Build a Hermes Knowledge Base That Self-Improves
📈 ALL Systems: https://bit.ly/4kol0y5 🩵 Free Resources: https://bit.ly/3RNNDLa 🎤 Glaido: https://bit.ly/4eGoI3R Hermes has the best memory of any AI agent, but there's one thing it just can't do. Once you fix it, you give Hermes a self-improving knowledge base built on Andrej Karpathy's LLM wiki principles. This is the workflow for anyone who wants Hermes to know your inbox, meeting notes, calls and expert research, not just you. You and I wire up an Obsidian wiki, connect it to Hermes and Claude, and set automatic cron jobs to ingest everything from Granola and NotebookLM. It saves you hours and gets you genuinely further ahead. *** CORE Software *** 📝 NotebookLM: https://notebooklm.google.com/ ☁️ Claude: https://claude.ai/ 🧠 Karparthy Wiki: gist.github.com/karpathy/442a6bf555914893e9891c11519de94f 🚀 AntiGravity: https://antigravity.google/ 🤖 OpenAI: https://openai.com/ 📓 Obsidian: https://obsidian.md/ 🎙️ Granola: https://www.granola.ai/ ⌚️ Stamps: 00:00 - Hermes Memory Superpowers 00:43 - Why LLMs Feel Like Amnesia 01:18 - How Hermes Memory Works 01:36 - The One Blind Spot 02:22 - A Knowledge Base That Grows 02:40 - Karpathy's LLM Wiki Idea 03:37 - Setting Up Your Wiki 04:02 - Building The Folder Structure 05:41 - The Injection Workflow 06:51 - Connecting The Vault To Hermes 07:54 - Building An Obsidian Skill 09:34 - Indexing Any Article You Find 11:53 - The Two-Way Memory System 12:35 - Auto Ingesting Meetings With Granola 13:45 - Scaling It With Notebooks 14:15 - The Next Piece Of The Puzzle Build a self-improving Hermes knowledge base with an Obsidian LLM wiki, connect it to Claude and NotebookLM, and set up automatic cron jobs with Granola so your AI agent never forgets your meetings, emails or expert research. #Hermes #ObsidianRag #NotebookLM #ClaudeCode #AIAgents #LLMWiki
