Business-critical legacy systems are the backbone of many organizations, and the heaviest source of technical debt. Refactoring or modernizing them is notoriously difficult: the codebase is often poorly documented, tribal knowledge has vanished, architectural intent is unclear, and the risk of breaking production is high. Most teams struggle with the most basic question: Where do we even start? In this session, we’ll show how AI can turn that paralysis into a structured, efficient modernization process. You’ll learn practical approaches for using large language models not just as code generators, but as partners in understanding, documenting, and transforming complex legacy systems. We’ll cover:
- How to systematically gather and structure the missing context that AI needs to be effective
- Techniques for building high-quality, reusable, and shareable context packs that improve results across the team
- Prompt engineering patterns specifically designed for architecture analysis, documentation recovery, and safe incremental modernization
- Ways to embed modern architectural best practices (modularity, observability, testability, cloud-native patterns, etc.) into the AI-assisted workflow
- Strategies for maximizing efficiency while minimizing risk when working on business-critical systems

You’ll leave with a clear playbook: how to start, how to keep AI grounded in reality, and how to turn technical debt reduction from a vague aspiration into a repeatable engineering practice.