Your Documentation Is Still in a 1970s Filing Cabinet — AI Won't Fix That

Most documentation is still organised into files, folders, and hierarchies — a direct inheritance from the 1970s Xerox PARC desktop metaphor. That model made sense when office workers were the target users, but fifty years later, it's a straightjacket. Knowledge doesn't live in one folder. A component accessibility decision touches design, engineering, content, and support. Where do you file it? The tree structure forces a single location, making every other path harder.

People don't browse like librarians

Information Foraging theory (Pirolli & Card) says humans follow clues, not hierarchies. They search before browsing, stop after a few levels, ask a colleague, or create duplicates. The information exists — it's just buried three drawers deep. Sound familiar? That's your team's Confluence wiki.

AI exposes the problem

Modern AI retrieval systems don't care about folder paths. A design token page gets pulled because it mentions "colour contrast," not because it lives under Design System → Foundations → Accessibility → Colour. Folders are a storage mechanism, not a knowledge architecture. AI is revealing a workaround humans have been tolerating for years.

From storage to knowledge architecture

Tree-based structures aren't going away soon, but we can layer multiple discovery paths on top: search, metadata, tagging, cross-linking, related content, and semantic relationships. Chase McCoy's piece on design systems as knowledge graphs nails it — understanding relationships between concepts is more valuable than knowing where a file is stored. Tools like Obsidian already prove this with tag and link graphs that show how ideas connect, not where they're filed.

What helps humans helps AI

The same characteristics that make docs discoverable for people also help AI retrieve them: clear structure, meaningful headings, useful metadata, consistent language, and strong relationships between concepts. Accessibility has been preaching this for years — don't rely on a single path. The future isn't a bigger filing cabinet with better labels. It's a connected body of knowledge discoverable from multiple directions by both humans and machines.