Industry executives and experts share their predictions for 2024.  Read them in this 16th annual VMblog.com series exclusive.

Application Modernization in 2024: Bridging the Gap between Expectations and the Reality of GenAI

By Miten Marfatia, Founder and CEO, EvolveWare

Enterprises are well aware that they need to modernize — legacy programmers are increasingly hard to find, outdated legacy systems pose a bigger and bigger risk as time goes on, and organizations want to capitalize on  new technologies. Meanwhile, the excitement around GenAI has permeated every industry, and application modernization is certainly no exception. The following predictions explore the impact GenAI will have on application modernization in 2024 — from high expectations to understanding obstacles and realities through greater experimentation.

High hopes for GenAI will drive a surge in application modernization activity, but overoptimistic expectations will get a reality check. Modernization will be as urgent as ever next year – due to the shrinking talent pool, the advent of AI, and dangerously antiquated legacy systems. But there will be a key difference: high hopes that GenAI will reduce modernization cost and time will significantly boost enterprises’ appetite for these initiatives during a period of economic uncertainty. The resulting scramble to experiment with GenAI for more efficient modernization will expose its limits – GenAI technology will not be ready to make a measurable impact on modernization in 2024, and that reality check will take hold within the first half of the year.

Enterprises will hesitate to fuel LLMs with their own code, curbing the impact of GenAI on modernization efforts. Though interest in applying GenAI to modernization will surge next year, enterprises will be hesitant to supply their own code to train LLM models, due to security concerns and the fact that their software’s code represents their intellectual property. This hesitation will significantly limit the near term impact that GenAI will have on modernization processes, given that any GenAI-enhanced modernization technology would require the massive amounts of legacy code that reside within organizations to properly train a model and thus achieve accurate and useful results. 

GenAI experimentation for application modernization will start with code documentation and source code transformation. With the advent of AI-powered tools over the last few years and the promise that GenAI will further streamline modernization efforts, organizations will aim for increasingly complex modernization strategies such as refactoring their monolithic applications and creating microservices in 2024. GenAI models for application modernization will first be developed in areas where significant data is available for modeling, likely starting with documentation of legacy applications where code is translated to a plain English description for use by business personnel, and transformation of source code to modern code.

The breadth of impact by implementing AI models will depend on the existing capabilities of each modernization tool. Tools that generate the same syntax pseudocode from source code written in multiple languages will provide the greatest impact by developing models using pseudocode to generate English descriptions of the source code. These same models will also be useful in translating extracted rules in pseudocode format into rules with descriptive English summaries and relevant details. From a code transformation perspective, significant advantage will lie with tools that provide refactoring capabilities prior to transformation. Quality of GenAI models is dependent on the quality of data used for modeling and using refactored data will result in high quality and efficient modern code being generated from GenAI models.

Conclusion

Hopes are high for the promise of GenAI and application modernization, and we’ll certainly see progress by way of experimentation. But will GenAI make a material impact, i.e., substantial savings in cost and time, on modernization efforts? Not likely in 2024. Organizations must have realistic expectations on when GenAI will have a reliable and useful role in application modernization initiatives. The good news is that well-established modernization tools with a very high level of supervised AI already exist, and depending upon the phase of modernization that is being conducted, these tools are already able to provide a 40-90% reduction in effort and cost. So even as enterprises aim to strike the balance between future aspirations and current applications of GenAI, successful modernization is already within reach.

##

ABOUT THE AUTHOR

Miten Marfatia is the founder and CEO of EvolveWare, a global leader in automating the documentation, analysis and modernization of software applications. An early player in the application modernization industry, Miten has helped revolutionize the approach to modernization by incorporating automation and ML into a unique platform that is successfully guiding large organizations such as the State of New York, U.S. Dept. of Defense, and Deloitte Consulting clients through complex digital transformation initiatives. Prior to EvolveWare, Miten was a founder of Perisol Technology, which provided state of the art storage solutions to corporations and government agencies, and prior to that, founded Silicon Electronics, a distributor of cutting-edge products to India’s nascent computer manufacturing industry.

Miten holds a M.S in Electrical Engineering and a M.B.A in Finance, Investment & Banking from the  University of Wisconsin at Madison. He earned his B.S in Electrical Engineering from the University of Bombay in India.