About the The Great Decoupling: Using AI Agents to Accele... category

The Great Decoupling: Using AI Agents to Accelerate Legacy OSGi to Client Extension Migrations

Presenter: Daniel de Francisco, Lead Sales Engineer EMEA, Liferay

The architectural shift in Liferay DXP toward a decoupled model is no longer a suggestion—it is the standard for cloud-ready, upgradeable platforms. However, for many organizations, the “migration mountain” consists of thousands of lines of Service Builder logic complex JSP/OSGi Portlets and other OSGi extensions. Manual refactoring is slow, error-prone, and expensive.In this session, we will showcase a revolutionary approach: using Autonomous AI Agents to orchestrate the end-to-end transformation of legacy modules into modern, decoupled architectures.We will demonstrate a workflow where specialized AI agents:- Deconstruct Service Builder: Analyze service.xml and Java implementation classes to automatically generate Liferay Object definitions and REST schemas.- Logic Migration: Extract business logic from *LocalServiceImpl and refactor it into Actions or external microservices connected via Client Extensions.- UI Evolution: Transform legacy JSP Portlet views into modern, framework-agnostic Custom Element Client Extensions (React/Vue/Vanilla), mapping Liferay taglibs to REST-based data fetching.The highlight of the session will be a Live Demonstration: watching an AI “Migration Agent” ingest a legacy CRUD Portlet and output a ready-to-deploy Liferay Object configuration and a decoupled frontend extension. Attendees will leave with a practical framework for using AI to turn months of technical debt into days of automated refactoring.Key Takeaways:The AI “Prompt Engineering” for Liferay: How to feed DXP-specific context (OSGi, Taglibs, Service Builder patterns) to LLMs for high-fidelity code generation.Mapping Patterns: A blueprint for translating Service Builder persistence logic into Liferay Objects and Client Extension Actions.Architectural ROI: How to build a “Migration Factory” that reduces modernization effort by 70% while maintaining code quality and security.