Contact us
For a business with an old .NET, Java or COBOL system and a license renewal coming. AWS Transform’s AI agents read and convert the code, and our engineers prove it still behaves the same, in your own AWS account.
A one-day workshop with the engineers who’d do the work. AWS funding typically covers the workshop.
An AWS Premier Tier partner, one of about 300 among more than 130,000, with the AWS Migration and Modernization Competency.
An old mill whose loom runs on punched cards nobody can read: AI agents read them card by card, our engineers prove the new cloth matches the old, and the rent on the old mill stops.
The renewal notice arrives: $38,100, due on 1 March. It pays for an order system written in 2012, and one of the two people who wrote it has left. Nobody can say what all its pricing rules do, so nobody will touch it. The plan to replace it gets scoped in years, the business case dies in committee, and the renewal gets signed again.
Every Windows Server and SQL Server license is rent, paid on code nobody has changed in years.
That part just changed. AWS Transform’s AI agents now do the reading, the documenting and the converting that used to eat the first year. AWS says a full-stack .NET and SQL Server modernization runs up to 5x faster, and mainframe programs take months, not years.
One order system, ten renewalsRenews 1 March
paid to keep it running,
2017 to 2026
Example
AI agents read it first. A person on your side signs off what it does. And some of it never has to move at all. Six examples.
A food distributor, 120 people
Its SQL Server licenses renew on 1 March.
AI agents read the 214 pricing and order rules in its 2012 .NET order system into plain words.
The sales director signs the pricing rules.
A software company, 45 people
Every new customer adds to its Windows bill.
AWS Transform ported its product from .NET Framework to .NET 8 on Linux, and SQL Server to Aurora PostgreSQL, the stack together.
Its engineers approve every pull request.
An insurance broker, 70 people
Its quoting runs on Java 8, out of support, and the security audit flagged it.
One upgrade to Java 21 rolled across all 38 of its repositories at once.
The head of underwriting signs the rating rules.
A payroll bureau, 90 people
The two people who wrote its payroll retire next year.
AI agents extracted the tax and pay rules from its COBOL, on a mainframe since 1994. Payslips moved first, and three domains stay where they are.
The payroll manager signs each tax rule.
A building supplies merchant, 160 people
Support is ending for the servers under its seven .NET modules, bolted onto an old ERP.
AI agents found which modules anyone still uses.
The operations director decides what goes.
A coach operator, 200 people
Its booking and ticketing run on Java 8 and a Spring two versions behind, and the security list grows every week.
One Java and Spring upgrade rolled across every repository at once.
Its engineers review it like any other commit.
Examples
AWS Transform’s AI agents read, document and convert. They don’t decide what’s worth converting, write the tests you never had, or settle who owns a rule nobody wrote down. That’s the work of our forward deployed engineers (FDEs), who work inside your company, in your own AWS account.
AWS Transform’s assessment lists every application, language, framework and dependency, so the dependency map is generated, not guessed.
Why:Nobody pays to convert code nobody needs, and a license you can drop goes on day one.
AI agents read the code and write its business rules down in plain words. On a mainframe, AWS Transform extracts and documents the business logic.
Why:A rule nobody wrote down can’t go missing in the new code, and nobody argues later about whose rule it was.
Characterization tests record what the system does today, right or wrong. We build them from real inputs: last quarter’s orders, replayed through the old code.
Why:Converted code can’t behave differently at month-end without a test failing first.
AWS Transform does the bulk. It ports .NET Framework to .NET on Linux and moves SQL Server to Amazon Aurora PostgreSQL, the whole stack together. It rolls one Java upgrade across every repository at once, and breaks a mainframe into domain services.
Why:No year-long rewrite, and no AI-written code reaches your customers unchecked.
The new piece goes in behind a routing layer. Traffic moves across a slice at a time, a group of customers or one kind of order, and if something goes wrong the route goes back.
Why:There’s no switch-over weekend.
Every converted app ships the same way: build, test, scan, deploy and roll back, all as code, on AWS CodePipeline and CodeBuild.
Why:Tech debt never piles up into another modernization project five years from now, and the last license goes, not only the first app.
The same order system, decision by decisionOn the golden path
renewed on 1 March
Build, test, scan, deploy and roll back, all as code, and the 1 March renewal not signed.
Example
What a client said
Armakuni were the engineering partner who made the AWS migration work. They came in with a clear plan, ran it cleanly, and the team that came out the other side had levelled up.
Marcus East, former CTO, Experian
The workshop is Estimate your legacy modernization with AI, with the engineers who’d do the work. Bring your engineering, architecture and platform leads, and whoever signs the license renewals. Your code is optional. AWS funding typically covers the workshop.
The ranked backlog you leave with8 apps, ranked
a year off
the next renewal
Example
60%
faster deployments
4.2x
more releases
$2.1M
saved in year one
Medians across AK Evolve clients, as published on armakuni.com. Behind them are 1,300+ engineers and more than 500 customer launches on AWS.
Direct to consumer: about 1,000 policies and 2,000+ quotes a day, on two production servers
An outage on its legacy LAMP stack kept coming back for six to ten months. Every restart erased the evidence, and with no observability every incident was a guessing game.
In a 60-hour engagement we kept the evidence from the last outage and found the cause: Apache thread exhaustion, about 800 MySQL queries per request. Connection pooling and session timeouts fixed it without putting policy binding or payments at risk, and request tracing and Amazon RDS Performance Insights closed the visibility gap. The retainer continues.
Automotive logistics in North America, hundreds of thousands of finished vehicles a year
Dealers phoned operations to chase delays, which were found after they had cost the customer. Shipping costs ran on weekly reports, looking back.
In a fixed-scope modernization over six to nine months we built Glovis America’s first real-time dealer tracking portal: a microservice backend with a bridge to its Oracle data, an IoT telemetry pipeline and automatic exception alerts. Shipping costs fell 15% and data handling capacity rose 40%. Dealer status calls fell sharply, and we still deliver against the roadmap.
Got another? Bring it to the workshop.
What stalled it was usually the reading: people working out by hand what code nobody understands. AWS Transform’s AI agents now do that part. The rest we keep small: one app at a time, the tests written first, and a go or no-go before each piece goes live. Nothing switches all at once, so no other work has to pause.
AWS funding typically covers the workshop. AWS says its AWS Transform AI agents for assessment, Windows and .NET, and mainframe carry no usage charge. After that, the licenses you retire pay for the next phase. The roadmap maps AWS funding to each phase where a workload qualifies. AWS decides what it funds, and we put the case together with you.
Then the next renewal is the date to plan to. The workshop prices what each app costs to keep, and the roadmap lands the move before that date, starting with the apps you can switch off.
The workshop decides which domains move first, and which never move. Nothing changes without tests that prove it still behaves the same. The old path stays until the new one has carried real work.
Good, and plenty of teams should. The workshop decides what to point it at, and we can build the tests that prove what comes out. Your engineers spend their time on the rules while the AI agents do the reading. Whatever we build stays in your AWS account: the converted code, the tests, the pipeline and the runbooks.
One day with the engineers who’d do the work, and whoever signs the renewals. You leave knowing what your old code does, what to switch off, and a costed plan to move the rest before the date. AWS funding typically covers the workshop.