AveriSource Enables Rapid AWS Modernization for a Major U.S. Bank

The Customer

One of America’s top financial services institutions responsible for high-volume transactions and critical workloads.

The Challenge

After several false starts, thefirm had to unearth and document every business rule hidden inside more thanhalf a million lines of legacy COBOL and assorted copybooks, CICS tables (PCT, FCT,PPT), JCL, PROCs, control card files, symbolic maps, security profiles, and BMS macros that held the application together.

Success depended on four non-negotiables:

• Translate cryptic code into plain English. Thousandsof domain-specific variables and abbreviations had to become readable for business analysts.
• Build a living data dictionary. Analysts needed clear lineage from screen input to transaction output while filtering out dead code.
• Produce low code models for Java. The team wanted automatically generated, standards-compliant starter services that would deploy cleanly on AWS.
• Minimize the burden on SMEs. Prior attempts collapsed under the weight of manual overhead, this pass had to cut human effort significantly to stand a chance.

 All of this had to happen on anaggressive timeline so downstream teams could begin rewriting for the cloud.

The Approach

Source Code Processing

Inventory & Discover

Using the AveriSource platform, the project team mapped the entire environment in hours rather than weeks. The platform identified every language in play, flagged missing or unused objects,mapped all entry points, and produced a baseline inventory that project leaders could trust for resource and schedule planning.

Rule Analysis & Business Process Mapping

With AveriSource Analyze, business analysts (many with no prior mainframe background) navigated intuitive rule trees, routine flow and control flow diagrams, annotated source views, and automatically generated data lineage reports. Rule-chaining capabilities stitched individual rules into complete end-to-end logic, showing exactly how a transaction moved through the system.Meanwhile, auto-generated call-chain diagrams tied code back to real world business functions, highlighting redundant processes that could be retired.​

Rule Extraction Methodology

AveriSource’s proven methodology married process mapping with unique modernization artifacts, allowing analysts to extract requirements ten times faster than manual review. The result was a business-centric view of the codebase that dramatically reduced technical debt and blind spots while surfacing commodity processes ripe for elimination.​

Business Rule Chaining & Dependency Analysis

With the legacy environment fully cataloged, the AveriSource Analyze rule-chaining engine was used to connect the dots between thousands of scattered logic fragments. Instead of reviewing rules one program at a time, analysts could trace a single variable or business concept end-to-end—across COBOL paragraphs, CICS transactions, JCL steps, and control card parameters to determine how each decision flowed into the next.

Since each chained rule was annotated in plain language and linked to its exact code location, subject matter experts could validate intent in minutes rather than days. The result was a step-change in discovery speed, where documenting complex business logic could be accomplished ten times faster than prior manual efforts. It also provided a rock-solid foundation for generating clean, consistent Java in the subsequent transformation phase.

Low Code Model Generation

Finally, AveriSource Transform turned the cleansed logic into production ready Java microservice skeletons complete with REST endpoints and standardized naming conventions, giving developers a significant head start once the application moved to AWS.

The Results

• 10× faster business-rule documentation versus prior manual attempts.
• 50% reduction in hands-on SME effort, freeing experts for higher-value work.
• Full coverage of 500K+ lines of COBOL and associated assets, eliminating migration blind spots.
• Automatically generated Java templates accelerated cloud-native development.
• Successful modernization of this application after multiple failed attempts.