💡

Applying Generative AI to Application Modernization

Jul 10, 2024

Applying Generative AI to Application Modernization

Introduction

  • Speakers: Leandro Santos & Aion Bam from McKenzie & Company
  • Topic: Application modernization using Generative AI (Gen)
  • Themes: Gen for Cloud, Cloud for Gen
    • Goal: Accelerating migration and modernization to the cloud
    • Relationship: Symbiotic – Gen can accelerate cloud migration and no enterprise-grade Gen AI use case exists without public cloud

State of Legacy Systems

  • Issues: Aging systems cause business and technical limitations
    • Business Side: Limits ability to change, innovate, and respond to market pressures
    • Technical Side: Brings risks (resiliency, continuity, cybersecurity), higher operating costs

Challenges in Modernization

  • Business Case Challenges: Justifying costs and benefits are tricky
    • High development costs over years
  • Technology Challenges: Lack of documentation, source code, and clear business logic
    • Resource constraints

Generative AI as a Game Changer

  • Automation: Automates labor-intensive tasks involving tedious developer work
  • Reverse Engineering: Understands business logic, tests more effectively
  • Unlocks: Automating the coding and documentation process

Methodology & Approach

  • Multi-Agent Orchestrated Approach: Codifies subject matter expert labor into workflows
    • Cognitive agents, operational agents, decision-based and critical-thinking reasoning agents
  • Benefits: Significant gains in modernization process speed and cost reduction
    • Increase modernization process speed by 40-50%, reduce costs by 30-40%

Use of Generative AI in Modernization

  • Stage-Based Approach:
    1. Taking code and generating documentation
    2. Creating human-readable test cases
    3. Generating modernized target-state systems
  • Composite Data Use: Code, observability data, business process integration, etc.
  • Rationalizing Systems: Simplifying, rationalizing systems, and improving data models

Demonstration - Mainframe Transformation

  • Example: Open-source COBOL code modernization
  • Steps:
    1. Integration in development environment
    2. Code synthesis to high-level documentation
    3. Generating new features and modern code
  • Outcome: Reduction in developer work, modernized and efficient code

Case Study - U.S. Bank

  • Legacy System: Online transaction processing through CICS, batch processes
  • New System: Modern Java components
    • Generated angular user interface, API endpoints, revised data models
  • Evaluation: AB testing for code maintainability, readability, correctness

Key Takeaways

  • Value Proposition: Reduced modernization costs and improved ROI
  • Gen-Assisted Engineering: Next wave in software development transformation
  • Methodology: Agile development, CI/CD automation, DevSecOps practices

Recommendations for Enterprises

  1. Re-evaluate modernization cases: Identify which cases now justify modernization with new methodology.
  2. Create a POC/MVP: Develop methodology, validate tool effectiveness, and stand up a transformation 'factory.'
  3. Scale: Expand to more domains and technologies, build competence and efficiency.

Conclusion: The time to start modernizing with generative AI is now!

Full transcript