Workflow
AI商业模式
icon
Search documents
周鸿祎谈Manus:广告模式不work了,要向用户直接收费
Xin Lang Ke Ji· 2025-07-23 04:02
周鸿祎认为,现在AI的运营成本很高,用户对智能体的要求越高,完成任务越多,我们过去跟ChatGPT 聊天的时候消耗token是很少的,现在智能体实际上替你完成一个任务都要花几百万个token,你跟人问 答案哪花得了几百万个token,几千个token就了不起了,所以这种广告模式就不work了,就要向用户直 接收费。 "2025(第二十四届)中国互联网大会"于7月23日-25日在北京举行。360公司董事长周鸿祎谈及近日 Manus将国内业务转移至海外这一话题时表示,Manus的问题比较复杂,我对他们也不是很了解,所以 只能简单的说,可能在国外 AI起来之后,商业模式发生了变化,跟互联网最大的不一样,互联网过去 是这样,运营成本很低,所以只要有足够多的用户,Dau足够高,就可以有广告模式,羊毛出在猪身 上。 他总结:中国的付费习惯可能还是比欧美市场可能要差一些,但是我觉得这是一个过程,整个互联网如 果大家都不要太内卷,现在你比如说你想收费巨头都免费对吧?所以对Manus这种公司来说,可能它既 然融了资,它就要有收入。所以可能还是商业模式变现的问题。但是,我在考虑国内还有一个巨大的市 场。我觉得Manus的思路是自 ...
Manus免费开放注册:下一个AI爆发时刻来袭?
3 6 Ke· 2025-05-14 03:48
Core Insights - Manus has opened free registration, which aligns with the recent trend of "periodic silence and sporadic bursts" in AI news, leading to a surge in related discussions [1][3] - The product, previously controversial due to its "hunger marketing" strategy, is now accessible to all users, offering initial rewards and daily free points for usage [3][4] - Manus differentiates itself from other AI search products by requiring points for task execution, which may limit user engagement [3][4] Product Features - Manus is designed to integrate information effectively, showcasing a structured workflow that breaks down tasks into manageable components [5][17] - The platform emphasizes source tracing within its workflow, aiming to reduce AI hallucinations, a notable criticism of competitors like DeepSeek [5][9] - Users can generate comprehensive reports and detailed task lists, although the speed of processing remains a concern [14][16] Market Positioning - Manus is positioned as a pioneering general-purpose AI agent, indicating a shift in the AI landscape towards collaborative and outcome-driven models rather than just tool provision [20][21] - The recent AI summit highlighted a new consensus that the value of AI applications will surpass the models themselves, focusing on deliverables and results [20][21] - Manus's approach reflects a broader industry trend towards task planning and execution frameworks that leverage multiple intelligent agents [25][24]