From 20-Year Legacy Code to Cloud-Native: How One FMCG Giant Transformed 16 COBOL Applications in 12 Months
Executive Summary
A leading FMCG company successfully migrated 16 legacy COBOL mainframe applications to modern cloud-native architecture, eliminating costly mainframe dependencies while enabling real-time processing for retail sales, inventory management, and supply chain operations.
The transformation leveraged Generative AI tools to accelerate development and reduce manual analysis effort.
“The biggest challenge was modernizing 20+ year old COBOL systems while maintaining business continuity. Everything was interdependent, and we needed a solution that could handle the complexity while accelerating our digital transformation.”
Avkash Chauhan
AI and Digital Engineering SVP
Client Overview
The client is a prominent FMCG organization managing complex retail operations, including sales processing, inventory reconciliation, stock replenishment, and financial posting across multiple distribution channels and retail partners.
Their earlier tech stack included:
- 16 Legacy Applications
- 5TB Data Volume
- 20+ year-old System
Business Challenges
The client’s 20+ year-old mainframe infrastructure was becoming increasingly costly and inflexible, hampering the company’s ability to compete in the FMCG market.
- Legacy Infrastructure Constraints
Expensive COBOL-based mainframe systems with increasing Broadcom licensing costs were draining IT budgets while limiting development agility and making it difficult to attract modern development talent. - Operational Inefficiencies
Batch-only processing created delays in inventory updates and sales reporting, while manual error handling and outdated “green-screen” interfaces reduced productivity and increased operational risk. - Integration Complexity
Complex interdependencies with SAP and external retailer systems, combined with minimal documentation, created fragmented workflows that were difficult to maintain and nearly impossible to enhance.
The transformation leveraged Generative AI tools to accelerate development and reduce manual analysis effort.
A5E’s Approach and Solution
A5E Consulting implemented a comprehensive modernization strategy combining cloud-native architecture with Generative AI acceleration, following proven methodologies for large-scale legacy transformations.
The project was executed in a phased manner, encompassing:
Discovery & Planning (6 months)
The initial phase included detailed functional workshops with business stakeholders, complete system dependency mapping across all 16 applications, database schema analysis, and architecture design.
- A comprehensive asset inventory was conducted using Amazon Q and Claude to analyze 5,000+ COBOL, PL/I, and JCL files per application.
- The team conducted extensive SME interviews to capture undocumented business logic and created detailed migration roadmaps with risk assessments and rollback strategies.
Development & Migration (6 months)
This phase included extensive parallel testing, user training programs, and phased cutover execution with geographic rollout strategies.
- Microservices were developed parallely using GitHub Copilot for code generation, implementing Java Spring Boot, Node.js, and Python services.
- The data migration was executed using enterprise tools including AWS DMS and Qlik Replicate to move 5TB of data from DB2 to PostgreSQL.
- The team completed a comprehensive CI/CD pipeline implementation with automated testing, quality gates, and deployment orchestration.
A specialized team of 20+ professionals, including enterprise architects, mainframe experts, cloud developers, and data specialists was formed. They were supported by dedicated PMO and governance structures to ensure seamless execution and risk management throughout the transformation.
Key Solution Features
- Real-Time Processing
Event-driven architecture enabling real-time inventory updates, sales processing, and order fulfillment, replacing batch-only operations. - API-First Design
RESTful APIs with comprehensive monitoring provides seamless integration with retail partners and internal systems. - Legacy Format Support
Maintained compatibility with EDI and fixed-width files while developing modern JSON/REST interfaces for new integrations. - Phased Migration
A geographic and retailer-based rollout strategy minimized business disruption with comprehensive rollback capabilities.
Technology Stack Transformation
Legacy Stack | COBOL batch jobs, DB2 database, Sterling File Gateway, manual processes |
Modern Backend | Java Spring Boot, Node.js, Python microservices with AWS Lambda and ECS |
Database Layer | Amazon RDS (Postgres SQL/MySQL) replacing mainframe DB2 |
API & Integration | AWS API Gateway, EventBridge, Kafka for real-time processing |
User Interface | React dashboards replacing terminal-based CICS interfaces |
GenAI Tools | Amazon Q, GitHub Copilot, Claude for development acceleration |
Key Outcomes & Business Impact
35% Cost Optimization
Eliminated expensive mainframe licensing costs and reduced operational expenses through cloud-native architecture and automated processes
40% Code Generation Boost
GenAI tools accelerated boilerplate API development, microservice scaffolding, and automated test case generation across all workstreams.
50% Faster Development, Zero Downtime
Accelerated development cycles using GitHub copilot, modern tooling, and CI/CD pipelines while maintaining 100% business continuity. Generative AI-assisted documentation and testing saved 50% of time, further enhancing delivery speed.
100% Legacy App Migration
Seamless API-first architecture supporting both legacy partners and modern retail systems with flexible EDI and fixed-width file format compatibility.
5TB Successful Data Migration
Seamlessly migrated 5TB of critical business data from legacy DB2 mainframe systems to modern PostgreSQL databases using enterprise grade tools.
Conclusion
This mainframe modernization demonstrates how strategic planning and Generative AI acceleration can transform 20+ year-old legacy systems with agile, cost-effective solutions. The 35% cost reduction and real-time processing capabilities positioned the client for sustained competitive advantage. By eliminating mainframe dependencies, the organization gained the agility needed to respond quickly to market changes while serving as a blueprint for other enterprises modernizing legacy systems.
Contact A5E Consulting today to simplify your manufacturing operations with proven SAP S/4HANA expertise.
About Avkash Chauhan
Avkash leads growth initiatives in North America and provides technical leadership across client engagements. With over 20 years of experience in AI, ML, cloud, and enterprise software, he has held leadership roles at Microsoft, H2O.ai, and Roambee, and holds a Master’s in Computer Science. Consult with Avkash on your Gen AI strategy—get in touch with us today.
About A5EConsulting
A5E is a global consulting company partnering with clients to chart, build, and deliver their. Digital Transformation journey. We execute a blend of expert process consulting services and top-tier technology platforms at organizations, making them more agile.
Discover how A5E’s expertise in generative AI consulting can enhance your services. Contact us today!
Related Case Studies
A fast-growing translation technology company was hitting scalability limits. Its “Uber for Translators” model still relied heavily on human interpreters—driving up costs and slowing real-time communication. With over 200 languages supported, the company needed a breakthrough to stay competitive.
A5E Consulting partnered with the client to reimagine its translation platform using generative AI. By combining LLM-based translation with optional human oversight, the team built a hybrid system capable of delivering 95%+ accuracy at a fraction of the cost. Legacy issues—like handling gender nuance, cultural context, and fragmented inputs—were addressed through custom model tuning and a centralized “Model Context Protocol.”
The result was nothing short of amazing. What previously cost $1.50/min now became a flat monthly fee under $100. Customers could engage in multilingual conversations, access AI-based proofreading, and rely on seamless real-time interpretation across devices—all through a simple QR-based interface.
Key Outcomes:
- 95%+ translation accuracy
- 400ms average response time (60%+ faster)
- 60%+ reduction in customer costs
- Real-time access to 200+ languages
Discover how A5E helped this global language tech company with GenAI.
Download the case study to learn more.