The operation teams of capital markets face many hurdles in post-trade lifecycle, such as errors in booking, delays in trade settlements and inaccurate regulatory reports. Artificial intelligence and machine learning (AI/ML) technologies like Intelligent Document Processing (IDP), which automate data extraction from documents, can be of great assistance in overcoming these obstacles. This offers potential benefits like employee productivity, enhancement of cash flow, and reduction of operational and regulatory risks.
The article discusses an automation system which has been designed to process derivative confirms at a large scale, using the AI services provided by Amazon Web Services (AWS). Key aspects of this service include Amazon’s Textract and Serverless technologies. The former is a managed ML tool that extracts handwriting, text, and data from scanned documents effortlessly. The latter, a suite of event-driven services designed to manage data and run code, enabling applications integration without managing servers.
In the system described in the article, derivative confirmation documents received from customers are stored in Amazon S3. An event notification automatically activates a message in an Amazon SQS queue on completion of the S3 object upload, which then brings an AWS Lambda function into play. This function triggers the Amazon Textract API and applies a fuzzy match using the document schema mappings saved in Amazon DynamoDB.
Additionally, a web-based human-in-the-loop UI allows for review of the document processing pipeline, offering the possibility of updating schemas in order to train services for new formats. This human-in-the-loop UI uses Amazon Cognito, providing authentication and access control.
The authors note the positive outcomes of combining machine and human intelligence, such as increased accuracy and efficiency. These outcomes make human-in-the-loop an applicable process for various deep learning AI projects, including computer vision, natural language processing (NLP), and transcription.
They conclude, the automation of derivatives confirmation enhances the capacity of operation teams, saves processing time, reduces costs and enhances workforce productivity. This sets a standard for many back-office documents processing use cases in the capital market industry.
To ensure the safety of the above-mentioned automation process, a thorough review of the security aspects of Amazon Textract is recommended along with strict adherence to the guidelines provided. Further information about the pricing of the solution can be obtained through the pricing details of Amazon Textract, Lambda, and Amazon S3.
