Core Viewpoint - The "Houxi" AI government service project in Fujian Province has been recognized as a benchmark in the AI governance sector, showcasing significant application effectiveness and innovation value, and has been included in the provincial-level AI centralized signing list, marking a high level of recognition at the provincial level [1][2]. Group 1: Project Overview - The "Houxi" AI government service intelligent agent is the first in Fujian to achieve full coverage of intelligent consultation, real-time processing, and assisted approval functions, and is recognized as a national innovation in government service AI [2]. - The project has received multiple honors and is included in significant reports, providing valuable reference cases for AI applications in the national governance sector [2]. Group 2: Technical Support - The project is based on the YanYun Infinity platform, a result of collaboration between the company and Peking University, which integrates various AI and data capabilities to overcome data barriers and link government data effectively [3]. - It utilizes deep learning on 150,000 government service policy documents and over 4 million words of local knowledge to create a "government brain" capable of precise service [3]. Group 3: Service Model Innovations - The "Houxi" project introduces a nationwide first "search, ask, and process" integrated intelligent service model, simplifying traditional multi-step processes into a single dialogue flow [4]. - It features a "non-intrusive" collaboration model that allows cross-departmental data sharing without altering existing government system architectures [4]. - The project includes voice interaction capabilities that support both Mandarin and local dialects, enhancing user experience [4]. Group 4: Application Effectiveness - Since its launch, the "Houxi" project has significantly improved service efficiency, providing 24/7 service equivalent to 1,000 staff members, addressing common issues faced by citizens [5]. - The project has reduced form filling by 74% and application materials by 26%, streamlining the process for users [5]. - The accuracy of online consultations has reached 100%, and the first-time approval rate for application materials has increased by 80%, optimizing human resource allocation in government services [5]. Group 5: Replicability and Future Prospects - The success of the "Houxi" project has created a replicable model for digital government construction, adaptable to different regional data environments and business processes [6]. - The project’s design allows for local customization based on regional policy characteristics and dialects, lowering the barriers and costs for other regions to adopt similar AI governance solutions [7]. - The company aims to leverage its experience from the "Houxi" project to promote AI governance solutions across more regions, driving the intelligent transformation of government services [7].

Core Points - Shenzhen has accelerated the transformation of its government services from "transactional" to "professional" with the certification of 30 staff members as the first batch of "Government Service Staff" at the highest level [1][2] - The "Government Service Staff" role, officially recognized by the Ministry of Human Resources and Social Security, involves providing consultation, material review, and service guidance in government service venues [1] - The certification process included a rigorous assessment of practical skills, focusing on the ability to handle complex issues and adapt to the "AI + government" model [1] Group 1 - The certification for the advanced level (Level 3) had high entry requirements, including relevant educational backgrounds or a minimum of 10 years of experience in related fields [2] - The talent service hall has implemented innovative measures such as one-stop services and has established nine frontline service stations in key areas to enhance accessibility for citizens and businesses [2] - The talent service hall has an annual business volume exceeding 600,000 transactions, demonstrating its commitment to meeting the needs of enterprises and the public [2]

Group 1 - The core viewpoint of the report is that Minsheng Securities has given a "recommended" rating to Bosi Software (300525.SZ) based on several factors [1] Group 2 - Q3 performance is under short-term pressure due to tightened budgets from downstream clients [1] - Gross margin has improved year-on-year, and expense control remains robust [1] - The company is fully embracing AI, with multiple product lines launching AI+ products [1] - Significant policy announcements have been made, providing strong support for AI+ in government affairs [1]

Investment Rating - The report maintains a "Recommended" rating for the company, considering its leading position in the fiscal IT sector and potential benefits from the upcoming digital government construction wave [3]. Core Insights - The company experienced a revenue decline of 7.67% year-on-year, with total revenue reaching 1.148 billion yuan in the first three quarters of 2025. The net profit attributable to shareholders dropped by 78.81% to 0.07 billion yuan [1]. - In Q3 2025, the company reported a revenue of 0.442 billion yuan, down 15.97% year-on-year, and a net profit of 0.049 billion yuan, down 32.45% year-on-year, primarily due to tightened budgets from downstream clients [1][2]. - The overall gross margin improved to 65.9%, an increase of 2.1 percentage points year-on-year, attributed to the company's strategy of enhancing the proportion of sustainable revenue [2]. - The company has embraced AI comprehensively, launching multiple AI+ products across various sectors, including public expenditure services and smart healthcare solutions [2][3]. - Recent government policies have provided strong support for AI+ applications in the public sector, which is expected to drive growth in the industry [3]. Financial Forecasts - The company is projected to achieve net profits of 0.323 billion yuan, 0.380 billion yuan, and 0.433 billion yuan for the years 2025, 2026, and 2027, respectively. The corresponding PE ratios are estimated to be 32, 27, and 24 times [3][4]. - Revenue forecasts for the upcoming years are as follows: 2.229 billion yuan in 2025, 2.581 billion yuan in 2026, and 2.905 billion yuan in 2027, with growth rates of 4.0%, 15.8%, and 12.5% respectively [4][9].

Investment Rating - The report maintains a "Recommended" rating for the company [4][3]. Core Insights - The company reported a significant recovery in Q3 2025, with revenue reaching 5.04 billion yuan, a year-on-year increase of 7.36%, and a net profit attributable to shareholders of 0.39 billion yuan, up 42.33% year-on-year [1]. - The company is actively upgrading its AI product matrix, having launched the industry's first Agentic platform, which aims to enhance efficiency and reduce costs for tax service institutions [2]. - Strong policy support for AI in government sectors is expected to drive growth, as new guidelines have been issued to promote the deployment of AI models in public administration [2]. Financial Performance Summary - For the first three quarters of 2025, the company achieved a total revenue of 14.26 billion yuan, an increase of 11.10% year-on-year, but the net profit attributable to shareholders decreased by 4.74% to 1.1 billion yuan [1]. - The company’s expenses for sales, management, and R&D for the first three quarters were 2.57 billion yuan, 1.70 billion yuan, and 3.37 billion yuan, reflecting increases of 20.1%, 13.6%, and 3.4% respectively [1]. - The projected net profits for 2025, 2026, and 2027 are 2.02 billion yuan, 3.00 billion yuan, and 4.29 billion yuan, with year-on-year growth rates of 80%, 48%, and 43% respectively [3]. Valuation Metrics - The current market capitalization corresponds to a price-to-earnings (PE) ratio of 105 for 2025, 71 for 2026, and 50 for 2027 [3]. - The projected earnings per share (EPS) for 2025, 2026, and 2027 are 0.50 yuan, 0.74 yuan, and 1.06 yuan respectively [3]. Product Development - The company continues to enhance its AI product offerings, focusing on a new product matrix that includes a cockpit, smart accounting, and smart compliance advisory, aiming to significantly improve operational efficiency for tax service providers [2].

Core Viewpoint - Shenzhou Holdings has launched the "Houxie" AI government assistant in Zhangzhou, marking a significant advancement in AI-driven public service capabilities in China, with full coverage of intelligent consultation, real-time processing, and auxiliary approval functions [1][2]. Group 1: AI Government Assistant "Houxie" - "Houxie" is the first AI assistant in Fujian, designed to enhance government service efficiency and user experience, embodying the meaning of "things done well" [1]. - The assistant utilizes Shenzhou Holdings' "Yanyun Infinity" platform, which integrates various data sources and employs deep learning to analyze over 150,000 government service policy documents and a local knowledge base of over 4 million words [1][2]. - "Houxie" addresses common issues in public service, such as inefficient consultation and repetitive processes, by introducing intelligent pre-examination features that streamline the service experience from multiple visits to a single completion [1][2]. Group 2: Technical Infrastructure and Solutions - The successful implementation of "Houxie" relies on Shenzhou Holdings' end-to-end integrated solutions, which include AI infrastructure, data intelligence platforms, and industry-specific applications [2]. - The "Yanyun Infinity" platform, co-developed with Peking University, has upgraded its capabilities to overcome data silos and integrate diverse government data, facilitating effective linkage of policy documents, processes, and electronic certificates [2][3]. - The platform employs a no-code business modeling approach to create digital twins of real-world government processes, providing practical training scenarios for AI models [2][3]. Group 3: Future Prospects - Shenzhou Holdings aims to replicate the "Zhangzhou model" in other regions, promoting the intelligent upgrade of government services through data intelligence technology [3]. - The company is focused on enhancing the convenience of smart governance for more citizens and businesses, contributing to the development of digital government initiatives [3].

Investment Rating - The report maintains a positive outlook on the computer industry, particularly in the context of AI applications in government affairs, with key targets identified for investment [3][7]. Core Insights - The issuance of the "Guidelines for the Deployment and Application of AI Foundation Models in Government Affairs" by the Central Cyberspace Affairs Commission and the National Development and Reform Commission is expected to accelerate the demand for AI in governance [3][7]. - The report highlights the broad market potential for AI applications in government, with the AI+government service market in China projected to reach 2.54 billion RMB in 2024 [3][7]. Summary by Sections Investment Recommendations - Key investment targets include Guotai Epoint Software, Hikvision, Zhejiang Dahua Technology, Beijing eGOVA, Geovis Technology, and Piesat Information Technology, with related targets being Fujian Boss Software, Zhiyang Innovation Technology, Linewell Software, and CHINA INFORMATION DEVELOPMENT INC. [3][7]. Guidelines Overview - The guidelines emphasize the use of "East Data West Computing" and integrated computing networks, promoting a unified model reuse approach and enhancing data governance to support AI model training [3][7]. - Four high-frequency scenarios for AI model application are identified: government services, social governance, office work, and decision support [3][7]. Financial Support and Market Potential - The guidelines propose strengthening financial support for AI applications in government, indicating a broad market potential for AI+government applications [3][7]. - The report notes that since 2025, domestic AI models like Deepseek have seen increased engagement from various regions and departments, indicating a positive trend in technology deployment and system development [3][7].

Core Viewpoint - The issuance of the "Guidelines for the Deployment and Application of Artificial Intelligence Large Models in the Government Sector" by the Central Cyberspace Affairs Commission and the National Development and Reform Commission is expected to accelerate the implementation of AI in government applications in China [2][3]. Group 1: Guidelines Overview - The guidelines provide a framework for government departments to deploy AI large models, emphasizing the need for a unified approach to infrastructure and data governance [2][3]. - The guidelines advocate for a model reuse strategy to avoid isolated models and enhance data quality for effective AI training [2][3]. Group 2: Application Scenarios - Four high-frequency application scenarios are identified: government services, social governance, office administration, and decision support [3]. - Specific applications include intelligent Q&A, auxiliary processing, policy service delivery, smart monitoring, and emergency response [3]. Group 3: Market Potential - The guidelines highlight the broad market potential for AI in government applications, with a projected market size of 2.54 billion RMB in 2024 [3]. - The emphasis is on securing funding for AI deployment in the government sector and promoting the integration of domestic AI technologies [3].

Investment Rating - The report maintains a "Cautious Recommendation" rating for the industry [6] Core Insights - The recent policy document issued by the Central Cyberspace Administration and the National Development and Reform Commission provides strong guidance for the AI+ government sector, outlining 13 application scenarios and emphasizing funding support and innovative business models [4][21] - The AI+ government industry is expected to thrive with several leading companies launching new products and applications, such as the AI billing system by Boshi Software and the AI government service assistant by New Point Software [13][21] - The release of DeepSeek V3.2 and its adaptation by domestic chip manufacturers like Cambricon and Ascend signifies a closed loop in domestic AI capabilities, integrating computing power, models, and applications [4][21] Summary by Sections Market Review - During the week of October 9-10, the CSI 300 index fell by 0.51%, the SME index decreased by 1.03%, and the ChiNext index dropped by 3.86%. The computer sector (CITIC) declined by 2.04% [31] Industry News - AMD and OpenAI have entered a four-year chip collaboration, leading to a significant stock price increase for AMD [23] - The Ministry of Industry and Information Technology and six other departments released a plan to promote the integration of "5G + industrial internet" with AI [27] Company News - Soran Technology announced plans to acquire a 55% stake in Kunyu Lancheng Technology for 74.25 million yuan, with a valuation of 135 million yuan for the target company [29] - ST Guohua disclosed significant litigation progress, with a court ruling in its favor regarding a case involving shares and cash [30] Weekly Insights - The AI+ government sector is poised for growth due to robust policy support and the emergence of innovative applications. Key companies to watch include iFLYTEK, Deepin Technology, and Boshi Software [4][21] - The domestic computing power industry is also highlighted, with companies like Cambricon and Inspur being key players [22]

Core Viewpoint - The Jiangsu Emergency Management Bureau's "Ning'anqing" model has been recognized as an excellent practice in the construction of a strong digital province, showcasing advancements in AI integration within emergency management [1][2] Group 1: Model Development and Features - "Ning'anqing" is the first vertical emergency management model in China, developed in collaboration with the Jiangsu Data Bureau and Qingtian Technology, utilizing the DeepSeek-R1-671B model [1] - The model addresses issues such as knowledge isolation, data fragmentation, and inefficient decision-making in emergency management [1] - It integrates over 200,000 pieces of data, including regulations, emergency plans, and case studies, to create an intelligent governance model for risk identification and emergency response [1] Group 2: Operational Efficiency and Applications - The model leverages natural language processing technology and over 6 million emergency knowledge points to generate official document drafts, significantly enhancing administrative efficiency [2] - It consolidates multi-dimensional data sources into a comprehensive risk information pool, allowing for precise risk prevention and control measures [2] - "Ning'anqing" facilitates cross-departmental inspections and risk assessments, promoting standardized enterprise inspections and more effective emergency decision-making [2] Group 3: Future Development and Goals - The model is widely applied within the emergency system and aims to further integrate data and multi-modal technologies for customized and professional applications in risk assessment and disaster warning [2] - The focus will be on advancing digital governance and intelligent transformation in urban safety management [2]