Google teased a DESIGN.md standard for AI design systems… but it looks like it’s locked to Stitch instead of being open

Why Read This from an HCI Perspective

This article shows why ‘standardization’ matters in AI-driven design workflows, and why the standard isn’t just a file format issue but an interoperability problem. From the perspective of HCI/UX practitioners and researchers, it lets you quickly review key concerns such as transferring context across tools, preserving design intent, and making agent interpretation possible. In particular, it prompts you to think about how design systems in the generative AI era should be mediated between documents, rules, and code.

CIT's Commentary

From a CIT perspective, it’s more appropriate to interpret DESIGN.md not as a ‘new format’ but as an experiment in how AI reads design systems. The issue isn’t markdown’s simplicity itself; rather, it’s whether that simple document can reliably express and convey the system’s structural constraints, component relationships, and even exception rules. If it stays confined within Stitch, then it’s less a standard and more a product feature. On the other hand, if tools like Figma, Cursor, and Storybook are designed to share the same underlying meaning model, then it could evolve into an operating system for design in the AI era. Still, the more practical question isn’t whether everything can be expressed in a single document, but rather which layers of information should be common and which should be separated by tool.

Questions to Consider While Reading

• Q.To become a true standard, what level of meaning model and schema would DESIGN.md need?
• Q.When expressing a design system’s rules in markdown, how should you balance human readability with AI interpretability?
• Q.What design principles are needed to increase interoperability across tools while preserving each product’s unique workflow and constraints?