Modernizing 73 legacy COBOL services to Java/Spring Boot via an AI-assisted workflow
Overview
Industry
Location
Real estate
USA
Technology Used
AI
COBOL
Java
Python
REST
Spring Boot
Project overview
The project focused on modernizing a legacy application based on COBOL by replatforming it to a cloud-ready environment and transforming its services in Java and Spring Boot. The client needed a solution that would reduce dependency on legacy technology, prepare the system for future cloud migration and accelerate the modernization process without losing control over quality. In response, we built an AI-assisted modernization workflow that supported the analysis of existing COBOL services, generated their Java/Spring Boot equivalents and validated the behavior of the new services. A key element was the shadowing process that compared the performance of the new Java services with the original COBOL services in terms of response consistency and execution times.
Client background
A property intelligence company that provides analytics and workflow solutions across the real estate ecosystem. By processing real-time data signals throughout the life cycle of properties, the company identifies hidden risks and new opportunities for real estate agents, mortgage lenders, insurance carriers and industry innovators.
Goals
The goal of the project was to modernize legacy COBOL services into a Java/Spring Boot application, thereby reducing the client's dependency on legacy technology and specialized COBOL expertise. This involved building a repeatable, AI-assisted modernization workflow designed to minimize manual effort. Maintaining absolute functional consistency with the original COBOL implementation was paramount. This was achieved by validating new services through shadowing, which meticulously compared execution times and response consistency against the legacy system. Ultimately, the project aimed to enhance overall system maintainability, observability and scalability, preparing the application for seamless future deployment in a cloud-ready environment.
Results
73 services successfully migrated
73 legacy COBOL services – spanning over 3,500 COBOL files and more than 500,000 lines of code – were migrated into a modern application built on Java and Spring Boot.
55% reduction in human delivery effort
The taken approach cut manual engineering hours by over half compared to standard manual migration or basic LLM-assisted methods.
90% automation in rewrite workflow
The AI-assisted modernization pipeline achieved roughly 90% automation across the code rewrite process.
Substantial cost reduction
Achieved a significant decrease in modernization expenses compared to the baseline approach.
4-to-6-hour processing cycle
The automated AI workflow processed individual legacy services in under six hours before team review, followed by ~8h of additional manual testing and ~4h UAT of deployment – about 16h total vs. ~36h manually.
Validated functional parity
The new services achieved strict response consistency and optimized execution times through systematic shadowing against the original COBOL process.
Repeatable enterprise blueprint
The project delivered both a cloud-ready application foundation and a repeatable workflow that can be seamlessly applied to future legacy modernization initiatives.
Comparable performance
Post-optimization performance stays within ±25% of the legacy services, keeping results at a broadly comparable level.
What our client says
We'd love to hear from you!
Fill out the form - we'll get back to you as soon as possible
