Jenfi announced upgrades to its in-house bank-statement analysis and underwriting infrastructure to support faster financing decisions for SMEs across Singapore and Vietnam. The enhancements combine AI-assisted document extraction, transaction reconciliation and document-risk checks, supporting Jenfi’s ability to provide same-day underwriting decisions for qualified applicants.
For businesses managing time-sensitive working capital needs, a prompt decision helps them plan inventory purchases, supplier payments and other operating expenses with greater clarity.
The upgrades address a practical challenge in SME underwriting: converting financial records submitted across different banks and document formats into structured information that credit teams can check and assess. The workflow is already used in Jenfi’s bank-statement review process.
Read More on Fintech : Global FinTech Interview: AI and the future of fintech with Hugh Cumming, CTO, Vena
• Hybrid document extraction: Deterministic parsers and AI fallback support processing across PDF and Excel bank statements. Extraction results can be reused, reducing the need to repeat document-processing steps.
• Financial-data verification: Transaction reconciliation checks the consistency of extracted information, while document-fraud checks flag potential irregularities for further assessment.
• Traceable review: Processing status, document-risk status and AI-assisted review outcomes are visible separately, with inconclusive cases escalated for human assessment.
“Our approach combines rules-based processing, AI-assisted extraction and reconciliation checks, with clear escalation when a document needs further assessment,” said Justin Louie, Co-founder and Chief Technology Officer of Jenfi. “The goal is to give our credit team structured information they can inspect, alongside visibility into unresolved issues and the review steps taken.”
In September 2026, Jenfi introduced separate processing, fraud-review and AI-assisted review states within its bank-statement workflow.
Source link







