AI-Augmented Design: How I Use LLMs to Ship Better Tokens, Palettes, and Specs in Liferay DXP
Presenter: Diego Alonso, Assoc Product Designer, Liferay
Every AI talk at DEVCON so far has been about backend stuff, we’ve been talking about RAG pipelines, MCP servers, agentic orchestration. This keynote comes to change this. I’m a Product Designer, and I want to show what happens when you put AI at the center of your design system workflow. I’ll demo real workflows from my daily work on Lexicon and Liferay DXP, that help me take better decisions and be better informed.- WCAG analysis on autopilot: I feed color palette JSON exports to an LLM and get back contrast-ratio matrices, Delta E perceptual distance reports, and compliance verdicts across four color modes. This used to take me days of manual spreadsheet work.- Token validation at scale: Cross-checking 400+ semantic tokens against their resolved values in Light, Dark, HC Light, and HC Dark, catching mismatches and getting fix proposals before any code is written.- Design-to-Jira pipeline: Turning design decisions into structured Jira tickets with acceptance criteria, expected vs. actual behavior, and implementation hints.- Decision docs that actually get used: Generating structured decision summaries that work as both design rationale and engineering specs at the same time.I’m not going to tell you which AI tool is the best or give you prompting tips. I want to show concrete, demo-driven workflows where AI collapses the gap between design intent and engineering implementation. Everything I’ll demonstrate ships real output that’s already in Liferay DXP: token files, palette definitions, accessibility reports, and component specifications.Whether you’re a designer, a frontend engineer, or a PO, I think you’ll find patterns here you can bring to your own Liferay workflow. And I’ll be honest about where AI helps, where it hallucinates, and where you still need a human making the call.