Manus launched | Expert: The difference with DeepSeek is to solve "how to do it," but its effectiveness awaits practical testing

Following DeepSeek, the world's first general artificial intelligence software (AI Agent) developed in mainland China, Manus, has recently gone online. This product has demonstrated performance surpassing that of OpenAI's comparable large models in the GAIA benchmark test, and within four hours of its release, its official website traffic even exceeded ten million. The overnight success of Manus has amazed the market, but it has also faced continuous controversy and has been questioned as a "shell" product. However, industry experts point out that Manus and DeepSeek are two different types of AI software, with Manus aimed at solving "how to do it specifically," pushing the technology of intelligent agents forward significantly. Manus aims to solve "how to do it specifically" Manus is referred to as the world's first AI intelligent agent. According to information released by the development team, Manus bridges the gap between concept and execution, "while other artificial intelligences are only generating ideas, Manus delivers results." Demonstration videos released by the development team show that Manus can complete tasks such as resume screening, real estate research, and stock analysis. Related news: Manus AI goes online | China develops the world's first AI Agent that can directly write PPTs, analyze stocks, etc. However, since its launch, Manus has been surrounded by controversy. Due to the invitation-only internal testing restrictions, most people, including industry insiders, have been unable to experience the innovative capabilities of this AI Agent, leading to accusations of "hunger marketing" in the industry. The mainland's "Science and Technology Innovation Board Daily" has also suggested that Manus does not have its own research model but instead uses the currently strongest programming development models, Claude and Alibaba's Tongyi model, and further trains them, with its functions being a combination of multiple applications. The highlights of Manus lie in engineering innovation: technological integration (essentially a "shell"), and smooth interaction (a simple and user-friendly UI interface). Related news: Manus AI | Invitation codes are being sold online for ¥100,000, sellers claim authenticity and do not negotiate prices Wei Wei, a professor at the School of Computer Science and Technology at Huazhong University of Science and Technology, stated in an interview with China News Service, "Manus and DeepSeek are not the same thing." DeepSeek is a reasoning language model, with its core being information generation and processing, excelling at answering "what should be done"; Manus is a general-purpose AI intelligent body that primarily utilizes large models or various tools to complete tasks, focusing on solving "how to do it specifically," with the core being task allocation and collaboration. Wei Wei pointed out that using AI intelligent bodies to complete different tasks is not a first for Manus. However, Manus's innovation lies in turning multi-agent collaboration into a product, integrating the most advanced AI capabilities into a seamless user experience, allowing for more complex tasks to be completed in a simpler way, thus pushing the technology of intelligent bodies forward significantly. The effectiveness remains to be tested in practice Related news: Manus launched | The founder Xiao Hong behind it adheres to the philosophy of "less structure, more intelligence." Wei Wei stated that based on the current video demonstration, Manus is indeed very intelligent. However, it is currently limited to internal testing, lacking comprehensive evaluations from technical experts and general user feedback, so its specific effectiveness still needs to be tested in practice. Pan Helin, a member of the Expert Committee on Information and Communication Economy of the Ministry of Industry and Information Technology, believes that although Manus, as an AI intelligent body, does not innovate in basic algorithms, it achieves remarkable results by innovating through engineering applications and mixing various large models to achieve a comprehensive experience