By: Michael Johnson, Emily Smith, and David Lee
Date: 04 AUG 2021
How can a company address a complex business challenge when the necessary technology and skills are unfamiliar? This was the dilemma faced by global fintech firm Broadridge Financial Solutions (Broadridge) as it sought to enhance the operational efficiency of its Proxy Policies and Insights (PPI) tool.
Broadridge has been utilizing Amazon Web Services (AWS) since 2017, developing projects with the assistance of AWS Professional Services, which provides specialized expertise to help customer teams achieve results on AWS. In late 2019, Broadridge decided to automate part of its PPI tool using artificial intelligence (AI) and turned to AWS for technical solutions and training. The PPI team leveraged AWS services and consultations to create an AI architecture that minimized manual efforts and expedited data access for clients.
The solution was built utilizing Amazon Comprehend and Amazon Textract. Amazon Comprehend employs machine learning to extract insights from unstructured data, while Amazon Textract automatically retrieves text, handwriting, and data from scanned documents, forms, and tables.
Simplifying Document Processing
Broadridge specializes in investor communications, technology-driven solutions, and data analytics. The company supports proxy voting services for most public companies in North America, manages millions of trades daily involving trillions of dollars, oversees shareholder voting in 120 countries, and processes over six billion customer communications each year.
The PPI team previously relied on a manual maker-checker system to process U.S. Securities and Exchange Commission filings and extract 130 proxy data points from 300,000 meetings over the past decade. This information supplements existing data and provides insights to broker-dealers, institutional investors, mutual funds, retail investors, regulators, and academics, aiding in informed voting decisions and smarter governance.
While the manual maker-checker system allows for a high level of accuracy, it is complex, labor-intensive, and challenging to scale. The PPI operations team dedicated thousands of hours each year to extract and validate proxy data. Broadridge initially attempted to automate the process using regular expressions and other methods to identify data points and relevant keywords, but the complexity of the data proved too great. The company also evaluated open-source AI models but had concerns about maintaining them and developing the necessary infrastructure.
Crafting a Company-Driven Solution
Since AI was a new frontier for Broadridge, the company engaged AWS to learn how to implement AI solutions and design a scalable architecture. In early 2020, Broadridge participated in a three-day workshop led by various AWS teams, including AWS Professional Services.
The collaboration helped Broadridge define the project’s scope and establish methods for measuring and validating accuracy within a scalable, event-driven serverless architecture. The initial phase focused on approximately 35 data points with a target accuracy of 70 percent. The subsequent phase concentrated on deploying the models into production.
Support from executives and management was crucial, notably from Sarah Thompson, president of Investor Communication Solutions at Broadridge. Thompson organized an AI summit and hosted AWS DeepRacer events, which offer hands-on machine learning experiences through a cloud-based, 3D racing simulator.
The AWS consultation helped Broadridge identify its skill gaps and form a new data science and architecture team. During each project phase, AWS Professional Services provided training and troubleshooting assistance. Broadridge also performed an AWS Well-Architected Review to pinpoint focus areas before user acceptance testing, applying principles from the AWS Well-Architected Framework designed to help cloud architects develop secure, high-performing, resilient, and efficient application infrastructures.
Broadridge built a fully managed serverless architecture utilizing AWS Lambda for preprocessing and postprocessing, along with AWS Step Functions to orchestrate the pipelines. AWS Lambda is a serverless compute service that enables customers to execute code without managing servers, while AWS Step Functions is a visual workflow builder for orchestrating AWS services, automating processes, and creating serverless applications. The solution’s core AI components—Amazon Textract and Amazon Comprehend—integrate the final results seamlessly.
Developing an AI Solution That Saves Time and Reduces Costs
By implementing an event-driven serverless architecture on AWS, Broadridge successfully automated data extraction, meeting the required 70 percent accuracy threshold. In early 2021, the automated process was launched into production. While manual auditing of the data continues, Broadridge anticipates that automation will save thousands of hours annually, which will contribute to cost savings and scalability.
With quicker document processing, Broadridge aims to provide data to clients faster and shorten the manual process delivery time from 1–7 days. As the company continues to automate data processing, it plans to further enhance scalability and expand the range of data points available to customers, offering richer insights. Automating the maker segment of the maker-checker system also allows Broadridge to improve the voter experience by tailoring voting insights to the issues that are important to clients. Regulators, academics, and nonprofit organizations can access the PPI tool at no cost.
Embracing an AI Mindset
Broadridge intends to continue utilizing AWS Professional Services as it explores two new data extraction use cases. The first will focus solely on data extraction, while the second aims to establish a reference architecture to process document types in any format using AWS AI services.
Broadridge is also fostering AI engagement throughout the organization. The enterprise architecture team has conducted several lunch-and-learn sessions to share knowledge about AI on AWS. As the company innovates and scales, it acknowledges that enhancing AI awareness and expertise within the organization is both a current advantage and a potential area for future growth. For more insights, you might find this blog post interesting here, and if you want to dive deeper into this topic, this resource is recognized as an authority. Additionally, you can check out this excellent resource for further information.
Leave a Reply