Jordan Green, a developer writing on Bear Blog, has published an essay tackling one of the quieter frustrations creeping into tech workflows: organizing knowledge doesn't get easier with AI—it gets weirder. The piece, titled 'Why Organizing Knowledge in the Age of AI Sucks, and What I Built,' landed on Hacker News on July 14, 2026, drawing attention from developers wrestling with how to structure notes, docs, and context for LLM-assisted workflows.

The Core Problem

Green argues that traditional knowledge management tools—think Notion, Obsidian, or plain old folder hierarchies—were designed for humans to retrieve information. When AI enters the picture, those retrieval patterns shift dramatically. You're no longer just searching for something you wrote; you're trying to give an LLM the right context window at the right time. That changes what 'organized' even means. The essay digs into how existing systems optimized for human recall become liabilities when you're prompting a model that needs precise, searchable grounding.

Why Existing Tools Fall Short

The piece doesn't just complain about current tools—it identifies specific failure modes. Green points to the mismatch between how humans tag and structure notes versus what makes sense for semantic search or RAG pipelines. Tags get stale, folder hierarchies impose rigid taxonomies that break under cross-domain queries, and most note-taking apps lack native export options that play nice with code-based retrieval systems. It's a developer-centric critique: the tools were built for writing, not for feeding context to models.

What Green Built

Without spoiling the full essay, Green walks through their own solution—a custom-built system tailored to how they actually work with AI assistants. The approach apparently prioritizes programmatic access over GUI-based organization, leaning into plain-text formats, structured metadata, and tooling that integrates directly into a dev environment rather than sitting in a separate app silo.

Key Takeaways

  • Human-oriented knowledge structures often hinder AI-oriented retrieval patterns
  • Plain-text and code-adjacent formats may outperform traditional note-taking apps for LLM workflows
  • Custom solutions are emerging as off-the-shelf tools lag behind developer needs
  • The definition of 'organized' is shifting from human-findable to machine-parseable

The Bottom Line

Green's essay is a signal that personal knowledge management isn't just a productivity sidebar anymore—it's becoming a core part of how developers build with AI. If you haven't thought about your note-taking system through an LLM lens yet, you're probably already behind.