AWS Transform: Agentic AI Automates Legacy Code Modernization and Reduces Technical Debt

At the re:Invent 2025 conference, AWS announced a sweeping enhancement to its application‑modernisation service, AWS Transform, introducing a new suite of Agentic AI capabilities designed to tackle technical debt and heavy manual modernization tasks.

The update promises to dramatically reduce workload by enabling agentic automation that accelerates modernization for Windows .NET, VMware, and mainframe environments, while adding a Custom capability to modernise proprietary languages, frameworks, and internal syntaxes. Official data shows that AWS Transform Custom can modernise hundreds or thousands of applications at speeds up to five times faster than traditional manual workflows, with feedback‑driven agents refining performance over time.

Real‑world examples include Air Canada reducing project time and cost by 80% for thousands of AWS Lambda functions, QAD cutting a two‑week process to three days and saving more than 7,500 developer hours annually, and Thomson Reuters automating 1.5 million lines of code per month, lowering costs by 30%.

For Microsoft‑centric environments, the update provides full‑stack modernization, migrating systems toward open‑source and cloud‑native options such as PostgreSQL, and can reduce operational costs by up to 70% by removing licensing dependencies.

AWS Transform also introduces three new AI agents for deep code analysis, rule extraction, and automated documentation, plus a task‑planning agent for test plans and automation scripts, while extending support to landscape‑level VMware migration and broader network and security ecosystems. According to AWS, the new composability program will let partners integrate their tools into AWS Transform, expanding industry‑specific workflows in regulated sectors such as finance.