Trade finance is a very document and human intensive process. Especially, processing Letter of Credit (LC) in banks are 100% manual. LC supporting documents usually comprises of multiple unstructured documents like MT700, Bill of lading, invoices, Bill of Exchange, etc. All these documents are in different layout and format. LC also involves many international trade rules and standards like UCP and ISBP to be checked and verified which today requires many human checkers with strong domain knowledge. Therefore today banks need to put multiple human checkers to process each letter of credit case to avoid human errors. Each human check takes at least 40 mins to process one case. Therefore it is very time-consuming and repetitive.
Powered by 6Estates’ state-of-the-art Machine Reading Comprehension (MRC) and Optical Character Recognition (OCR) technologies as well as domain knowledge of UCP & ISBP banking rules, LC Automize is an AI engine which is able to read LC documents, understand them, extract key information and conduct the rules check automatically. It can greatly reduce the dependency of human LC document checkers to conduct first-level document checks and reduce turnaround processing time to enable the bank to scale up its trade finance operations exponentially.
Convert LC supporting documents into digital format through OCR technology. Check and auto-correct the converted text using NLP technology.
Extract key information from various documents using MRC technology and convert it to structured data.
Construct knowledge base and rules engine in trade finance and automatically conduct compliance checks in accordance to international trade standards.
Collect user behaviour for incremental machine learning so as to improve the accuracy of the AI model.
Technology Highlight
Optical Character Recognition (OCR),Machine Reading Comprehension (MRC),Information Extraction
Reduce human Dependency on mundane tasks
Scale up trade finance operations exponentially
Cut down on operational and compliance risk
Accumulate trade rules and domain knowledge