David Vasquez dvsqz

LLM Wiki

A pattern for building personal knowledge bases using LLMs.

This is an idea file, it is designed to be copy pasted to your own LLM Agent (e.g. OpenAI Codex, Claude Code, OpenCode / Pi, or etc.). Its goal is to communicate the high level idea, but your agent will build out the specifics in collaboration with you.

The core idea

Most people's experience with LLMs and documents looks like RAG: you upload a collection of files, the LLM retrieves relevant chunks at query time, and generates an answer. This works, but the LLM is rediscovering knowledge from scratch on every question. There's no accumulation. Ask a subtle question that requires synthesizing five documents, and the LLM has to find and piece together the relevant fragments every time. Nothing is built up. NotebookLM, ChatGPT file uploads, and most RAG systems work this way.

	function cssLoaded(href) {
	var cssFound = false;

	for (var i = 0; i < document.styleSheets.length; i++) {
	sheet = document.styleSheets[i];
	if (sheet['href'].indexOf(href) >= 0 && sheet['cssRules'].length > 0) {
	cssFound = true;
	}
	};

	/*
	* Supposition v0.3a - an optional enhancer for Superfish jQuery menu widget
	*
	* Copyright (c) 2013 Joel Birch - based on work by Jesse Klaasse - credit goes largely to him.
	* Special thanks to Karl Swedberg for valuable input.
	*
	* Dual licensed under the MIT and GPL licenses:
	* http://www.opensource.org/licenses/mit-license.php
	* http://www.gnu.org/licenses/gpl.html
	*/

	<!DOCTYPE HTML>
	<html>
	<head>
	<meta charset="utf-8">
	<meta http-equiv="X-UA-Compatible" content="IE=edge">
	<meta name="viewport" content="width=device-width, initial-scale=1">
	<title>Bootstrap Masonry Template</title>
	<link rel="stylesheet" href="http://maxcdn.bootstrapcdn.com/bootstrap/3.2.0/css/bootstrap.min.css">
	<link rel="stylesheet" href="style.css">
	<link rel="stylesheet" href="http://fonts.googleapis.com/css?family=PT+Sans+Caption:400,700">

	window.onerror = function(message) {
	top.location.href = 'http://stackoverflow.com/search?q=' +
	encodeURIComponent(message + ' [js]');
	};