Introduction
Many banks, insurers and governments still run their most important systems on COBOL and mainframes. These systems work, but skills are retiring, costs are high and changes are slow. TCS MasterCraft TransformPlus is TCS's tool for modernising such legacy applications: it analyses legacy code, extracts business rules and helps convert applications to modern languages like Java. With generative AI, two product directions are possible. One is to push for fully automated conversion, where code goes in and Java comes out with minimal human effort. The other is AI-assisted, human-in-the-loop modernisation, where the tool proposes changes and engineers review, refactor and redesign. The question asks which the product should push. I'll clarify, identify the trade-off type, build product understanding, form a hypothesis, choose metrics, design an experiment, plan the analysis, set a decision framework and recommend.
In trade-off questions, the answer is rarely "one or the other forever". Strong candidates define the conditions under which each option wins, test it, and recommend a default with clear exceptions.
Step 1
Clarifying Questions (3 minutes)
- Who are the clients? Assume large banks and insurers with 5–50 million lines of COBOL, often existing TCS clients.
- What does "fully automated" mean? Automated translation of COBOL to Java with automated test generation, aiming for functional equivalence with little manual coding.
- What does "human-in-the-loop" mean? AI proposes code, documentation and refactoring; engineers review and redesign into modern architecture (for example, microservices).
- What is the business goal? Win more modernisation deals and deliver them faster, with fewer production incidents, while protecting margins.
- Constraints: Regulated industries; zero tolerance for functional errors in calculations such as interest or premiums.
Practice similar questions
Subscribe to access the full answer