Leading Global Retail and Commercial Bank Extracts Legacy COBOL Using Slingshot on Google Cloud

A leading global retail and commercial bank needed to modernize legacy mainframe systems supporting core banking services, spanning 300+ critical batch feeds. But the logic behind these systems was embedded in decades-old COBOL programs, making it difficult for teams to understand how critical services interacted.

Using Sapient Slingshot on Google Cloud, the bank analyzed nearly three million lines of legacy code and extracted the business rules embedded within it. The result was a clear, traceable set of specifications that accelerated modernization planning and reduced delivery risk. 

In just eight weeks, we helped the bank extract and verify the business rules in the code, turning them into clear specifications and modern designs, significantly accelerating the transition to a modern architecture. 

Modernizing the bank’s core systems required understanding how hundreds of legacy COBOL programs interacted across 300+ batch feeds and services.

In practice, teams had to manually trace thousands of lines of interconnected code across programs to understand how the systems behaved. Even small changes required significant investigation to ensure existing business rules were preserved.

This process slowed delivery and made modernization difficult to scale. Engineering teams could spend weeks validating system behavior before making changes, delaying modernization initiatives and increasing operational risk.

Because these systems support critical customer payments, the bank also needed clear traceability between legacy code and modernization designs to demonstrate compliance and provide regulators with evidence that all functionality remained intact. 

Slingshot’s code to specification (code-to-spec) capability analyzed nearly three million lines of legacy COBOL across hundreds of programs and batch feeds.

Using Gemini, the platform dynamically extracted business rules embedded within the code and converted them into structured, reviewable specifications. Slingshot generated program overviews, process flow diagrams, field mappings and business-readable documentation that made previously hidden system behavior visible. 

Rather than simply summarizing code, Slingshot’s dynamic business rule extraction process identifies the underlying logic within legacy programs and converts it into structured specifications. This creates traceability between the original code, extracted rules and future modernization designs. 

The verified specifications were then used to generate 200+ implementation-ready backlog items, helping teams move more quickly from legacy analysis to modernization planning. 

The results were measurable and immediate:

• 95% specification accuracy through dynamic business rule extraction (DBRE)
• 70-85% reduction in code to specification effort
• 50% reduction in effort from specification to design
• Analysis per feed reduced from 35 days to 5 days
• Nearly three million lines of code converted into verified, documented specifications in eight weeks

Beyond the metrics, the bank gained clear visibility into system behavior, reduced operational and compliance risk and established a predictable roadmap aligned to its upcoming technology commitments.

Explore how AI-powered Code-to-Spec helps you uncover, understand and modernize legacy systems with speed and confidence.