Ant Consumer Finance Deploys Vertical AI Model in Dispute Mediation

Ant Consumer Finance has integrated its proprietary vertical AI model, named 'Jiangxiaotun', across customer rights protection workflows, helping human mediators resolve 36,000 financial disputes in 2025 with a 99% resolution rate. Speaking at the Inclusion Conference on the Bund, General Manager Jiang Hao revealed that the firm's AI customer service handling ratio and user satisfaction have both surpassed 90%. In addition to frontline service, the company's multimodal detection and graph-mining tools have helped public security organs crack down on 14 organized underground fraud syndicates exploiting synthetic claims.

The deployment demonstrates how regulated consumer finance firms are moving generative and analytical AI beyond generic chat into high-stakes, emotionally sensitive compliance operations. By keeping human mediators at the center of final judgments while using models for fact synthesis and fraud detection, Ant Consumer Finance offers an operational blueprint for balancing operational efficiency with consumer protection mandates.

Ant Consumer Finance has integrated an in-house specialized artificial intelligence model into its consumer protection and dispute resolution operations, resolving tens of thousands of customer conflicts while aiding police investigations into organized fraud, company leadership announced at the 2026 Inclusion Conference on the Bund. Speaking at the event, Ant Consumer Finance General Manager Jiang Hao stated that the firm's proprietary vertical large model, dubbed "Jiangxiaotun," now underpins frontline customer service, illicit gray-market governance, and mediation workflows. In 2025, an AI-assisted mediation system helped human mediators resolve 36,000 financial disputes, achieving a 99% dispute resolution rate, according to the company. Automated agents now handle more than 90% of customer service inquiries, with reported user satisfaction also exceeding 90%. Jiang emphasized that artificial intelligence applications in finance must be measured by their ability to resolve substantive operational issues rather than raw algorithmic complexity. Traditional automated customer service systems typically operate by matching customer queries against standard knowledge bases to reduce headcount costs. In contrast, Ant Consumer Finance connects user profiles, risk parameters, and behavioral data to allow the AI model to infer user intent within specific operational contexts. The deployment comes as China's consumer finance sector pivots from rapid balance-sheet expansion toward risk-managed sustainable operations, accompanied by tighter regulatory requirements across the lifecycle of consumer lending. However, the rise of generative AI has also lowered the barrier for illicit rings to manufacture fraudulent evidence, fabricate medical or financial hardship documents, and conduct coordinated proxy complaints. To counter these threats, Ant Consumer Finance deployed multimodal detection algorithms alongside large-scale financial graph mining to identify coordinated fraud patterns across voice, text, imagery, and transaction networks. According to company disclosures, the system has helped public security authorities crack down on 14 organized underground fraud syndicates. For dispute settlements, the lender established an on-site mediation workstation using an "AI-accompanied" framework. Under this arrangement, the AI platform extracts key case facts, evaluates user sentiment, and drafts proposed resolution packages, while accredited human mediators retain sole authority over final assessments and communication.