In this video I’ll show you exactly how to build an AI-maintained personal LLM knowledge base, (often called an AI second brain) on any topic with no coding required. It’s the exact same method used by Andrej Karpathy (who popularized it) and Lex Fridman. You’ll walk away knowing how to collect sources, get Claude to compile them into a structured personal wiki, and query it to get answers no single article could give you. And I’ll show you in real time how to build one using a real hypertrophy wiki I made to help me build muscle as fast as possible.
This isn’t about pasting things into ChatGPT. This is a completely different way to use AI, where Claude reads all your sources and writes you a personal Wikipedia that grows over time, lives on your computer, and never forgets anything.
What you’ll learn:
What Andrej Karpathy’s LLM Knowledge Base actually is (explained for non-technical people)
The three-phase workflow: Collect → Compile → Query
How to set up Obsidian and Obsidian Web Clipper for free
The exact prompt I use to compile a wiki from raw sources
How to query your wiki to get answers no single article could give you
Why this beats just asking ChatGPT every time
Tools used in this video (all free except Claude Code):
Obsidian — http://obsidian.md/
Obsidian Web Clipper — browser extension
Claude — claude.ai
Claude Code — for accessing local files
Link to the prompt to create the wiki:
https://benstevenson.notion.site/Wiki-Creation-Prompt-33bbe0f4c24f8090bf84cb20c3d553c8?source=copy_link
I also created a free prompt pack with 25 of the best prompt to use with your wiki. These help you learn more, improve the wiki, find gaps in your knowledge, and hone your thinking skills about the topic. You can get it here:
https://www.sciencebased.ai/resources/wiki-prompt-cheat-sheet
Timestamps:
00:00 – Why you should care
00:46 – Personal LLM Knowledge Base Overview
03:01 – Tutorial
source
Leave a Reply