For stock trading, check the Jin Qilin analyst research reports - authoritative, professional, timely and comprehensive, helping you uncover potential thematic opportunities! Source: China Fund News. Introduction: Brokerages collectively move onto third-party AI platforms as the battle for position in the Agent era begins. By Sun Yue, China Fund News reporter. From integrating DeepSeek, to launching Skills, to recently flocking onto third-party AI platforms, the securities industry is embracing external AI ecosystems faster than most people expected.
According to an incomplete tally by the reporter, more than 10 brokerages have now connected to third-party AI platforms, covering Kimi, WorkBuddy, Qwen, Volcano Engine and others. Interestingly, many large brokerages did not choose "either/or" but instead connected to multiple platforms simultaneously. "Different platforms have different capability strengths, and we will select the best one based on the scenario," said a person in charge of information technology at CITIC Securities Company Limited (ASX: 600030). A year ago they were competing on self-developed large models; why are they now moving their own AI capabilities into third-party platforms?
In the view of industry insiders, expanding new user touchpoints is the primary consideration - by outputting investment research, market quotes and other capabilities through Skills or agents, brokerages can reach more potential clients. For financial institutions, traffic competition is already an unavoidable battlefield, and connecting to the mature ecosystems of leading platforms is one of the simplest and most direct paths. The "positioning battle" for brokerage AI Agent deployment has begun. Brokerages are accelerating their embrace of external large models and the Agent ecosystem.
In early September, Tencent's WorkBuddy open platform officially launched, and GF SEC (ASX: 1776) became the first brokerage institution to join the ecosystem zone; the two sides jointly released the "GF Securities" Buddy application zone, with 12 self-developed Skills and 9 expert capabilities launched in the first batch. In late September, the six major AI skills of China Securities' self-developed Dragonfly Skill went live on the WorkBuddy skill plaza. On September 7, Alibaba's Qwen open platform launched more than ten financial agents, allowing professional AI services from different financial institutions to be accessed through a single entry point.
Four brokerage agents were among the first to join: Guotai Haitong Securities Co., Ltd. (ASX: 601211) "Lingxi," Industrial Securities' intelligent investment assistant, Soochow's "Xiucai," and CICC Wealth. Guotai Haitong stated that this cooperation marks the official integration of brokerage professional investment services into everyday public AI scenarios. In mid-September, Moonshot released the Kimi financial industry AI solution, entering the core business processes of financial institutions with a systematic architecture of "authoritative data sources + professional skills + security compliance," and brokerages including China Securities Co., Ltd. (ASX: 601066) and CICC have already implemented or co-built it in their businesses. Even earlier, Baidu, Alibaba and Volcano Engine had all opened agent capabilities to the financial industry.
Brokerages joining third-party AI platforms have mainly formed two paths: one leans toward professional investment research capabilities and tool output, namely joining AI office workbenches such as Tencent WorkBuddy, mainly listing in Skills form; the other focuses on service reach for a broader range of investors, namely joining general AI assistant platforms such as Alibaba Qwen, where brokerage AI capabilities exist in the form of independent conversational agent Bots, focusing on mass wealth management dialogue services. In the view of industry insiders, brokerages joining third-party AI platforms is an innovation in the securities financial service model.
China Securities stated that, taking joining WorkBuddy as an example, the company's professional financial investment research capabilities have been transformed into a universally accessible "mass dialogue flow," completely breaking down the barriers to professional financial resources, allowing users to access institutional-grade professional data capabilities on a general Agent platform. On the other hand, actively connecting to the mature agent ecosystems of leading internet platforms is also expected to bring traffic to brokerages' own Apps. Yang Ling, a securities industry analyst at Analysys Qianfan, said that expanding new user touchpoints is the primary consideration for brokerages' layout.
By outputting investment research, market quotes and other capabilities through Skills or agents, brokerages can expand their reach to potential clients. In the future, users may more often directly raise investment needs through AI assistants, and brokerages need to enter these platforms in advance to secure a professional service entry point in the new user decision-making chain. She believes that the real value of this model of connecting to third-party AI platforms lies in first building user awareness through professional content and tools, then gradually converting to their own Apps, investment advisory and wealth management services. Whether it can ultimately become an effective channel depends on whether a conversion closed loop can be formed between external traffic and the brokerage's own client system.
Beyond cooperation at the AI capability level, business collaboration between the Tencent ecosystem and brokerages is advancing in depth. On September 15, Tencent Marketing disclosed that in 2026 it had paved a full "account opening customer acquisition" chain with brokerages - more than 40 traditional brokerage headquarters have placed customer acquisition campaigns during the year, with budgets at leading brokerages reaching the tens of millions of yuan level, and branches and business departments also accelerating their entry. Self-developed and external agents complement each other. Brokerages densely connecting to external ecosystems does not mean giving up self-development.
"Our overall strategy is centered on a self-developed foundation, with external ecosystems as capability supplements," said a person in charge of information technology at CITIC Securities. Introducing external agent ecosystems is not about handing over core capabilities to third parties, but more about technical benchmarking, capability reference, and introducing some external Skills as needed to enrich the skill material library within the company's platform. At the platform cooperation level, CITIC Securities stated that at the current stage the company continues to conduct technical evaluation and capability verification of multiple mainstream domestic agent platforms, introducing some external skills as needed for scenario testing and capability supplementation.
"The first criterion in our selection is data security, isolation capability and compliance traceability," the person pointed out. This positioning directly responds to the concerns of business colleagues - if a third-party public cloud skill platform is chosen, internal business data cannot be securely accessed; but conversely, if a unified internal skill platform is not used, it is difficult for employees' individual work experience to accumulate, and overall business efficiency is hard to continuously improve. "Different platforms have different capability strengths, and we will select the best one based on the scenario. External skills are only supplementary material; the real core asset accumulation and co-building and sharing of the enterprise skill library all happen on the internal Harness platform," the person said.
In the view of GF Securities, brokerage AI and general large models are not an either-or opposition. Judging from the current implementation path of brokerage AI applications, most are fine-tuning or post-training based on general large models, combined with brokerages' unique application scenarios to provide services to employees or clients. Therefore, the knowledge foundation of brokerage AI is built on proprietary domain data such as brokerage market quotes, trading, positions, investment research and regulatory standards, superimposed with the research frameworks and business experience accumulated by institutions over the long term. In addition, general large models mostly exist in an independent Q&A form, delivering information and content; brokerage AI must be embedded in real business processes, operating together with permission systems, risk control rules, compliance review and responsibility chains, delivering executable, traceable and accountable business results.
Xin Zhiyun, Vice President and Chief Information Officer of GF Securities, said the industry is entering a new stage of AI agent development, and differentiated data assets, self-controllable AI technology and all-scenario implementation capability will become key levers in brokerages' long-term competition. GF Securities has formulated an overall AI layout blueprint, and on the basis of building the three core elements of computing power, data and algorithms, it focuses on planning in thematic areas such as AI investment advisory, AI investment research and AI investment banking. Technology competition has shifted from an "arms race" to capability differentiation. Under the technology wave, brokerages' real-money investment continues to increase.
From the 2025 brokerage annual reports, Guotai Haitong Securities ranked first with an investment scale of 3.235 billion yuan, Huatai Securities invested as much as 2.679 billion yuan, and the information technology investment scale of brokerages including China Merchants Securities, China Securities, CICC, GF Securities, China Galaxy, Guosen Securities and Shenwan Hongyuan also exceeded 1 billion yuan. But the question is, with investment-output results not yet fully clear, is this "arms race" style of investment sustainable? CITIC Securities believes that industry AI competition is shifting from comparing large model parameters to comparing AI large-scale implementation and enterprise knowledge asset governance capabilities.
Its investment strategy adheres to value orientation, prioritizing scenarios that can quantitatively improve efficiency and front-load control of business risks, and does not make purely conceptual investments. "The efficiency improvement of human-machine collaboration is more sustainable than simple manpower expansion," CITIC Securities stated. Facing the objective situation of talent flow in the industry, capacity built purely by piling on people is fragile. Only by transforming individual experience into intangible core assets that the company can permanently reuse can stable and sustainable organizational productivity be formed. From the perspective of the industry landscape, CITIC Securities judges that future investment will continue to differentiate: leading institutions have the ability to build self-developed platforms, accumulate company-level skill assets and solve large-scale implementation challenges; small and medium-sized institutions will more directly connect to external ecosystems for lightweight pilots, and there will not be an indiscriminate "arms race" across the entire industry.
The value of AI investment, in the short term, is work-hour savings; in the long term, it is building the core foundation that carries business data and enterprise skill assets. GF Securities believes that AI's reconstruction of brokerage content services is not limited to cost reduction and efficiency improvement, but also pushes content services from a cost center to a growth engine, from an auxiliary tool to a core capability, and from traffic operations to long-term client management, helping brokerages build a differentiated moat in the wealth management transformation and complete the strategic upgrade from "content services" to "intelligent services." Sina statement: This news is reprinted from a Sina partner media outlet. Sina publishes this article for the purpose of conveying more information and does not mean it agrees with or confirms its views or descriptions. The article content is for reference only and does not constitute investment advice. Investors operate at their own risk based on this.