Legacy Application Modernisation for a UK Not-For-Profit

Discover how Arrk Group used AI tools to modernise and redevelop a customer’s legacy systems.​

A large UK not‑for‑profit union and trade association representing hundreds of member organisations, each with thousands of individual members. The organisation runs a buying consortium that handles high volumes every day across purchasing, inventory allocation and payment syndication.

The Customer’s mission critical business operations relied on a fragmented, 20‑year‑old set of desktop and client‑server applications. The systems required significant manual intervention and could not scale with demand. Operational complexity and staffing costs continued to increase. Several software and hardware components had reached end of life, creating business continuity risks.

When the customer approached Arrk, suppliers were chasing payments and members were experiencing delays in receiving inventory. Knowledge of critical fixes was limited to a small number of staff. The legacy environment was no longer sustainable and posed a significant risk to business operations.

Our first priority was to restore basic functionality and stabilise operations. We temporarily downgraded selected software components, operating systems and hardware to known stable versions. This returned the platform to a stable operating state and gave the Customer time to plan the modernisation programme while maintaining business operations through manual intervention where required.

Using Arrk’s Remarrk™ Application Modernisation Framework, we produced a practical roadmap focused on scalability, reliability, security and cost control. The plan moved the Customer away from a monolithic architecture and manual processes towards an automated, cloud‑based platform capable of supporting future growth.

Older Java Versions and Coding Styles
We used AI‑assisted tools to update the codebase to supported Java versions and adopt modern language features. This accelerated refactoring and reduced manual rework. We also used AI to generate and update unit tests to improve coverage.

Monolith to Microservices
The legacy application was a monolith. Using domain‑driven design, we split the system into microservices, each aligned to a specific business area. This improved service isolation and made future enhancements easier. AI‑assisted design reviews helped test service boundaries early.

Manual Deployments
We implemented an automated pipeline so builds, tests, security checks and deployments run consistently. This removed bottlenecks and reduced release risk.

Data discrepancies during like‑for‑like testing
Differences appeared during parallel runs due to inconsistent databases. We created a fresh copy of the live database and used it as the single comparison source for side‑by‑side validation.

Daily snapshots and a clean cutover
While both systems ran in parallel, we took a daily snapshot from the live database and loaded it into the new system for test runs. On the planned cutover date, we made the old system unavailable and directed all users to the new platform. Any issues were resolved within the new system rather than rolling back to the legacy platform.

Ongoing Support & Cost Optimisation
To address concerns about cloud cost and security, Arrk provided its Commercial Support Flex Model. The service includes:

• Lower monthly cloud infrastructure costs through right‑sizing and automation.
• High availability and predictable stability.
• End‑to‑end information security managed by a single team.
• Stabilised operations during a period of disruption, followed by a planned modernisation programme.
• Reduced operational risk by replacing end‑of‑life components and reducing key‑person dependencies.
• Faster change cycles with automated pipelines and microservices.
• Lower running costs and an improved security posture on AWS.

Having successfully modernised the customer’s legacy system, we implemented:

Cloud platform: AWS
Architecture: Event‑driven and container‑based microservices

• Compute and integration: AWS Lambda, AWS ECS (containers), API Gateway.
• Messaging and events: Amazon SQS, Amazon EventBridge.
• Storage and delivery: Amazon S3, Amazon CloudFront.
• Communications: Amazon SES

Application layers

• Front end: React for a clearer, faster user experience.
• Back end: Refactored Java codebase using Spring and Spring Boot to integrate with AWS services.

Engineering practices

• AI‑assisted coding and testing to accelerate refactoring and improve coverage.
• CI/CD for automated builds, tests, security checks and deployments.
• Automated backups for SQL Server on AWS.

The result is a stable, scalable system aligned with business requirements. Rather than a like-for-like replacement of the previous platform, the solution was designed to support future development and change.

Working with Arrk, the customer moved from an unsupported legacy environment to a stable, scalable platform designed to support future business requirements.