Technical debt costs US enterprises an estimated $2.41 trillion per year, according to Accenture research cited by AWS. For most organisations, roughly 30% of engineering time gets absorbed by maintaining legacy systems – work that’s necessary, but produces no new business value.
That’s the problem AWS has been trying to solve with AWS Transform, its agentic AI service for enterprise application modernization. Transform launched in May 2025, gained substantial new capabilities at re:Invent in December, and in January 2026 was positioned squarely at VMware customers looking for an exit. SDxCentral reported execution time reductions of up to 80% on some workloads.
The headline numbers AWS is publishing are significant. Customers have used Transform to analyse an estimated 1.1 billion lines of code and save more than 810,000 hours of manual effort, according to AWS’s own re:Invent announcement. If your organisation has been deferring an application modernization project because the business case never quite worked, it’s worth understanding what has actually changed – and what hasn’t.
What AWS Transform Actually Does
Transform isn’t a replacement for AWS Application Migration Service (MGN) or the existing 7 Rs framework. It’s a layer of specialised AI agents that automate the tasks that traditionally eat the most hours in a modernization project: code analysis, dependency mapping, refactoring, documentation, and test plan generation.
AWS has built specialised agents for four workload types:
- Windows .NET applications, with up to 5x acceleration across application, UI framework, database, and deployment layers. AWS reports operating cost reductions of up to 70% when moving off Windows and SQL Server licences to open source equivalents.
- Mainframe workloads, where Transform extracts business logic from COBOL and JCL code, decomposes monoliths into domains, and generates refactored Java or microservice specifications.
- VMware environments, with automated discovery, network translation, wave planning, and EC2 instance optimisation.
- Custom code, where the agent learns organisation-specific patterns from examples and applies them across large multi-repository codebases.
The underlying architecture uses three layers: specialised AI agents trained for specific transformation tasks, an agentic coordination layer that manages workflows and dependencies, and a natural language interface that lets business and technical teams work from the same view of the project.
Why This Matters for Application Migration
Most application migrations stall in the same place: the assessment and planning phase. Teams spend months mapping dependencies, documenting undocumented systems, and trying to understand code nobody has touched in a decade. By the time they’re ready to move anything, priorities have shifted and budgets have been reallocated.
Agentic AI changes that sequencing. Discovery that used to take a quarter can now take a week. Code analysis that required senior engineers reading through legacy systems line by line can run in parallel across hundreds of applications simultaneously. That compression doesn’t just make migrations faster – it makes more migrations viable. Projects that would never have survived the business case under manual estimates suddenly look achievable.
This is particularly relevant right now for two groups. First, organisations with significant VMware footprints who are reassessing their position following Broadcom’s licensing changes. Second, organisations running Windows Server and SQL Server workloads where licensing costs have quietly become a material line item.
What AWS Transform Doesn’t Do
AWS Transform is genuinely impressive, but it’s worth being clear about what it isn’t.
It’s not a strategy. Transform will accelerate whatever migration path you point it at, but it won’t tell you whether lift-and-shift, replatform, or refactor is right for a given workload. That decision still requires understanding the business context, the application’s lifecycle stage, the team’s operational capacity, and the commercial drivers behind the move.
It’s not a replacement for architectural judgement. The agents produce high-quality transformations, but the output still needs review. AWS is explicit that reliability depends on codebase consistency, the quality of examples provided to custom agents, and the complexity of underlying frameworks. Real-world code is rarely as clean as demo scenarios suggest.
It’s not operations. Once applications land in AWS, they need to be monitored, secured, cost-optimised, and kept compliant. Transform accelerates the move; it doesn’t run what’s on the other side.
Where the Human Partnership Still Matters
The most useful way to think about AWS Transform is as a force multiplier for engineering teams, not a substitute for them. The agents are excellent at executing well-defined transformation tasks at scale. They’re not a replacement for the strategic work that determines what to transform, in what order, and toward what target architecture.
That strategic work looks like this in practice. Before any code moves, you need a clear disposition strategy across your application portfolio – which workloads justify a reimagine, which need a straightforward replatform, and which should simply be retired. You need a target AWS architecture that accounts for compliance requirements, cost optimisation, and the operational model your team can actually support. You need a migration wave plan sequenced around business priorities rather than technical convenience. And you need the operational capability to run modernised applications once they’re live.
This is where a managed AWS partner adds value that agentic AI cannot. Transform handles execution; the strategic frame around it – assessment, architecture, prioritisation, and ongoing operations – remains a human problem.
Modernize Your Applications With Confidence
As an AWS Premier Tier Services Partner, Opti9 helps organisations plan and execute application modernization projects that deliver measurable business outcomes. Whether you’re evaluating a VMware exit, modernising legacy .NET workloads, or planning a broader migration, we bring the architectural judgement and operational expertise that sits around the tooling.
Get in touch today to discuss your application modernization strategy
