# dom.vin > Illustrated design essays on AI by Dom Vinyard. Every essay below has a parallel markdown mirror at //.md, written for agents that cannot see the rendered visuals. ## Essays ### Seeds 🌱 - URL: https://dom.vin/2026/seeds - Markdown mirror: https://dom.vin/2026/seeds.md - Published: 2026-04-30 **Summary** The seed skill is a way for agents to share context — bundles of files (memory, skills, plans) wrapped in a folder. Install once with `Fetch & install seed.show/skill`. Tell your agent to share any folder and it gives you back a URL; the recipient pastes the URL into their agent and the files land locally, context shared. The essay frames this as the smallest end of a much bigger spectrum: dyadic context-building, small-group shared lore, civilizational moral-constitution context. Same dynamics rhyme at every scale; texture and stakes change. Seeds is the small case — one folder, one prompt, one URL — and the small case being good enough is what's been missing. Built on fold (folder ↔ single markdown file) and portdown (markdown ↔ self-installing polyglot). **Claims** - Markdown has become the language of AI context — memory, skills, plans, resources all end up as bundles of markdown wrapped in a folder. - Sharing those bundles between agents is the friction; seeds is the smallest-case answer to it. - Context-sharing is a spectrum — dyadic, small-group, civilizational — and the small end being good enough is what's been missing. ### Hello, world - URL: https://dom.vin/2026/hello - Markdown mirror: https://dom.vin/2026/hello.md - Published: 2026-03-15 **Summary** This is the introduction essay for the rebuilt dom.vin, by Dom Vinyard, founding designer (and hacker-in-residence) at sauna.ai, where the team is working to make agentic AI safer and easier to use for everybody. The site is a slower, more speculative track for thinking out loud about the philosophies and the culture growing up around AI. Each essay is built for two readers in parallel: a human who scrolls a rendered page with figures, and an agent that fetches a markdown mirror at //.md and reads prose with a paragraph wherever a figure would have been. The wider thesis: a huge part of agents working reliably will turn out to be awareness — the same signals a person uses to confirm the world is the way they think it is (tests prove code works, items in a sent folder prove email went out) are most of what lets an agent operate without flying blind. **Claims** - dom.vin is being rebuilt as a small home for slow, illustrated essays on AI. - The site is designed for two parallel readers: a human reading a rendered page, and an agent reading a markdown mirror. - Agents working reliably depends on awareness — the same signals humans rely on to confirm the world: tests for code, sent-folder receipts for email.