DXC XACT is a proprietary automated conversion tool that reads business logic directly from the CA Gen encyclopedia — the source of truth — and converts it to clean, maintainable Java or C#. Unlike approaches that attempt to re-engineer compiler output or rely on generative AI, XACT works from the original models, preserving decades of business logic exactly.

The output is cloud-native, API-first, and containerization-ready from day one, with support for modern CI/CD pipelines, QA automation, and microservices architectures. Legacy GUI and TE3270 interfaces are simultaneously transformed to modern front-end web pages.

CA Gen does not store applications as conventional source code. Business logic lives in a proprietary encyclopedia of structured models that only CA Gen's compiler can read — not as Python, COBOL, or any format that large language models have been trained on at scale.

This creates three compounding problems: sparse training data on CA Gen's action diagram syntax and entity relationship notation; an inability to reconstruct the 'why' behind decades of undocumented business rules; and context window limitations that cannot hold a large, tightly coupled CA Gen estate.

XACT bypasses these limitations entirely by reading directly from the encyclopedia rather than from compiler output. This is why XACT achieves 100% automated conversion where generative AI tools produce unreliable, incomplete results that require extensive manual correction.

The assessment is a three-week, fixed-price engagement that gives organizations a complete picture of their CA Gen estate before committing to full conversion.

DXC analysts map every action block, dependency, flow and architectural touchpoint in the estate, then produce a detailed modernization roadmap and validated business case. The output includes a clear scope, conversion timeline and cost projection — structured so clients can present the investment case internally.

Clients can exit after the assessment with no obligation to proceed. This matters because it separates discovery from commitment: organizations facing an upcoming Broadcom renewal for CA Gen can run the assessment within the current license cycle and use the findings to negotiate or plan an exit before the next renewal hits.

One of the largest US healthcare payers — serving over 100 million customers — migrated its customer-facing system backend and mid-tier from CA Gen on zOS to a native Java stack on AWS using XACT. The engagement converted 11.4 million CA Gen objects and 36,000 action blocks, generating $50 million in five-year cumulative savings and a projected $20 million five-year revenue increase through improved agility.

A second US healthcare insurance provider used XACT to modernize its core administration platform on Azure Kubernetes Service. The project eliminated CA Gen vendor lock-in, increased release frequency from 2 to 26 per year, mitigated a 10x licence cost increase risk, and avoided approximately $350 million in costs that a conventional RIP and replace approach would have incurred.