在刚刚落下帷幕的2026年中国国际服务贸易交易会（下称“服贸会”）上，企业级AI Agent基础设施平台“词元无限”携面向金融行业的“数字劳动力”解决方案，在“数字友好解决方案”展区首次亮相。
本届服贸会“数字友好解决方案”展区以“源自北京、服务世界”为目标，由北京市经济和信息化局指导，北京数字经济企业出海创新服务基地承办，共汇聚全市40余家行业龙头与专精特新标杆企业，集中展示了全场景数字人平台、国产万亿级大模型、低空飞行管控系统及智能机器人矩阵等数字服务成果，并将数字技术、标准体系与服务模式整合为一体化出海解决方案。
据介绍，当前金融行业对人工智能的关注，正从“能不能用”转向“能不能规模化、安全可控地进入核心研发生产流程”。面对核心代码与数据留域、存量系统复杂、研发规范严格、质量责任可追溯等要求，金融企业需要的已不仅是能力更强的通用Agent，而是真正掌握在企业手中的“主权Agent”。
词元无限认为，主权Agent的核心，是企业能够掌握人工智能进入生产系统后的技术选择权、安全控制权和持续进化权，具体表现为三个特征：全自主，即企业保有模型选择权和技术路线控制权，避免关键能力受制于单一模型或外部平台；高安全，即Agent行为可治理、过程可追踪、结果可审计，核心代码与数据不脱离企业安全边界；自进化，即企业知识能够持续沉淀，Agent能力随业务实践不断迭代，逐步形成企业自身的智能资产。
基于主权Agent能力，词元无限提出了“数字劳动力生产体系”，推动人工智能从研发辅助工具走向可进入真实生产流程的数字化生产力。在词元无限看来，AI不仅需要生成代码，还要理解企业真实系统、协同执行任务、完成测试验证，并在全过程中形成清晰的责任链条；能够解释“为什么这样写、过程如何发生、最终由谁负责”，才具备进入金融关键研发流程的资格。
本届服贸会期间，词元无限重点展示了覆盖金融研发全生命周期的智能化产品矩阵，贯通需求、设计、开发、测试、交付与运维等环节。方案支持私有化部署、代码与数据留域及国产算力适配，并兼容多模型、多工具统一接入。
真实场景验证方面，在与神州信息的合作中，词元无限在“零组织调整、零流程侵入”条件下引入InfCode平台，需求澄清时间减少70%，代码可用率达88%以上，综合提效39%、成本降低20%。目前，该平台已服务金融、通信、工业等领域近50家头部客户。
本届服贸会上，词元无限CEO杨萍分享了金融企业如何从AI Coding走向数字劳动力、金融核心生产系统为何需要主权Agent等议题，探讨人工智能进入关键研发流程需要解决的安全、质量、控制权与责任问题。
面向海外市场，词元无限正依托广西算力集群深耕东盟市场。借助北京数字经济企业出海创新服务基地的出海服务体系，其主权Agent数字生产力方案将获得从海外商机对接、跨境项目撮合到本地化落地支持的全链条服务保障。词元无限表示，会和数字产业上下游伙伴一道，推动以主权Agent构建数字劳动力的规模化落地，让智能技术转化为自主可控、可验证、可持续进化的组织生产力。
本文来源：https://finance.eastmoney.com/a/202609143872923820.html
若有版权问题，请来信告知：13951431913#139.com 或在右方提交评论。
At the recently concluded 2026 China International Fair for Trade in Services (CIFTIS), the enterprise-level AI Agent infrastructure platform "Infinite Word" made its debut in the "Digital Friendly Solutions" exhibition area, showcasing its "Digital Labor Force" solution tailored for the financial industry.
The "Digital Friendly Solutions" exhibition area at this CIFTIS, with the goal of "originating from Beijing, serving the world," is guided by the Beijing Municipal Bureau of Economy and Information Technology and hosted by the Beijing Digital Economy Enterprise Overseas Innovation Service Base. It gathers over 40 industry leaders and specialized "new" benchmark enterprises from the city, concentrating on displaying digital service achievements such as full-scenario digital avatar platforms, domestic trillion-parameter models, low-altitude flight control systems, and intelligent robot matrices. It also integrates digital technology, standard systems, and service models into an integrated overseas solution.
According to the introduction, the current focus of the financial industry on artificial intelligence is shifting from "can it be used" to "can it be scaled up, safely and with control, to enter the core R&D production process." Facing requirements such as keeping core code and data within the domain, coping with complex existing systems, strict R&D norms, and traceable quality responsibility, financial enterprises no longer just need more capable general-purpose agents, but rather "sovereign agents" that are truly under enterprise control.
Infinite Word believes that the core of a sovereign agent is that the enterprise can control the technical choice, security, and continuous evolution rights of artificial intelligence entering the production system. This is specifically manifested in three characteristics: full autonomy, meaning the enterprise retains model selection and technology route control to avoid being dependent on a single model or external platform; high security, meaning agent behavior is governable, processes are traceable, and results are auditable, with core code and data not crossing the enterprise security boundary; and self-evolution, meaning enterprise knowledge continues to accumulate, agent capabilities iterate with business practice, and gradually form the enterprise's own intelligent assets.
Based on sovereign agent capabilities, Infinite Word has proposed a "digital labor force production system," pushing artificial intelligence from a R&D auxiliary tool to digital productivity that can enter real production processes. In Infinite Word's view, AI not only needs to generate code but also understand the enterprise's real system, collaborate to execute tasks, complete testing verification, and form a clear chain of responsibility throughout the process. Only by being able to explain "why it is written this way, how the process happens, and who is ultimately responsible" does it qualify to enter the critical R&D process of finance.
During this CIFTIS, Infinite Word focused on displaying an intelligent product matrix that covers the entire lifecycle of financial R&D, bridging requirements, design, development, testing, delivery, and operations. The solution supports private deployment, keeping code and data within the domain, and adapting to domestic computing power, while being compatible with unified access for multiple models and tools.
In terms of real-scene validation, in cooperation with China Systems, Infinite Word introduced the InfCode platform under conditions of "zero organizational adjustment, zero process intrusion." The time for requirement clarification was reduced by 70%, and code usability reached over 88%, achieving a 39% increase in overall efficiency and a 20% reduction in costs. Currently, the platform serves nearly 50 leading customers in finance, communications, industry, and other fields.
At this CIFTIS, Infinite Word CEO Yang Ping shared topics such as how financial enterprises transition from AI Coding to digital labor forces and why financial core production systems need sovereign agents. He discussed the security, quality, control, and responsibility issues that artificial intelligence entering critical R&D processes needs to solve.
Looking toward overseas markets, Infinite Word is deepening its presence in the ASEAN market by leveraging the Guangxi computing power cluster. With the support of the Beijing Digital Economy Enterprise Overseas Innovation Service Base's overseas service system, its sovereign agent digital productivity solution will receive full-chain service guarantees ranging from overseas business matching and cross-border project facilitation to localized implementation support. Infinite Word stated that it will work with partners in the digital industry chain to promote the large-scale implementation of building digital labor forces with sovereign agents, transforming intelligent technology into autonomous, controllable, verifiable, and sustainably evolving organizational productivity.
Original title: "Sovereign Agent" Debuts at 2026 CIFTIS, May Reshape Financial AI Productivity