Automating Competitor Rate Monitoring with Gen AI in a leading one stop money shop.

Customer Business Overview

The client is a leading provider of money transfer and payment services, with a particular focus on remittance services to Latin America and other global regions. As one of the key players in the remittance market, the client competes with major companies like Western Union, MoneyGram, and various digital platforms in the U.S.-to-Latin America corridor.

The client’s agent locations often work with multiple money transfer providers, giving customers a choice of service. In this competitive landscape, fees and foreign exchange rates are crucial factors in winning customer business. Exchange rates can vary significantly across locations and tend to be more favorable for in-person transactions compared to online rates.

Business Challenge

To maintain a competitive edge, the client needed to closely monitor its competitors’ exchange rates across different agent locations. The company had representatives collecting competitor rate information through images taken at various agent sites. However, the varying quality and formats of these images made them difficult to process and utilize effectively, requiring manual intervention and leaving valuable data underutilized.

Technical Challenge

The client required an AI-powered solution to automate the extraction and processing of exchange rate data from the collected images. The key technical challenges included:

Developing an optical character recognition (OCR) system to accurately extract text, tables, and other relevant data from the diverse set of images.

Organizing the extracted data into a structured format, such as JSON, by mapping key-value pairs, table rows and columns, and other elements into a consistent structure.

Integrating the data extraction processes seamlessly into the client’s existing AWS ecosystem, leveraging services like Amazon S3 for storage and AWS Glue for data transformation and ETL processes.

Creating datasets tailored to the specific analysis needs, and integrating them into a Data Lakehouse architecture for efficient querying and reporting.

Developing intuitive dashboards and visualizations to enable The client teams to analyze and compare competitor exchange rates across locations, optimizing their own rates to gain a competitive advantage.

Solution Details

Strata Analytics, an AWS Partner with deep expertise in machine learning and data analytics, collaborated with the client to design and implement a comprehensive solution leveraging AWS AI/ML services.

Image Upload and Processing: The client’s representatives uploaded images of competitor rates to an Amazon S3 bucket, triggering an AWS Lambda function to initiate the data extraction process.

Text and Data Extraction: Amazon Textract, a powerful OCR service, was employed to extract text, tables, forms, and other relevant data from the images. This included detecting key-value pairs, structured data like tables, and unstructured text. Due to the existing variation in the data structures, over the initial output of Textract APIs, a multimodal LLM (Claude 3.5 Sonnet) powered by Amazon Bedrock was used to arbitrage the conflicting output and complete those cases where gaps were detected.

Data Structuring: The extracted data was organized into a structured JSON format using an multimodal LLM (Claude 3.5 Sonnet) hosted by Amazon Bedrock. This involved mapping key-value pairs, table rows and columns, and other elements into a consistent structure via prompting.

Database Integration: The structured data was then inserted into an Amazon DynamoDB database, with each image or document corresponding to a new entry or record, and fields populated based on the extracted data.

Historical Tracking and Versioning: A versioning system was implemented to track changes over time, enabling historical comparisons and temporal queries by storing timestamps and other metadata.

Data Analysis and Reporting: The structured data was integrated into the client’s Data Lakehouse architecture, leveraging Amazon Athena and Amazon QuickSight for efficient querying and reporting. Intuitive dashboards were developed to visualize competitor exchange rates across locations, enabling The client’ teams to optimize their own rates for increased competitiveness.

Business Outcomes and Success Metrics

The AI-powered solution delivered by Strata Analytics enabled The client to gain a significant competitive advantage in the remittance market. Key business outcomes and success metrics included:

Operational Efficiency: Automating the data extraction and analysis process reduced manual effort by over 80%, enabling the client’s teams to focus on strategic decision-making.

Data-Driven Insights: The comprehensive dashboards and reporting capabilities provided The client with valuable insights into market trends, enabling proactive strategies to maintain a competitive edge.

By leveraging AWS AI/ML services and Strata Analytics’ expertise, The client successfully transformed a manual, inefficient process into a scalable, data-driven solution.

*This article was elaborated with the help of Generative AI, through the AIML Case Study Generator application on the AWS Party Rock toolkit.


Posted

in

by

Tags: