AWS Transform - How Agentic-AI Automation Changes Migration Cost and Delivery

AWS Transform - How Agentic-AI Automation Changes Migration Cost and Delivery

If you have read our guide to AWS migration costs and process, you already know the shape of the spend: most of a migration budget is people’s time, not the infrastructure bill. AWS Transform aims straight at that line item. In this article we treat it as a tool that moves cost and timeline, not as a single feature. We covered the narrow case of Azure-to-AWS .NET code conversion separately in our article on AWS Transform and .NET code conversion; here we look at the whole platform.

What AWS Transform is

AWS Transform is an AWS service that uses specialised AI agents to automate the migration and modernisation of enterprise workloads. It grew out of the transformation capabilities announced as Amazon Q Developer at re:Invent 2024 and reached general availability in 2025. Instead of one “click to migrate” wizard, it is a set of agents that break down specific, time-consuming tasks in a transformation project and run them with a human in the decision loop.

AWS points AWS Transform at three large classes of workload, plus a fully customisable path:

One thing to settle up front: the Azure-to-AWS .NET code conversion we covered separately is one feature inside AWS Transform, not the whole service. The platform is much broader, and it is the platform, not a single SDK conversion, that changes the migration bill.

What it automates, and why that moves the economics

The value of AWS Transform comes from automating the phases that, in a classic migration, are the most labour-intensive and repetitive. It is not a single code-writing agent but several agents working across different stages:

• Discovery and dependency mapping. An agent inventories the estate and finds the links between applications, servers, and databases. Done by hand this is weeks of work, and one missed dependency can derail a production cutover.
• Wave planning. Grouping applications into migration waves by risk and dependency, instead of laying out the schedule in a spreadsheet.
• Code analysis and conversion. For mainframe: extracting business logic from code nobody fully understands any more. For .NET: porting the framework version and rewriting SDK calls.
• Network conversion (VMware). Translating the source network configuration into AWS constructs.
• Validation and transformation planning. Checking that the transformed code incorporates the changes, and sequencing the transformation itself.

Why does this matter for cost? Because a migration is priced mostly in human hours, and these phases, discovery, dependency mapping, mechanical code refactoring, are where most of those hours go. AWS states that AWS Transform makes transformation projects up to 4x faster. That is an AWS figure, not our measurement, so treat it as a marketing ceiling rather than a guarantee. For wider context, a McKinsey estimate cited in AWS documentation puts the time reduction from generative AI at 30-40% for cloud migration. In a published case study, the company Vector, with AWS Partner Slalom, reported cutting VMware Cloud migration time by 34%. These are third-party numbers, but consistent in order of magnitude: the real speed-up is tens of percent on the labour-heavy phases, not a whole project done instantly.

Pricing: agent-minutes as a new budget line

Here is the change you have to understand when budgeting, because AWS Transform bills differently from classic migration tooling. The model is based on agent-minutes: you pay for the time agents actively work server-side, not for user idle time or client-side operations like local builds and file reads. Because several agents work in parallel on one task, the total agent-minutes can exceed the wall-clock time.

From AWS’s published pricing (as of 2026, verify current rates before budgeting):

• Base rate: $0.035 per agent-minute, with a 1-minute minimum billing increment.
• Not charged: user idle time and CLI-side operations (file reads, builds).
• Free monthly quota for AWS-managed .NET transformation: within the quota the rate is $0.00 per agent-minute.
• Scale: AWS’s own example puts a ~50k-line .NET Framework solution at roughly 2,800 agent-minutes.

The practical takeaway: automation lowers the cost of people’s time, but agent-minutes are a new budget line that wasn’t there before. In most projects that is a good trade, because an engineer-hour costs many times more than an hour of agent work, but it is not free and you have to cost it separately. The right budgeting question is not “how much do I save on the tool” but “how much does the total project cost drop when cheaper agent-minutes replace some of the expensive engineering hours”. We give whole-project cost ranges by company size in the AWS migration cost calculator; what matters here is that AWS Transform changes the ratio of labour cost to tooling cost, it doesn’t zero it out.

What AWS Transform won’t do for you

This is the part that decides whether a project lands on budget and on schedule. The agent speeds up execution, but it does not make the decisions or own the outcome.

Three caveats that drive the cost in practice:

• The decision of what to move stays with a person. The agent will map dependencies and speed up conversion, but it won’t tell you that half the estate should be retired rather than migrated. That triage is often the biggest saving in the whole project, and no tool automates it.
• Validation is not testing. AWS Transform validates that the transformed code incorporates the changes made. That is not proof the application does what it should in production. Your test strategy and the cutover don’t leave the schedule.
• Data migration is not code conversion. Moving databases and volumes is a separate step, still handled with tools like AWS DMS, and it is often the bottleneck for a near-zero-downtime cutover.

Where AWS Transform fits in the migration process

The simplest way to think about it is as an accelerator inside the phases we already laid out in the migration process guide, not a replacement for the process.

For VMware estates, AWS Transform plugs into exactly the discovery and wave-planning phases we described in the VMware to AWS migration guide, with the rehost execution itself still run by AWS MGN. For .NET, the modernisation agent covers the framework port and code conversion. For mainframe, it is the analysis and refactoring layer that previously needed a team fluent in COBOL.

What this means for migration funding

There is one more cost dimension AWS Transform doesn’t change but is worth keeping in mind: funding. Migrations from VMware, mainframe, or Oracle often qualify for the AWS Migration Acceleration Program (MAP), which covers 70-80% of a migration partner’s cost. AI-agent automation and MAP funding work at two different levels: AWS Transform lowers the number of hours a project takes, and MAP lowers how much of the remaining cost you pay out of your own budget. Together they can bring the out-of-pocket cost of a migration down to a fraction of the original quote. How partner-led funding works is covered in our article on how to choose an AWS migration partner.

FAQ

What is AWS Transform? It is an AWS service that uses AI agents to automate the migration and modernisation of enterprise workloads: VMware, mainframe, and .NET/Windows, plus a customisable path for non-standard code patterns. The agents run the labour-intensive phases, such as discovery, dependency mapping, and code analysis and conversion, with a human in the decision loop.

Does AWS Transform do the whole migration for me? No. It accelerates the mechanical, repetitive phases, but it doesn’t decide what to move, design the target architecture, migrate data, or test the application end to end. It is an accelerator inside the migration process, not a replacement for a plan.

How much does AWS Transform shorten a migration? AWS markets transformation projects as up to 4x faster, and McKinsey (cited in AWS documentation) estimates a 30-40% reduction in cloud migration time from generative AI. In a published case study, the company Vector reported a 34% cut in VMware Cloud migration time. These are AWS and third-party figures, not our measurement, and the real result depends on the specific estate.

How much does AWS Transform cost? It bills per agent-minute: $0.035 per minute of active server-side agent work, with a 1-minute minimum (as of 2026, verify the current pricing). User idle time and CLI-side operations are not charged, and AWS-managed .NET transformation has a free monthly quota. Agent-minutes are a new budget line, but usually far cheaper than the engineering hours they replace.

How is AWS Transform different from AWS MGN and DMS? MGN rehosts servers to EC2, and DMS migrates databases. AWS Transform works a layer up: it automates the analysis, planning, and code conversion, the conceptual and refactoring work that precedes the move itself. On a project these tools complement each other rather than compete.

Does a migration with AWS Transform qualify for MAP funding? The tool and the funding programme are two separate things. VMware, mainframe, and Oracle migrations often qualify for AWS MAP regardless of whether you use AWS Transform. Automation lowers the hours, and MAP lowers your share of the remaining cost.

Summary

1. AWS Transform uses AI agents to automate the labour-intensive phases of VMware, mainframe, and .NET migrations: discovery, dependency mapping, wave planning, and code analysis and conversion. Those phases took the most hours, so automating them genuinely moves the project economics.
2. The agent-minute pricing model is a new budget line. Cheaper than the engineering hours it replaces, but not zero. Budget the whole project cost, not just the saving on the tool.
3. Automation does not replace the decisions or the ownership. What to move, the architecture design, data migration, and end-to-end testing are still yours, and that is where most of the risk sits.

If you are planning a VMware, mainframe, or .NET migration to AWS and want to work out how much automation would actually shorten it, and where a plan and funding are still needed, explore our cloud migration guide or talk to us directly.

Planning a VMware, mainframe, or .NET migration to AWS?

Book a free 30-minute call. We work out which phases AWS Transform genuinely accelerates, design the target state and the data migration it doesn't cover, and check whether the move qualifies for MAP funding.

Book a call Send us a message