COBOL has often been likened to asbestos. Just as that toxic material is extremely difficult to remove from old buildings, COBOL is deeply embedded in massive, vertically integrated mainframe stacks, and replacing it involves redesigning everything from the data architecture to the storage and transaction processing. Rewriting COBOL manually is a years-long effort that most organizations keep putting off because of the expense and the massive risk involved.

Hypercubic wants to convince those companies that its autonomous AI agents can finally solve their problems by automating the entire COBOL modernization process. They’re powered by models that have been trained on massive volumes of COBOL code and they also know how to rewrite all of its specific functions in modern programming languages.

When they start out with a new COBOL application that needs to be modernized, they map the entire codebase first of all. While doing this, they recover all of the business logic that’s buried inside that code, and then use this information to automatically generate the documentation for it, which has never previously been written.

“The problem is not simply translating COBOL syntax,” Gurrapu said. “The real challenge is recovering decades of hidden business logic, understanding how systems behave in production, and proving that a modern replacement preserves what matters.”

Using this AI-generated documentation, it then becomes much easier to generate the modern code needed to create an identical replica of the original app, complete with all of its historical data. According to Gurrapu, Hypercubic’s agents can completely modernize legacy apps in just a couple of months, as opposed to the years required for manual transformation. “We are building agents to make that work dramatically faster and more reliable,” Gurrapu said.

The startup is already working with one of the largest banks in Latin America, as well as a leading Caribbean retailer that serves more than 50 million customers annually, to revamp their most critical applications. With today’s funding, it’s going to invest in building out the capabilities of its AI agents and also expand its capacity to serve even more enterprise customers.

Image: Hypercubic