Xebia’s customers face significant challenges due to outdated technology and design of some of their applications. Their legacy systems make it nearly impossible to innovate rapidly, severely restricting their agility in an increasingly competitive market. Rather than focusing on new developments and features, their teams are often stuck in an endless cycle of fixing bugs and addressing customer issues.
The use of legacy technologies further compounds the difficulty in hiring skilled developers. New developers prefer modern, widely-adopted technologies, and the shortage of talent comfortable with legacy systems increases the risk of future maintenance and continuity.
Customers express growing concern regarding the long-term reliability of the software, perceiving the existing technological framework as a risk. Quite often performance issues prevent the organizations from successfully pursuing larger, more lucrative clients, severely restricting business growth.
The business challenge is also related to high risk of modernization effort – for large applications the length of modernization project is often estimated on 2 years or more which stops the organizations from even starting it.
Technical Challenges:
The existing solutions typically raise the following technical issues:
- Performance challenges due to outdated, often non-maintained technologies and missing capability to scale the resources or respond to traffic spikes.
- Lack of long-term support for the legacy frameworks and technologies which raises security issues.
- Limited platform flexibility – often locked-in to Windows or legacy platforms.
- Multi-tenant applications available often through remote desktop due to technology used.
- Limited testability of some solutions which increase the risk of application modifications.
- Inability to quickly start the modernization proces which requires strong knowledge of the existing application and high skills of both source and target technology.
Solution:
Xebia designed and implemented a sophisticated solution leveraging Amazon Bedrock and Amazon SageMaker AI. This solution is a compound framework consisting of the following elements:
1. Pre-modernization workshops:
To address long-standing challenges with legacy systems, Xebia conducts a strategic, multi-phase workshop designed to uncover both business and technical barriers to modernization. The process begins with deep discovery sessions, where Xebia partners with the client to understand not only the structural and architectural composition of the application but also the business motivations driving change. This comprehensive diagnostic phase allows the Xebia team to pinpoint technical pitfalls and inefficiencies that hindered scalability and agility.
Such workshop consists also of a proof-of-value phase where limited scope of the modernized application is migrated to the new, target technology, using Xebia’s proprietary tooling deployed on SageMaker AI and using Amazon Bedrock Claude LLMs.
2. Modernizer application:
Modernizer application allows to quickly assess the application codebase, delivering the documentation, technical debt analysis and recommendations for modernization effort. It also allows to quickly perform modernization experiments or retrieve business requirements directly from the code. Modernizer is built on ECS and RDS, and utilizes the Amazon Bedrock Claude models to perform tasks.
3. Modernization phase:
Modernizer application and customized experimentation platform on SageMaker AI allows to prepare the modernization plan and assess the benefits of using GenAI-enhanced modernization (GEM) throughout the whole modernization journey.
Advanced Implementation Features:
- Implemented advanced pattern matching modernization process using Claude models
- Implemented advanced orchestration platform to perform multistep agentic migrations
- Implemented automation platform to orchestrate mass modernization of the legacy codebase with results assessment and comparison stages (old vs new)
- Implemented versatile test generation tools to enable increasing code coverage along with the modernization effort
Outcomes and Benefits:
The implementation of the Amazon Bedrock-powered Modernizer solution delivered significant improvements:
Operational Efficiency:
- Eliminated manual process of creating documentation of the legacy applications
- Simplified starting modernization journey
- Simplified creating modernization experiments and sped up the proces of their evaluation
- Allowed to automate modernization process for large codebases
Business Impact:
- Improved the ability to modernize legacy stack
- Created the opportunity to start the modernization even though it initially was perceived as a long and not necessarily justified proces
- Allowed to create the possibility to utilize cloud to further modernize the technological stack and increase the overall agility
The solution demonstrates how Amazon Bedrock's advanced language models can be effectively used to solve difficult business challenges, such as legacy modernization. By combining multiple AWS services, including Bedrock and SageMaker AI with sophisticated prompt engineering, agentic approach and orchestration platform, Xebia created unique tooling for simplifying modernization journeys for their customers.
The automated modernization proces brings significant benefits in time/costs but also in the team’s frustration in the modernization journey. Modernizer can speed up the modernization process by at least 20%, but it typically reaches 40-50%, up to 80%.
