AWS is leaning on its partner ecosystem to speed up mainframe modernization projects
The company this month unveiled AWS Transform Composability, a platform designed to help service providers build specialized AI agents that handle specific parts of mainframe-to-cloud migrations. Partners can encode domain expertise into reusable agentic tools that plug into the platform, automating migrations while preserving core system functionality.
The platform marks the hyperscaler’s latest push to shorten modernization timelines from years to months. AWS Transform Composability converts and refactors code, data and business rules during migrations, improving upon traditional “lift, then fix” strategies.
Under the new model, partners design agents for particular industries or legacy environments — COBOL-based insurance claims systems or government record-keeping applications, for example. Each agent reflects industry knowledge, regulatory requirements and business logic that AWS said are often difficult to capture through generic automation alone.
Operationally, the process works as follows: Partners develop and containerize agents locally, upload them to AWS’s Elastic Container Registry and register them within the AWS Transform Composability framework. Once approved, agents contribute to multi-agent workflows that target industry-specific requirements.
AWS said automating domain-specific tasks reduces manual work and rework, which lowers expenses. According to the company, customers typically see cost reductions of 30% to 40% compared with traditional modernization approaches, depending on workload complexity and organizational factors.
For AWS partners — typically large managed service providers, system integrators and consultancies — Transform Composability offers a way to turn existing expertise into reusable modernization tools.
Partners can embed proprietary methodologies and industry-specific knowledge directly into the modernization process through custom agents, the hyperscaler said. This allows MSPs and other providers to deliver specialized services rather than a one-size-fits-all migration.
Partners also can extend their portfolios by building targeted solutions for specific industries or technical domains, such as healthcare claims processing or financial reporting systems. Over time, AWS expects agents to be used across multiple customer engagements.
The platform further allows partners to capture and retain internal expertise. Knowledge that once lived with individual specialists can be encoded into agents, reducing variability and turning institutional memory into reusable data.
AWS framed the platform as an ongoing project, with partners building and adding agents as requirements evolve. Early contributors to the platform include Accenture, Capgemini, IBM, Infosys, Deloitte and TCS.
