Descriptions:
Greg Isenberg hosts designer Mangto for a deep dive into Google’s design.md — an open-source Markdown file format that encodes a complete design system (typography, color palette, spacing rules, WebGL animations) so it can be passed directly to AI coding agents as a prompt attachment. The core idea mirrors how developers use agents.md or soul.md to give AI context: designers can now inject a consistent visual identity into any AI-generated interface, taking output from generic “vibe-coded” results to something that holds up professionally across landing pages, mobile screens, and motion design.
Mangto walks through his full workflow live: downloading both a design.md and its companion HTML template from the design.md library at Aura, combining them in a prompt, and generating a coherent landing page. He explains why pairing the two files matters — the HTML captures WebGL and three.js animation logic that the design.md alone may not fully encode — and how the same design.md file can be ported across different AI platforms without being locked to any single tool.
The conversation also touches on ChatGPT image generation for slide assets, the economics of running an AI-native startup (Mangto estimates spending close to half a million dollars in tokens), and a brief comparison with Google Stitch as an alternative. For founders and builders who want polished, consistent design without a dedicated design team, this episode offers a concrete, replicable workflow using currently available tools.
📺 Source: Greg Isenberg · Published May 06, 2026
🏷️ Format: Podcast
