Kyndryl is tapping into agentic AI technology from AWS to offer enterprise customers another way to integrate with or move mainframe applications to the AWS Cloud platform.
Specifically, Kyndryl will offer an implementation service based on the recently announced AWS Transform package, which analyzes mainframe code bases, decomposes them into domains, and modernizes IBM z/OS applications to Java using orchestrated AI agents, according to AWS. “The system’s architecture leverages deep learning models along with AWS extensive mainframe modernization expertise, accumulated from 19 years of enterprise migration projects,” AWS wrote in a blog post.
Traditional modernization approaches can be plagued by complexity and can require multi-year timelines, according to AWS: “Legacy mainframe systems contain millions of lines of COBOL, PL/I, and Assembler code, developed across decades. These systems house complex business logic and data relationships critical to daily operations. The monolithic architecture of mainframe applications restricts rapid changes and innovations,” AWS wrote. “Mainframe applications typically entail tightly integrated monolithic architectures that don’t easily translate to cloud-ready architectures. Complex business logic is often buried deep within legacy code, which often lacks clear documentation. Additionally, these systems often employ nested and intricate coding patterns that were developed decades ago but do not translate well with modern architectures.”
AWS Transform and its AI agent are designed to address those challenges: “Users can interact with AWS Transform’s agents to create customized modernization plans through an interactive dialogue, defining target patterns and objectives that map their unique transformation journey from analysis through code conversion. This reduces modernization timelines from years to months while maintaining functional equivalence,” AWS wrote.
For most organizations, conducting exploratory and deep analysis of source code and documentation can be challenging, time-consuming and prone to errors, Kyndryl said in a statement. “AWS Transform’s ability to translate complex code terms using agentic AI improves understanding of and working with application documentation. Kyndryl’s projections indicate that a project typically taking 12 months could potentially be completed in about 8 months, representing substantial time savings,” Kyndryl stated.
In addition, Kyndryl’s AI-powered open integration platform, Kyndryl Bridge, can consolidate cloud and mainframe automation and event handling to help customers identify those applications that can be modernized in new ways. Customers can then use AWS Transform to modernize and migrate mainframe applications to AWS, according to Kyndryl.
Google Cloud a partner, too
Kyndryl recently said it would be using AI technology in its mainframe modernization partnership with Google Cloud. Kyndryl’s Mainframe Modernization with Gen AI Accelerator Program includes Google Cloud’s AI and Gemini large language models, Mainframe Connector package, and BigQuery analytics platform. Google’s Gemini LLMs are integrated into the Google Cloud platform and offer AI-based help across services and workflows, Google stated.
The package supports features such as COBOL-to-Java application coding assistance, and it enables AI training using customer on-premise data, according to Kyndryl. With the Google service, the idea is to offer customers generative AI-driven code analysis and documentation of mainframe applications. Then, the service will help customers who are interested rewrite their mainframe applications to applications for Google Cloud, the companies said in a statement. Following an assessment, customers receive a modernization blueprint and plan, from Kyndryl Consulting. Testing, certification, and de-risking of the modernization project is also offered.