Recording a walkthrough still makes me grimace: too many takes, wrong energy when I'm on camera. Shipped it anyway — intro, walkthrough, demo. Then wired analytics to chapter selects + landing section depth. Video-assisted checkout intent should be measured, not guessed. Checkout the new landing, I'd love to hear some feedback! https://userefract.io

👋 I am building Refract, a premium SaaS foundation engineered for the long game. After scaling systems at Loom and Atlassian, I realized most MVPs are architectural traps. We’ve built a "pivot-proof" monorepo using Symmetric Legibility: an Astro marketing site for SEO, a React/MUI auth portal, and a functional core decoupled from infrastructure. It’s high-leverage code designed to map to your intent, whether read by a human or an AI. https://userefract.io