Julia Kordick

Migrating from COBOL with AI: A Moonshot Demo

A bank had 17 million lines of COBOL. We built a team of autonomous AI agents to migrate it to Java.

Migrating from COBOL with AI: A Moonshot Demo
#1about 2 minutes

Introducing an open source AI project for COBOL migration

A multi-agent AI framework developed with Bankdata modernizes their 17 million lines of COBOL to reduce costs and dependencies.

#2about 1 minute

How to set up and run the migration demo

The migration process is initiated by configuring an AI endpoint, adding COBOL source files, and running a single Docker command.

#3about 2 minutes

Understanding the multi-agent analysis and dependency mapping

AI agents perform reverse engineering on COBOL files and map complex, multi-level call chains and dependencies using GOTO commands.

#4about 1 minute

Orchestrating the migration with semantic kernel and logging

The process uses Semantic Kernel for robust agent orchestration and generates logs and markdown files to ensure resilience and traceability.

#5about 1 minute

Building a flexible source and target agnostic framework

The framework is designed to be adaptable beyond COBOL to Java by changing agent personas through prompting for different source or target languages.

#6about 1 minute

Reviewing the generated Java code and migration artifacts

The final output includes markdown reports, mermaid diagrams for visualization, and fully generated, modern Java code from the original COBOL.

Related jobs
Jobs that call for the skills explored in this talk.

Featured Partners

Related Articles

View all articles

From learning to earning

Jobs that call for the skills explored in this talk.