Astadia: Change the machine. Keep the business.

On a Friday afternoon, an insurance company handed over a sample of its old code. By Monday, according to Astadia’s account, that sample had been converted into Java. The weekend made an impression. But the interesting part came afterward: a proof of concept, an outside review, and an agreement to modernize six mainframe systems in stages. A demonstration opened the door. Evidence kept it open.

A weekend is not a migration

The insurer had already spent two years discussing modernization with large IT providers. Astadia says their proposals brought unwanted compromises: generic conversion approaches, extended code freezes or cumbersome ways of incorporating changes. The insurer needed to move its systems while continuing to develop them. Pausing the business was an expensive way to improve it.

This is Astadia’s territory. It works on IBM and Unisys mainframe applications and databases, combining specialist services with software that converts code, moves data and tests the result. The central promise is functional equivalence: the new application should perform the same business work as the old one. A different programming language should not quietly produce a different insurance policy.

The attraction is easy to understand. An old application may contain years of exceptions, contractual rules and operational knowledge. Its documentation may be less thorough. Reusing that logic offers a route between continuing to maintain the existing platform and commissioning a fresh implementation of everything it does.

“We believe in maximum impact with minimum intervention.”

Astadia’s published value of simplicity

A factory with an unusually strict inspector

Astadia calls its approach the FastTrack Factory. The name describes an effort to make conversion repeatable rather than an exercise in heroic manual programming. Its 2021 acquisition of Belgian migration specialist Anubex added refactoring capabilities to its existing replatforming business. Today, four tools explain much of the machinery.

CodeTurn translates source code into a target language, with rules that can be configured for a particular environment. DataTurn converts the accompanying data structures and contents into relational databases. Moving one without the other would be rather like relocating a restaurant’s kitchen and forgetting the ingredients.

TestMatch records transactions and replays them against the migrated application. It compares the recorded and replayed sessions and reports response times and throughput. Its value extends beyond migration: those recordings can become regression tests for later changes. The people who know how the application works can help demonstrate what “working” actually means.

DataMatch tackles batch results and database changes. It can compare an initial database state with the outcomes of the original and converted jobs. That third reference matters: a missing record could mean the new program failed to insert it, or deleted it incorrectly. The final states alone do not always explain the mistake.

When comparisons expose differences, FastTrack’s methodology calls for tuning transformation parameters, customizing tools or supplying missing source artifacts, then testing again. Production changes can enter later snapshots. The practical idea is a repeatable conversion loop that accommodates continuing maintenance, rather than a one-off translation everyone hopes will survive contact with reality.

The cloud is a destination, not the definition

One of Astadia’s bank accounts offers a useful surprise. The bank kept its business applications on the mainframe. It converted more than eight million lines of Natural code, used across branches in 15 countries, into Enterprise COBOL. Modernization meant changing the software stack, without changing the underlying platform.

Astadia reports that the project took 20 months and involved no more than 30 IT experts from both organizations. The bank achieved a return on investment within two years after completion, according to the account. That is a specific result for a particular project; it is not a timetable that comes free with the software.

Replatforming and refactoring solve different problems. Replatforming adapts applications to another operating environment, often retaining their language. Refactoring can change languages and data technologies. Astadia offers both. A buyer can preserve COBOL, convert to Java or C#, or make a narrower change where that better suits the workload.

The bill depends on what you are saving

Miami’s emergency dispatch system faced a choice between reusing its applications and rewriting them. Astadia’s historical case describes 80 applications and 180,000 lines of code. A rewrite was estimated at two to three years and more than twice the cost of rehosting. The selected modernization took six months.

With Micro Focus, Astadia converted Unisys COBOL for a Windows and .NET environment and moved the database to SQL Server. Its account reports a reduction of more than 50% in mainframe costs, saving at least $900,000 annually. Those are operating savings, not the migration invoice. The useful comparison is the cost of preserving a working application against the cost of recreating it.

For a prospective customer, the starting point is an assessment. Astadia’s AWS Marketplace listing offers workload inventories, complexity analysis, interfaces, dependencies and risk reports, with custom pricing. A sensible business case counts the conversion, testing and transition alongside future operations. The cheapest-looking destination can become expensive if reaching it requires rebuilding half the business.

First, discover what the system actually does

At an unnamed automotive retailer with more than 900 stores, the difficulty began with limited understanding of its own applications. Discovery uncovered unexpected conditions left by old code changes. Documentation, dependency mapping and knowledge transfer preceded the migration to AWS and Micro Focus COBOL.

The case illustrates what can become fragile before a migration even starts: the organization’s knowledge of its system. Astadia reports that deployment time fell from an average of three days to about an hour after introducing the pipeline and testing tools. The improvement followed both discovery and delivery work; changing the hosting address alone would be a thin explanation.

A $75 million piece of a larger business

Amdocs completed its acquisition of Astadia on November 2, 2023, for approximately $75 million net in cash, with potential additional consideration tied to performance. That disclosed purchase amount is different from a current valuation. It explains why Astadia now sits inside a much broader software and services organization.

Amdocs’ 2024 annual report linked the acquisition to a five-year AT&T mainframe migration and operations agreement. In September 2025, Astadia announced its place in Amdocs Studios’ Cloud Studio. Conversion tools can now be paired with a wider set of engineering, testing and managed services. In April 2026, Astadia reported Amdocs’ recognition as a global Leader in ISG’s mainframe solutions evaluation.

The business model combines enterprise migration software, project delivery and ongoing support. Tools are also available to partners and clients. In a market that includes broad providers such as Accenture and Atos, Astadia’s emphasis is the mechanics of legacy conversion and validation. Its CodeTurn description says converted applications are independent of proprietary Astadia runtime software - a concrete point for buyers to investigate.

Borrow the discipline before buying the machinery

The transferable lesson is straightforward: inventory the system, record representative behavior, choose the smallest useful change and make comparison repeatable. Ask a vendor to prove a relevant slice of your application, including its data and interfaces. A rapid code sample is encouraging; a tested, integrated workload is a better decision aid.

The approach fits when existing business logic is worth keeping. If the process itself needs replacement, faithfully reproducing it may preserve the very limitation a company wants to remove. Complete source material, representative tests and a viable target design are prerequisites. Functional equivalence also needs separate checks for performance, security and integration. Software can tell the same story in a new language; the buyer must decide whether it is still the right story.