IBM is deploying AI and building tools that can help clients move away from traditional coding languages including COBOL. 

Common Business Oriented Language — COBOL — became the bedrock coding language for many industries through the 1980s due its deterministic outcomes, Jeana Bolanos, president of AI implementation and consulting firm Sales E, previously told FinAi News. 

COBOL, which runs on mainframe architecture, provides banks with high reliability and resiliency, IBM CEO Arvind Krishna said during an interview with Bank of America Securities today.

By using COBOL, banks can ensure that outages last only milliseconds to seconds for the entire year, which is difficult to replicate in other server and coding language architectures, Krishna said. 

“I think that if you try to replicate that level of resilience on alternate architectures, your price would be three, four, five times of what it would be on a mainframe architecture,” he said. 

The alternative 

One alternative to COBOL is Java, SunWest Bank’s Chief Technology Officer Ben Xiang previously told FinAi News. AI can help FIs in the discovery and documentation phase of legacy code infrastructure. 

IBM has already made a move on that. 

IBM launched Watsonx Code Assistant for Z (“Z” represents mainframe), which can translate COBOL to Java, Krishna said. 

The tool can help FIs document “all kinds of spaghetti code where people have undocumented code — and you have no idea. ‘Is this something stubbed out or is it real’ — and it can comment it for you,” Krishna said. 

ALSO READ: Transitioning away from COBOL with AI 

In the past 40 years, COBOL has captured business logic operations in which deterministic outcomes are needed, Krishna said. 

“If you’re going to replicate it, you’ve got to make sure that that business logic is correctly there in an alternate language,” he said.  

“When your issue becomes that a single mistake in a million or a billion transactions a day gets the regulator breathing down your neck, you’re going to have to be really careful that what is translated is 100% accurate, not 90% accurate, not 95% accurate,” Krishna said. 

AI and software development 

AI tools can also help generate and convert AI code..  

IBM software developers can garner 30% to 70% productivity gains using agentic AI, Krishna said, adding that AI will help software developers to create new products quickly and phase out older ones. 

“We actually believe there’ll be a billion new applications leveraging agents and LLMs that are going to get created over the next five years,” Krishna said. “So clearly, if that billion gets created, it’s replacing some of what is there now.” 

Read coverage from FinAi Banking Summit here.