus AI must face significant technical challenges. During tests conducted by MIT Technology Review, the system experienced frequent crashes and timeouts during prolonged usage periods.
Furthermore, some system errors have reported service loads that are too high, suggesting that the IT infrastructure necessary for the optimal functioning of the AI agent is still a critical point.
Another limitation is the extremely restricted access to the platform. Less than 1% of the users on the waiting list have received an invitation to test Manus, while its Discord server already has over 186,000 members.
From an economic standpoint, the operating cost of Manus remains relatively competitive: each task execution costs about 2 dollars, according to the Chinese technology publication 36Kr.
Strategic collaboration with Alibaba Cloud
To tackle infrastructure challenges and expand the reach of Manus, Butterfly Effect has announced a partnership with Alibaba’s cloud computing sector.
According to the South China Morning Post, Manus will collaborate with the Qwen team of Alibaba to adapt to the needs of the Chinese market. Even though the implementation schedule has not yet been defined, the goal is to make Manus available on data processing platforms and artificial intelligence domestic models.
The advance of China in AI models
The collaboration with Manus comes at a time of profound innovation for Alibaba in the field of artificial intelligence. On March 6, the company announced the launch of the new model QwQ-32B, capable of competing with OpenAI o1-mini and with DeepSeek R1, one of the most powerful open-source models of the moment.
According to CNN Business, Alibaba stated that QwQ-32B offers superior performance compared to most competing models with a reduced number of parameters. The model uses only 32 billion parameters, much less than the 671 billion of DeepSeek R1, thus reducing processing costs and improving efficiency.
Artificial Intelligence at the center of the Chinese strategy
The development of Manus AI and Alibaba’s advancements in foundational models reflect a broader strategy of China for the artificial intelligence sector. The Chinese government has expressed clear support for “emerging and future industries”, placing particular emphasis on AI, quantum computing, and robotics.
In the next three years, Alibaba will invest approximately 52.4 billion dollars in AI infrastructure and cloud computing, an amount greater than the company’s investments in these sectors over the last ten years.
As highlighted by the MIT Technology Review, Chinese companies are not simply copying what their Western competitors are doing, but are forging their own path in the adoption of autonomous AI agents.
Manus AI is the demonstration that China is evolving beyond the simple replication of Western innovations, thanks to targeted investments, government policies of technological self-sufficiency, and an increasingly specialized academic ecosystem.
Artificial intelligence is no longer an exclusive technology of the West: we are entering an era in which different AI philosophies coexist, offering optimized solutions for diverse cultural and application contexts.
English (US) ·