Push Is On to Use AI Tools to Modernize Legacy Systems
Millions of lines of dated languages such as Cobol that run core banking applications, being eyed as a market for AI-modernization tools to ready the application suite for - more AI
By John P. Desmond, Editor, AI in Business
AI is ready to work with legacy software systems, a growing consensus indicates.
In banking, for example, IBM estimates 70 percent of global transactions are processed on mainframe systems running Cobol, a programming language dating to 1959. Another estimate put the total number of lines of Cobol in daily use at about 800 billion.
“AI is ready to play with legacy systems,” stated Martin Prescher, CTO of Autonomy and a member of the adjunct AI faculty at USC, in a recent account he posted on LinkedIn.. “What is different is that over the last decade the concepts behind AI and ML have been bolstered with unbelievably sophisticated toolsets (some free, many paid for) that make it possible to integrate with existing, and not very sophisticated/old school IT ecosystems.”
Further, “Unlike in the past, machines can not only follow algorithms and deterministic rules, but [they can] learn from the past and historic data to emulate human-like understanding and decision-making,” he stated. Also, “The new kids on the block are impressive models that emulate human-like creativity through large language models like GPT.”
AI toolsets are becoming more mainstream and easier to integrate with legacy systems, which means real projects need to get underway, beyond proofs-of-concept and pilots. Use cases Prischer is working include: auto-decisioning and customer qualification, identifying true patterns of risk levels and data sources needed to underwrite; next-best action tools that optimize sales and customer pull-through, helping loan officers and customer service agents be more effective; document recognition and contextual understanding, to help eliminate mountains of paperwork; and dynamic pricing, for novel and complex services such as usage-based insurance.
Legacy modernization is a target market for a growing number of firms offering AI-powered software and services.
IBM, for example, this week announced watsonx Code Assistant, a generative AI-powered assistant tailored for very specific use cases. “One of those use cases is converting decades-old code that run on IBM’’s mainframe systems to a more modern programming language,” stated a recent account in The Motley Fool, with a headline describing the new watsonx tool as a “game changer.”
“This represents an enormous opportunity for IBM to keep its mainframe systems relevant and generate additional revenue from its enterprise customer base,” the article author stated. The difficulty with Cobol is it is not widely taught in universities anymore, and is viewed as a bad career move for developers. “This puts companies running Cobol-powered workloads in a bind.” the author stated.
The watsonx Code Assistant for Z uses generative AI to map out a Cobol application and its dependencies, split it into modular parts, and convert the parts into modern Java code. Support for validation testing is planned for a future release, the account stated. This gives mainframe customers a path to modernize without abandoning the mainframe platform, It also enables IBM to continue and expand its relationship with mainframe customers, and oh by the way, IBM’s consulting business can help customer identify which applications to modernize and how to do it in the secure manner important to banks.
Survey Shows Legacy Apps Slow Move to AI
Moving to AI is the chief reason to modernize the computing infrastructure, and reliance on legacy systems is slowing down the transition, according to results of the Rackspace Cloud Modernization Survey 2023, recently reported in RT Insights.
Nine out of 10 respondents said moving to AI was the main reason to modernize.
The average responding company reported having 22 legacy systems or applications currently in use, and over half of them have not considered moving away from them, due to their critical necessity and the difficulty of replacing them.
“It is telling that even well into their cloud journey, the three most critical apps organizations say they need to upgrade are truly at the heart of the business, because they are at once the most important things to modernize and the most challenging,” stated Jeff DeVerter, chief technology evangelist at Rackspace Technology. “At the same time, it could be that the prospect of losing out on AI will motivate organizations to finally get off the sidelines when it comes to modernization of core systems.”
Luxoft, Unisys Among Companies Offering Legacy Modernization
One company offering services to modernize legacy systems is Luxoft, which provides software engineering services to its 425 global clients, 75 of them in Fortune 500 companies. In one example project called out on the company website, a mission-critical Cobol/IMS mainframe application for a Fortune 500 company was migrated. The operating system was switched from z/OS to RHEL; JCL was auto-transformed to Python;and IMSDB was transitioned to SQL Adapter; and performance was optimized with a GnuCobol compiler.
Another competitor in the legacy modernization market is Unisys, which has worked with Amazon Web Services as a managed service provider for six years.
Unisys positions for “AI-augmented application modernization,” with a suggested focus on updating the documentation for existing systems as a first step. A post on the Unisys website states, “In any company, the foundation for consistent growth and innovation relies heavily on comprehensive and up-to-date software documentation …. These older platforms often operate as the backbone of an enterprise, but are generally not well documented.”
Mohan CV, a principal solutions architect with Amazon Web Services (AWS), is quoted on the Unisys website stating, "Utilizing tools that leverage capabilities like generative AI can eliminate the heavy lifting associated with coding, enabling faster development and allowing developers to focus on the more creative aspects.”
The application modernization market was sized at $8.3 billion in 2022, and is expected to grow to $28.7 billion by 2030, according to 360 Research Reports.
Read the source articles and information from Martin Prescher on LinkedIn, in The Motley Fool, in RT Insights and from 360 Research Reports.
Click on the image to buy me a cup of coffee and support the production of the AI in Business newsletter. Thank you!