If you can read a normal note, you can read Markdown. It stays readable even before software turns it into a pretty page, and that’s the whole point: both people and AI can read it without a heavy app in the middle.
Think about a paper note where you use simple marks to help the reader: a title at the top, an underlined section, a short list, a circled word. Markdown is the digital version, but the marks are typed characters. A heading starts with a #, a list starts with a dash, a link puts brackets around the words and parentheses around the address. It looks technical because of the little symbols, but it’s closer to a clean note than to software.
How it shows up
This matters when you work with AI tools. A PDF can look nice to a human but add layout noise (headers, columns, footers, weird copied text) before the real text is clear. Modern models handle many text PDFs, so PDF isn’t forbidden, but Markdown is cleaner when the goal is for an agent to read, edit, and reuse the material. That’s why so much agent work uses it. A README is often Markdown, a skill can be a Markdown file, and a vault is often a folder of Markdown notes. YAML frontmatter at the top adds structured fields like date or status, while the body stays readable.
Why you care
For a non-technical person, the unlock is simple: Markdown lets you own the instruction, not just the app. Knowledge locked inside a proprietary tool needs that tool to show it back to you. Knowledge in Markdown files opens almost anywhere: Claude Code, Codex, Obsidian, GitHub. Not everything should be Markdown, since a spreadsheet should stay a spreadsheet and a contract should stay a Word document. When the goal is durable instructions, notes, process docs, or agent skills, it’s usually the cleanest format. When you see .md, don’t flinch. It just means, “This is a text note with a little structure.”