Generate suspicious transaction report drafts for financial compliance using generative AI

This article by Divyajeet Singh, Parag Srivastava, Sangeetha Kamatkar, and Vineet Kachhawaha demonstrates how AWS generative AI solutions can automate suspicious transaction reporting (STR) for financial compliance. The solution combines Amazon Bedrock's foundation models with intelligent agents to streamline the traditionally manual process of creating compliance reports. Using Amazon Bedrock Knowledge Bases, agents, and Lambda functions, the system creates draft STR reports by gathering transaction details and analyzing fraudulent entities through conversational AI interfaces.

• Automates STR creation using Amazon Bedrock agents with conversational workflows to collect required information
• Leverages retrieval augmented generation (RAG) with OpenSearch Service to enhance accuracy and reduce AI hallucinations
• Includes web scraping capabilities through Lambda functions to gather real-time fraudulent entity information
• Provides both automated CDK deployment and manual setup options with comprehensive testing instructions