Sinolink: Manus triggered the imagination space of Universal Agent, focusing on breakthrough innovations in AI application end.

Sinolink released a research report stating that Manus' application innovation based on large model performance upgrades has sparked capital market discussions about universal Agents. Although projects of this nature are still in the early stages, they have driven the modularization of toolchains (such as Ubuntu container operations) and scene expansion, proving that the industry's core barrier lies in engineering capabilities rather than single model performance. From a business model perspective, internet giants can embed existing applications or cloud platforms in the form of MaaS with the advantage of base models and computing power, making it easier to achieve scalable monetization; vertical industry enterprises have richer customer resources and industry know-how, and are expected to deepen scene-based Agents. It is recommended to pay attention to the domestic leading generative large model company Iflytek Co., Ltd. (002230.SZ); AI hardware is expected to become a new carrier for application landing, and it is recommended to pay attention to Hangzhou EZVIZ Network (688475.SH), ArcSoft Corporation (688088.SH), etc.; AI-related feature polishing can drive up C-end application monthly active users and payment rates, it is recommended to pay attention to Beijing Kingsoft Office Software, Inc(688111.SH), Wondershare Technology Group (300624.SZ), etc. Key points from Sinolink are as follows: This week, Monica.im released the world's first universal Agent product, Manus, which surpassed OpenAI and ranked first in the GAIA (Universal AI Assistant) evaluation. It can achieve full automation of the process "understand needs-analyze problems-automatically execute-deliver usable results". In the showcased Case, Manus supports functions such as travel planning, stock/business data analysis, animation/game generation, real estate selection, classroom assistance, workflow optimization, etc. Manus adopts a multi-agent collaborative architecture and utilizes multiple model collaborations to achieve task decomposition, tool calling, and path correction. It is currently in the internal testing phase. Manus is an application innovation breakthrough based on large model performance upgrades, sparking capital market discussions about universal Agents. Teams such as MetaGPT have quickly launched open-source versions (such as OpenManus, OWL), significantly reducing development thresholds by reverse engineering task disassembly logic. Although projects of this nature are still in the early stages, they have driven the modularization of toolchains (such as Ubuntu container operations) and scene expansion, proving that the industry's core barrier lies in engineering capabilities rather than single model performance. From a business model perspective, internet giants can embed existing applications or cloud platforms in the form of MaaS with the advantage of base models and computing power, making it easier to achieve scalable monetization; vertical industry enterprises have richer customer resources and industry know-how, and are expected to deepen scene-based Agents. It is recommended to focus on core targets in office, enterprise services, industrial software, and network security fields. In terms of the prosperity of specific industries, the bank believes that industries with high Prosperity and maintaining or upward trends in AI industry chain (computing power is at a high level of prosperity maintenance, applications are accelerating in terms of prosperity), intelligent driving (such as laser radar, vehicle-road cloud, etc.), C-end stock trading software (if the market can maintain a trading volume of over a trillion), Huawei industry chain, low-altitude economy, quantum, etc. are good and sustainable in terms of prosperity. Data elements, power IT, EDA, going abroad, and other industries are moderately good and sustainable in terms of prosperity, with an upward trend in prosperity, and strong possibilities that are better in areas such as new information technology, military industry IT, emergency government IT, with the potential for prosperity to increase, but the strength is uncertain, progressing upward by quarter, depending on the landing rhythm of the cyclical industry is security, network security, industrial software, capital market IT, etc., industries with weaker prosperity may rebound but with uncertain strength, with low prosperity and gradual upward trends, key industries include security, network security, industrial software, capital market IT, etc., and industries with weak prosperity are likely to be construction and real estate IT, medical IT, etc. Risk Warning: Risks of intensified industry competition; Risks of technological research and development progress falling short of expectations; Risks of cyclic fluctuation in downstream capital spending in specific industries.