Source: Guangdong Provincial Development and Reform Commission
In recent years, Guangdong has thoroughly implemented the decisions and policies of the CPC Central Committee and the State Council regarding the development of the digital economy and the optimization of the business environment. Using “artificial intelligence + bidding and tendering” as a breakthrough point, the province has vigorously promoted the deep integration of smart regulation and intelligent services in cities such as Guangzhou and Shenzhen—which have well-developed new infrastructure—and has accelerated the practical application of artificial intelligence technology in key stages of the bidding and tendering process, exploring effective pathways for AI to empower the high-quality development of the bidding and tendering sector, and promoting the standardized and healthy development of the bidding and tendering market.
Driven by Demand: Charting a New Course for “AI + Bidding and Tendering”
Guangdong has keenly seized the opportunities for transformation in the digital age and adhered to a “demand-driven, forward-looking” approach. By advancing the practical application of “AI + Tendering and Bidding” in cities such as Guangzhou and Shenzhen across dimensions including planning and design, data governance, and application validation, the province has laid a solid foundation for AI-driven development.
Focus on the issues at hand and effectively implement the “Three Combinations.”In response to the practical challenges in the traditional bidding and tendering sector—such as low evaluation efficiency, difficulties in regulation, and a lack of tangible benefits for market entities— by combining national policies with Guangdong’s specific needs, the regions have promoted the adoption of the *Guangzhou Implementation Opinions on Accelerating the Promotion and Application of Artificial Intelligence in the Tendering and Bidding Sector* and the *Shenzhen “AI+” Work Plan for the Housing and Urban-Rural Development Sector*, thereby charting a clear path for the implementation of AI applications in the tendering and bidding sector. By combining technological innovation with institutional innovation, we have advanced technological applications while simultaneously refining the supporting institutional framework. Guangzhou has clarified the primary responsibility of tendering entities, while Shenzhen has established technical guidelines to ensure that AI applications are conducted in accordance with established rules. Adhering to a combination of pilot applications and widespread implementation, Guangzhou selected the housing, municipal, and water conservancy sectors, while Shenzhen chose 72 projects to pilot AI-assisted evaluation. Using a strategy of “small-scale pilot projects to drive broader adoption,” they verified the technical feasibility before gradually scaling up the initiatives.
Strengthen system integration and solidify the data “foundation.”Guangzhou, guided by high-standard data governance, has established an intelligent data lake and data warehouse infrastructure, formulated unified data standards and management guidelines, and comprehensively utilized tools such as data analysis, rule engines, lineage analysis, to cleanse, integrate, and standardize 120 million multi-source, heterogeneous bidding and tendering data records from the trading platform. It has aggregated over 100 million data records covering project approvals, transactions, and regulatory oversight, with a total governed data volume of 160 TB, thereby supporting the creation of a dynamic, comprehensive panorama of bidding and tendering transactions. Shenzhen actively leverages its platform to integrate external data—including business registration, social security, and administrative penalty records—from business entities. By constructing cross-domain data correlation analysis models and a comprehensive profiling engine, the city deeply integrates on-platform transaction behavior data with external government data to generate profiles of public resource transaction projects and candidates, thereby providing objective data support for tenderers to implement the “separation of evaluation and award” and make scientifically informed award decisions.
Accelerate updates and iterations, and drive system optimization.Focusing on advancing the platform’s development and refinement, the initiative has achieved a leap from “functional” to “user-friendly” through multiple rounds of technical validation and model iteration, thereby laying the groundwork for comprehensive rollout. Guangzhou is promoting the deep integration of the “Guang Jiaoyi” digital trading platform with the DeepSeek large language model, leveraging the deep logical reasoning and semantic understanding capabilities of a domestically developed open-source large model with hundreds of billions of parameters. This has broken through the limitations of traditional business systems that rely on rule databases in intelligent applications, significantly improved the efficiency of compliance checks for bidding documents, driven a comprehensive upgrade of the “one-stop online service” experience, and empowered the platform to support decision-making in bidding and tendering. Shenzhen has deeply integrated blockchain, BIM, GIS (Geographic Information System), and artificial intelligence technologies to form a “technology iteration matrix” for digital and intelligent applications in the bidding and tendering sector. This has not only improved bid evaluation efficiency but also enabled intelligent review of technical bid documents through the semantic understanding capabilities of multimodal large language models, propelling bidding and tendering into a new phase of “smart transactions.”
Empowered by Digital Intelligence, Innovation Creates New Scenarios for “AI + Bidding and Tendering”
The true value of technology lies in its application. Guangdong is fully leveraging the advantages of its digital and intelligent infrastructure—including blockchain and big data—to drive the precise implementation of artificial intelligence in core bidding and tendering processes, such as eligibility screening, document review, and transparent oversight, thereby reshaping the paradigm of smart transactions through scenario-driven approaches.
Proactive Testing: Smart “Quality Inspectors” Strengthen Market Access Controls.Guangzhou has launched an AI-powered platform for compliance verification of bidding documents, establishing a three-tier knowledge base comprising “policies and regulations, negative lists, and historical cases.” The platform features 137 compliance checkpoints and 22 key information alerts, automatically screening for restrictive clauses in eligibility requirements and commercial terms to conduct intelligent compliance assessments of bidding documents, thereby enhancing fairness in the bidding market. Since its launch in November 2024, the platform has activated more than 120 regulatory accounts at the municipal and district levels, completed compliance checks on 3,100 construction projects, verified 380,000 compliance checkpoints, identified over 120 issues, and achieved an accuracy rate of 95%. Shenzhen has established an AI-powered qualification review system featuring “automatic parsing + cross-validation + risk early warning.” This system can accurately identify 52 items across seven categories of certificates and licenses—including business licenses, qualification certificates, corporate work safety permits, and project managers’ registered constructor practice certificates—enabling intelligent review, automatic generation of review results, and risk alerts, thereby providing decision-making references for tenderers.
Precision Review: An AI “Think Tank” Enables Efficient Standard Setting.Guangzhou has developed an Intelligent Analysis System for Abnormal Bidding and Tendering Behavior, featuring a four-tier risk monitoring system that includes core, early warning, anomaly, and extended indicators. The system establishes more than 80 risk monitoring indicators, such as hardware feature codes for bid documents, equity tracing of business entities, similarity of bill of quantities, analysis of abnormal quotes, and scoring tendencies of bid evaluation experts. It has innovatively introduced a “Smoke Index” early warning mechanism that assigns comprehensive risk scores to projects, enabling precise risk identification. Shenzhen is exploring the use of large AI models in the construction and government procurement sectors to organize and extract compliance checkpoints. By leveraging technologies such as OCR (optical character recognition), document parsing, and semantic analysis by large AI models to parse bidding documents into formatted data and segment them appropriately. The large AI model then provides feedback on potential violations of policies and regulations, along with detailed justifications and modification recommendations. Currently, this system covers 20 subcategories of construction projects and has completed AI-assisted evaluations for 16 government framework agreement procurement projects, with a technical review accuracy rate of 99.1% and a significant reduction in review time.
Transparent Oversight: A “Magnifying Glass” That Identifies Hidden Risks.Guangzhou uses data as its “radar” and algorithms as its “probes,” exploring the construction of a full-chain digital regulatory system featuring “pre-bid early warning, in-bid control, and post-bid traceability.” It has established a unified intelligent early-warning network capable of effectively detecting clues such as regular fluctuations in bid prices and anomalies in document similarities, and promptly pushing information on abnormal behavior to regulatory authorities. At the same time, physical trading venues have undergone intelligent upgrades, with the establishment of “one file per bid” digital archives. Bid evaluation experts are equipped with smart ID badges that integrate location tracking, facial recognition, and automatic audio capture technologies, enabling dynamic tracking of experts“ movements within the venue and issuing alerts for abnormal behavior, thereby powerfully driving the intelligent upgrade of regulatory capabilities. Shenzhen has established over 200 compliance monitoring points across six major functions—including data overview, early warning and prevention, daily supervision, audio-visual monitoring, complaint handling, and credit-based supervision—and developed intelligent analysis models for bid-rigging and collusion. To date, the platform has monitored 9,603 projects and triggered 25 risk alerts, significantly enhancing the precision of regulatory oversight.
Improved Mechanisms and Standards Drive New Developments in the Application of Artificial Intelligence in Bidding and Tendering
While vigorously promoting the application of technology, Guangdong has consistently adhered to the principles of “standards first, safety and reliability, and inclusiveness and prudence,” focusing on building a standardized and secure environment for artificial intelligence applications.
Prioritize Standards to Establish a Solid “Baseline” for Compliance.We are committed to treating compliance as the “bottom line” for AI applications. While vigorously promoting the application of innovative technologies, we will strengthen safeguards through standards to achieve a dynamic balance between innovation efficiency and compliance and security. Guangzhou has established a three-tier risk control standard comprising “data isolation, content moderation, and dynamic assessment.” Through rigorous data de-identification and isolation technologies, the city protects privacy data from disclosure; it has introduced advanced content moderation systems to automatically detect and filter harmful information, ensuring that generated content is wholesome and compliant; a dynamic assessment mechanism is employed to monitor and evaluate AI applications in real time, ensuring stable system operation and continuous optimization. Shenzhen, having participated in the drafting of the National Data Interface Security Monitoring Standard (GB/T 46796-2025), the “Shenzhen Work Plan for ”AI+’ in the Housing and Urban-Rural Development Sector” clearly defines the application boundaries and compliance standards for AI in bidding and tendering scenarios, establishing a security framework that covers the entire lifecycle of data collection, transmission, and storage, and solidifying the dual foundations of legal governance and technology.
A dual-track approach ensures safety and reliability.In the practice of “AI + Bidding and Tendering,” the division of labor model—“AI lays the groundwork, while humans set the tone”—achieves an organic integration of efficiency and authority, infusing bidding and tendering activities with intelligence and vitality. Guangzhou has established a “multimodal evaluation engine” to build a collaborative model of “AI-powered evaluation plus expert review.” Leveraging the “e-Zhao Zhisheng” smart trading platform, the system automatically verifies bidder qualifications, checks the compliance of commercial documents, and conducts preliminary assessments of the completeness of technical documents. Based on this, experts conduct in-depth analysis of anomalies flagged by AI, assign final scores to subjective items in technical bids, and adjudicate disputed evaluation points. Shenzhen has implemented a dual-track mechanism of “independent evaluation + human review,” in which AI independently conducts full-process eligibility and compliance reviews of bid documents and generates evaluation conclusions, which are then reviewed by experts. This “separation of human and machine” prevents human factors from interfering with the evaluation results.
Adopt an inclusive and prudent approach, and strengthen interagency cooperation.At the departmental regulatory level, Guangzhou and Shenzhen have implemented the “Guiding Opinions of the General Office of the Guangdong Provincial People’s Government on Promoting Inclusive and Prudent Regulation” and are exploring a “sandbox regulation” mechanism in the field of “AI + bidding and tendering,” allowing trading platforms to test new technologies in a controlled environment and providing ample room for “trial and error” in technological innovation; Through small-scale pilot programs, they are gradually refining algorithmic models and steadily advancing the comprehensive application of these technologies while ensuring safety. In conjunction with a special rectification campaign targeting bidding and tendering, they have deployed an “AI module” on the provincial online bidding and tendering supervision platform to screen, verify, and re-examine procurement projects; and establish a mechanism for industry supervisory departments to coordinate with public security authorities and disciplinary inspection departments in the referral and handling of leads on illegal and non-compliant activities, thereby creating a closed-loop deterrent combining “regulation, law enforcement, and disciplinary enforcement” to foster a clean and upright bidding and tendering environment.