At the China International Services Trade Fair held in Beijing on September 9, which drew over 90 countries and international organizations alongside 1,800 businesses for offline exhibitions, Qifu Digital President Zhou Xuqiang delivered a keynote address during the financial services segment at the 8th China Fintech Exchange. His presentation highlighted the company's "One Horizontal, N Vertical" enterprise-grade AI office platform and its smart agent matrix tailored for the banking sector, charting a course for data-driven innovation across the financial industry.
Zhou stressed that the past two to three decades of banking informatization, from system development to model construction, have leaned heavily on expert intellectual input. Today, however, combining large language models with smart agents can distill that expertise into scalable intelligence, empowering internal teams and relationship managers with the capabilities once reserved for specialists. This philosophy underpins Qifu Digital's comprehensive full-stack AI offering.
At the core of the "One Horizontal" pillar sits the Qwork enterprise AI office platform, forged within Qifu Technology-S (QFIN-S, 03660) and rigorously tested by thousands of its employees. Moving forward, Qifu Digital aims to extend these battle-hardened capabilities to a broader range of financial institutions. Through MCP and API protocols, Qwork integrates seamlessly with email, OA systems, instant collaboration tools, cloud documents, data platforms, risk management, operations, marketing, and even HR, finance, and administrative systems. Internally, over 200,000 cloud documents have been indexed and made accessible to the AI. The platform's four pillars—foundation model routing, agent orchestration, permission governance, and an AI security gateway—dramatically lower the barrier for employees to safely harness large models, striking a balance between data protection and productivity. Users can leverage natural language commands to handle complex tasks like collating business metrics or diving into customer analysis.
During his presentation, Zhou showcased two real-world applications of Qwork: one where the AI automatically generated a PowerPoint deck by pulling from authorized cloud files, historical speeches, and local data; another where it sifted through past emails, Teams conversations, and T5T records to pinpoint the most valuable banking clients and flag those with the most service issues—producing insights that mirrored manual analysis exactly.
The "N Vertical" dimension encompasses a suite of smart agents focused on banking's critical workflows, chiefly credit underwriting and risk control, plus marketing and growth. Qifu Digital has cultivated a diverse agent ecosystem, including AI due diligence officers, loan approvers, model developers, strategy architects, anti-fraud specialists, credit officers, wealth advisors, compliance supervisors, operations handlers, and growth strategists. These credit and risk agents have already been deployed at a national joint-stock bank, while the marketing and growth lineup has gone live at a regional listed bank in East China.
The AI strategy architect streamlines everything from data ingestion, knowledge and strategy library construction, and data cleaning, to strategy generation, deployment validation, and ongoing monitoring. Its value proposition is fourfold: it makes strategy development intuitive via natural-language adjustments; it accelerates delivery far beyond traditional methods; it enhances quality by automatically selecting high-value features, eliminating redundant variables, and letting all candidates vie for inclusion in model training, yielding sharper and more consistent strategies; and it ensures knowledge continuity by shifting logic from human minds to agents, enabling reuse and iteration. Meanwhile, the AI anti-fraud agent focuses on scrutinizing credit fraud cases, autonomously gathering fragmented, varied-format data from acquisition systems, risk engines, tracking platforms, collection logs, decision frameworks, and external sources as needed. It handles single-case reviews, cross-case aggregation, and syndicate link analysis, collapsing what once took hours or weeks of manual work into minutes while generating fresh anti-fraud rule candidates for human evaluation.
Zhou underscored Qifu Digital's central vision: converting the intellectual capital tied to expert brains into a reusable, transferable, and evolving intelligence via its agent matrix, giving more financial players—especially smaller banks lacking senior specialists—access to richer, more advanced professional tools. Qifu Digital remains committed to a dual-engine approach of platform infrastructure and scenario-specific agents, driving AI in finance from mere viability to true value, and from surface-level adoption to deep-seated integration.
Comments