Claude just killed ALL Note-Taking Apps. Here is proof.
For seven years, the note-taking world has been a race through apps: Evernote, Obsidian, Notion, Mem, Tana, Heptabase. Each promised control, flexibility, and the perfect scaffold for personal knowledge management. But what if all those tools — and the entire PKM industry — just became obsolete? In this video, the host doesn't just theorize about AI replacing note-taking apps. He builds a complete personal knowledge assistance system from scratch, live, using nothing but a folder on his desktop and Claude. The promise is audacious: no subscriptions, no vendor lock-in, no pre-built templates. Just plain text, a local database, and AI agents that work for you. But can a single folder and a chatbot really replace years of PKM infrastructure?
Puntos clave
PKM tools lock you into developer-designed structures; local folders plus AI let you build exactly what you need, swap AI models freely, and retain total ownership of your data.
You don't need to write code. The entire system — agents, database, interface — is built through plain-language conversation with AI, even in the terminal.
AI empowerment means giving each person their own AI team, not layering one AI over a whole organization. Individual superpowers drive collective efficiency.
A local SQLite database plus a simple HTML interface replaces Notion dashboards, Obsidian plugins, and premium PKM subscriptions — at a fraction of the cost.
The era of Personal Knowledge Management is over; the era of Personal Knowledge Assistance — where AI actively organizes, indexes, and surfaces your knowledge — has begun.
En resumen
Personal knowledge management tools as we know them are over. With Claude (or any AI), a single local folder, a SQLite database, and custom agents, you can build a fully personalized, interface-agnostic knowledge system that you control — no code, no subscriptions, no vendor lock-in.
Why PKM Tools No Longer Matter
All note-taking apps lock you into someone else's scaffold.
For years, productivity enthusiasts have hopped between Evernote, Obsidian, Notion, Mem, Tana, and Heptabase, each time adapting to a developer's vision of how knowledge should be structured. The host reveals the core problem: these tools are scaffolds built by others. Even Obsidian, celebrated for local control, formats files in ways that only Obsidian can properly visualize. You're always dependent on plugins, templates, and the roadmap of developers you've never met.
The breaking point came when Heptabase refused to open an MCP server for easy data access. That was the «final nail in the coffin»: if you can't freely access and manipulate your own knowledge, you don't truly own it. The solution? A single folder on your computer, a local SQLite database, and Claude as your orchestrator. The AI is swappable — Claude today, Gemini tomorrow, a local model next year. What you build is yours, tool-agnostic, and infinitely customizable.
This isn't about replacing apps with AI. It's about replacing rigid structures with fluid, personalized assistance. The host calls it the shift from Personal Knowledge Management to Personal Knowledge Assistance. PKM was about organizing information; PKA is about having AI actively surface, index, and work with your knowledge on demand.
The Philosophy: Empower Individuals, Not Organizations
Real AI advantage comes from giving each person their own team.
The Philosophy: Empower Individuals, Not Organizations
The host and his co-founder work with businesses worldwide on AI implementation, and they've identified a key misconception: companies think AI should manage the business from the top down. The real power is the opposite. «Any business owner and team leader who empowers their employees to learn these skills will make the overall business more efficient.» Give each person their own AI team — custom agents tailored to their field of expertise — and you create compounding individual advantages that lift the entire organization.
Building the System: From Empty Folder to AI Team
Three folders, one orchestrator, and plain-language prompts create everything.
Create a folder Right-click desktop, create a new folder (e.g., «PKA demo»). This folder will hold your entire system — no cloud, no app, just local files.
Open terminal and navigate Open Terminal, copy the folder path (Option+right-click on Mac), type «cd» and paste. You're now in the folder. Type «claude» to launch Claude Code.
Create inbox folders Manually create «owner's inbox» (for AI deliverables) and «team inbox» (for files you want AI to process). This gives clear input/output boundaries.
Introduce Larry, the orchestrator Tell Claude: «You are my AI assistant, Larry. You are only an orchestrator. You never do work yourself — you hire specialists.» Larry will create a team roster and agent definitions.
Hire specialist agents Larry hires Pax (senior researcher) and Nolan (HR director) by default. When you need something — e.g., organize scanned PDFs — Larry asks Nolan to hire a specialist (e.g., a librarian agent).
Add a local database Ask for a SQLite database to store knowledge, journal entries, contacts, projects, and file metadata. AI creates schema, tables, and indexes automatically.
What You Just Built (And What It Costs)
Live Demo: Organizing 120 Scanned PDFs in Seconds
Drag, drop, and watch AI OCR, categorize, and index.
To prove the system works, the host drags 120 scanned PDFs (invoices, receipts, documents — all in German and English) into the «team inbox» folder. He tells Larry: «Organize these, create folders that make sense, add them to the database, and find interconnections.» Larry immediately recognizes the files need OCR, delegates to a temporary «librarian» agent (or hires one via Nolan), and begins processing.
Within minutes, PDFs are renamed, sorted into year-based folders, and indexed in the SQLite database. The host can now query: «Show me all invoices from 2023» or «Find documents related to supplier X.» No tagging, no manual sorting, no Hazel rules. AI does it all, learning from the context of the folder and the database schema. The host notes: if something goes wrong, you simply interrupt (hit Escape), ask AI to undo, and refine. No complicated rollback procedures, no lost work — just conversation.
This is where the terminal shines over Claude Desktop's «co-work» mode. You see every step, every permission request, every agent action. The host runs multiple terminals — six to eight Larry instances — each managing a different domain (business, personal life, hobbies). This parallelism is the key to scaling AI assistance without chaos.
Building the Interface: From Overkill to Simple
First attempt built a full app; second built a single HTML file.
Why This Beats Obsidian, Notion, and Heptabase
The New Era: Personal Knowledge Assistance
AI doesn't just store knowledge; it actively assists you.
“We just entered a new era. The era of personal knowledge management is over and we are entering the new era of personal knowledge assistance.”
Key Numbers: What This System Delivered
Tokens, files, agents, and cost in one live session.
Personas
Glosario
Aviso legal: Este es un resumen generado por IA de un vídeo de YouTube con fines educativos y de referencia. No constituye asesoramiento de inversión, financiero o legal. Verifique siempre la información con las fuentes originales antes de tomar decisiones. TubeReads no está afiliado con el creador de contenido.