Document Processing · Enterprise Knowledge Search · Risk Management
Penghua Fund needed to reduce manual work in research, operations, and compliance tasks such as classifying financial events, summarizing announcements and meetings, checking fund-style drift, and answering internal knowledge questions from large document collections.
Reported impact
Penghua Fund accelerated analyst and operations work, raised the quality of internal Q&A, and created a reusable financial AI service layer that can support more business teams without starting from scratch for each use case.
Baidu AI Cloud and Penghua Fund built a financial large-model application layer on top of Qianfan and the BaiGe heterogeneous compute platform. The implementation packaged mature large-model capabilities for text classification, summarization, entity extraction, and document retrieval into eight business scenarios spanning research, operations, and office workflows. Through the Base Zhihui financial APaaS workbench, Penghua created visual prompt templates, tuned business logic, and assembled an internal AI application marketplace with twenty skills. The resulting Penghua Wenxin platform integrates market intelligence, A-share announcement summaries, report briefs, internal rules, and product knowledge so analysts and staff can query a shared knowledge layer, retrieve cited answers, and automate repetitive reading and classification tasks without rebuilding models for every department.
News, listed-company announcements, meeting notes, internal rules, product documents, and research reports are processed through Qianfan services running on BaiGe. Base Zhihui packages the prompts and skills for each scenario, and the Penghua Wenxin platform serves summaries, classifications, and knowledge-grounded answers back to analysts and operations staff through a shared internal marketplace.
Integrating news, A-share announcements, and research reports into the Penghua Wenxin platform saves more than 30% of the time spent collecting information (original: 大幅提升基金经理的资讯获取效率，节省了30%以上的时间).
Peak employee usage of the internal marketplace exceeds 60% (original: 目前公司内部员工使用率峰值超过60%).
Penghua and Baidu built eight scenarios including financial event classification, style-drift judgment, announcement summaries, and internal intelligent Q&A (original: 共建八大业务场景，包括财经事件分类及正负面判断、基金持仓风格偏移判断、企业公告摘要生成、会议纪要摘要生成、内部员工制度智能问答等).
The internal knowledge assistant answers questions with citation links and resolves more than 90% of issues while saving about 40% of search time (original: 每个回答附有引用链接…问题解答率超过90%，相比传统人工搜索，节省了约40%的时间).
Through the Base Zhihui financial APaaS workbench, Penghua built an internal AI application marketplace with 20 skills (original: 通过…基座智汇智能体（金融APaaS平台）…搭建了包含20个技能的智能应用市场).
Review how this reference was scored and trace each normalized claim back to its documented source.
Scoring methodology
Customer Reference Value measures usefulness as a customer reference. Evidence Confidence measures the quality and traceability of available evidence, not independent verification.
Traceability
Field coverage shows which facts each source supports. Dates are shown only when documented by the source.
Related
Mistral Sample Apr 24, 2026 Ardian Ardian needed to analyze large volumes of confidential due diligence material faster, but its teams were overwhelmed by scattered internal documents, NDA constraints, and traditional machine-learning approaches that did… 18,000+ documents since inception GAIA · Mistral models 
In-house Sample May 26, 2025 BNP Paribas BNP Paribas must design and distribute investment products tailored to each retail customer while handling very large volumes of information, complying with stringent regulation, and giving advisers and customers clear… several billion possible combinations algorithm · automated preparation of performance reports · internal conversational assistants In-house Sample May 8, 2024 Mapfre Mapfre wanted to improve customer experience and streamline insurance operations such as claims handling, risk assessment, underwriting support, document management, internal knowledge access, and digital content… 75 use cases under study for the application of generative AI ControlExpert · Cyberwrite · Holistic AI 
