| Code to Code |
Tools like Cursor, Calude, Copilot, Devin can be used with right prompts, contexts to convert code from legacy to modern applications. This is fairly close to vibe coding and usually discounts the business knowledge required to migrate. |
Small but more techincal oriented applications can use this approach. Should not need business background as this is programmer centric approach |
Pay attention to security, testing and maintainability. Usually the typical problems of vibe coded applications also apply to this approach, including requiring human verification and testing. Require some mastery of AI assisted coding |
| Platform Specific |
Tools that are specifically desinged for platforms like mainframe. They can assist the technical and business teams, by providing critical information like, how to break up the monolith, what contracts to comply for input and output etc.
The models are potentially pre trained for the language thus providing accurate understanding. This approach depends heavily on the vendor's expertise and knowledge of the legacy system.
|
Fits large and critical applications that match platform technology, like mainframe. |
These tools heavily require vendor experts to operate the tool and hence need continous engagement of tool and associated services. Customer can not do it on their own. |
| Documentation Generation |
Tools that generate code explanation documentation. This documentation is expected to be read by systems analysts, and generate specifications for technical team. |
Most used by companies today. Works for any type of project. Zero vendor lock in. Low risk and eliminate need to guessing or reading code. |
The productivity improvement is minimal, as most such tools generate more technical style documents and fits programmers and ignores the important underlying business processes. |
| Specification Generation |
Tools that generate detailed user scenarios, when business users provide explanation of business process or describe what they want to build. Technical teams can then convert scenarios into software |
Small and simple applications |
Useful for improving business analyst productivity |
| Petals |
Generates detailed specifications from legacy code. Embeds nicely into company's development processes. Provides RAG layer to help train and relieve from SME depenendency. Learns continuosly and 'Intelligence' becomes the asset, not documentation. |
All kinds of applications. Particularly suited when companies prefer no lockin and self operable. Zero risk of failure as human in loop is the fundamental design |
Iris tool from Petals.AI is a business aid for analysts and users. Legacy modernization is a critical use case, however can be used as a training aid for new employees, SME/Guru replacement as well. |