Enterprise Knowledge Search
Guangdong needed to modernize investment-attraction outreach with an assistant that could answer policy and resource questions accurately, cut the cost of manual consultation, and show a reproducible public-sector path for large-model adoption.
Reported impact
Guangdong strengthened the technical foundation beneath its AI招商 service, improved the speed and precision of resource and policy consultation, and reinforced the project as a reference implementation for public-sector large-model assistants.
This parallel 'Investment Guangdong AI Assistant' implementation emphasizes Baidu AI Cloud's Qianfan and BaiGe stack as the technical base for Guangdong's province-level招商 service. The department used Wenxin large-model capabilities to rebuild the UNIT dialogue experience and to connect structured resource data such as parks, plants, buildings, projects, land parcels, and events with unstructured policy documents. BaiGe heterogeneous compute underpins the large-model workload, while the assistant exposes structured matching, policy interpretation, summarization, and search as one public-facing workflow. In the Heyuan pilot, full local resource and policy datasets were used for learning, training, and fine-tuning, allowing the assistant to answer natural-language investor questions and to surface the right resource or policy result more efficiently than traditional web or relationship-based招商 processes.
Guangdong's resource inventories and policy documents are ingested into the assistant's shared knowledge layer, where Qianfan models running on BaiGe process the content for matching, summarization, and Q&A. Investor questions flow through the upgraded UNIT dialogue interface and return grounded resource or policy answers through the assistant.
Heyuan completed training and launch within two months, with current inquiry accuracy above 90% (original: 2 个月内完成了模型训练和上线…现阶段问询准确率达 90% 以上).
The AI assistant uses the Qianfan platform and the BaiGe heterogeneous compute platform (original: 千帆大模型平台；百舸异构计算平台).
The assistant can answer questions over structured investment resources and over unstructured policy documents with natural-language generation (original: 现阶段具备招商资源结构化数据的精准匹配问答；具备招商政策文本文档非结构化数据的自然语言问答和生成).
The project rebuilt UNIT with Wenxin capabilities and integrated Guangdong's massive resource and policy data (original: 依托百度文心大模型，对百度 UNIT 智能对话平台进行重构升级，集成广东省海量资源与政策数据).
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
Baidu AI Cloud Sample Sep 24, 2025 All-China Federation of Trade Unions The All-China Federation of Trade Unions needed a unified AI foundation that could support legal aid, rights protection, and other worker services at national scale instead of leaving union service digitization… Delivery cycle fell from 3-6 months to under 1 month. Qianfan AppBuilder · Qianfan ModelBuilder · Union Industry Model Baidu AI Cloud Sample Jun 19, 2025 Nanwei Software Co., Ltd. Nanwei needed a better way to help Fengning government users explore and analyze operational data because dashboards and cockpits could only display fixed views, and custom cross-team analysis cycles were too slow for… Baidu Jiuzhou Government Native Products · Qianfan AppBuilder · Qianfan ModelBuilder Baidu AI Cloud Sample Jun 19, 2025 Xi'an Public Security Bureau Traffic Management Department (西安市公安局交通管理局) Xi'an traffic police needed a scalable way to handle rising traffic-service inquiries, relieve hotline pressure, remove information silos between frontline officers and functional departments, and standardize both… Answer accuracy exceeded 96%. Qianfan AppBuilder · Qianfan ModelBuilder · Yong'anjun 