A small company without an established borrowing record can spend days waiting for a lender to verify basic information. A new Guizhou system is designed to compress that stage into minutes by joining datasets that normally sit in separate public-data systems. Xinhua reported on August 30 that the Guiqi AI agent was released at the 2026 China International Big Data Industry Expo in Guiyang, where it was presented as an application of public data authorized for operation.
Guiqi AI links data from industry and commerce records, social security and housing provident funds. The purpose is not to make a bank’s lending decision automatically, nor did the report claim it eliminates human risk assessment. Its stated function is narrower: speed up credit verification for small and micro enterprises by bringing evidence together in a form that can be checked more quickly.
According to Xinhua, the process has moved from several working days to the minute level. That is a meaningful operational change if it can be repeated safely, because small businesses often face a mismatch between their need for quick working capital and the time required to document their standing. The release also makes Guiqi AI a useful case study of how China’s data-policy agenda is being converted into a specific service rather than remaining a high-level strategy.
It follows the broader data-infrastructure discussion in EastFrontier’s analysis of Guizhou’s data centers and China’s AI development model. Data centers supply computing capacity, but a public-data agent tests another question: whether existing administrative records can be organized into a useful, accountable workflow for ordinary firms.
How Guiqi AI Uses Authorized Public Data for Credit Verification
The key phrase in the Xinhua report is “authorized operation.” Guiqi AI is built on public data that has been approved for use, rather than on an unspecified scrape of personal or corporate records. Its data links include business registration information, social-security data and housing provident-fund data. Those sources can provide different signals about whether a firm exists, operates and maintains formal employment relationships.
That does not mean the system has access to every fact a lender might want, nor does it mean each data field should be treated as equally relevant. A credit decision can involve revenue, cash flow, collateral, sector conditions and bank policy. The reported improvement is at the verification stage, where checking separate databases can delay an application before the lender has even reached a substantive decision.
The structure therefore matters as much as the AI label. If multiple public systems can be joined under a defined authorization framework, the agent can produce a common evidence layer for an applicant and a financial institution. The potential value is especially clear for firms that do not have a long conventional credit history but do have traces in formal administrative systems.
Tong Ling, founder and chairperson of Lanxiang Zhilian (Hangzhou) Technology Co., Ltd., said the system can enable small firms without existing credit records to obtain bank credit. That is a company-side assessment, not proof that every such enterprise will receive financing. The distinction is important. Faster verification may widen the set of firms that can be evaluated, but lending remains subject to a bank’s independent judgment and risk controls.
Why Minute-Level Checks Matter for China’s Small Firms
The difference between several working days and minutes is not only a technical metric. For a small manufacturer, retailer or service firm, a long wait can make financing less useful when inventory, payroll or supplier payments are immediate. A more rapid check can allow the business and lender to move to the next step sooner, provided the underlying data are accurate and the authorization process is clear.
China’s National Data Administration has made data circulation and application a central part of its AI policy agenda. The importance of usable data also sits behind China’s expanding AI compute capacity: hardware scale matters only when enterprise workflows can translate information into an actionable service. Guiqi AI is one downstream example, using authorized records to change a decision workflow.
The two initiatives should not be conflated. Annotation pilots focus on making raw data useful for training AI across sectors. Guiqi AI focuses on combining existing public datasets for a particular financial-service process. Yet both depend on common issues: data quality, permissions, interoperability and the ability to show how a result was produced.
The public-finance angle also helps explain why Guizhou has become a testing ground. The province has spent years building a data-industry identity, and the 2026 expo brought 372 exhibitors and more than 1,100 new technologies, according to the Xinhua report. A credit-verification agent is more tangible than a generic expo display because its intended user is identifiable: a small firm seeking a bank loan and a lender trying to verify the application efficiently.
The Limits of an AI Agent in Lending Decisions
The system’s promise should not be mistaken for a claim that algorithms can solve small-business financing on their own. A faster verification process can reduce administrative friction, but it does not create a business’s repayment capacity. Banks will still need to decide how to weigh the evidence, protect sensitive information and explain adverse decisions when they occur.
Data governance is particularly important where public records are involved. The usefulness of connecting business, employment and housing-fund information comes with obligations to determine who is allowed to access it, for what purpose and for how long. The Xinhua report identifies public data authorized for operation as the foundation. The next test will be whether the underlying rules are sufficiently clear for companies, local agencies and financial institutions to trust the workflow.
There is also a question of portability. Guiqi AI was launched in Guizhou, and the report does not say that it is a national banking product. Its value lies in demonstrating one local implementation. If other cities pursue similar systems, their results may differ because the completeness of public records, the structure of local authorization and the readiness of lenders are not uniform.
For China’s AI industry, the case is a reminder that some of the most consequential applications may look less spectacular than a new foundation model. Guiqi AI is not presented as a system that replaces bankers. It is a data-organizing agent aimed at a specific bottleneck. If it reliably reduces credit-verification time while preserving legal and financial accountability, it will offer a stronger model for public-data AI than any one-time expo demonstration.