After three local RAG plugins failed on my Obsidian vault, I ran the same content through NotebookLM (now Gemini Notebook) to see if the cloud version actually solves what I couldn't build…
read more →
If you’ve run a local LLM even once, llama.cpp was almost certainly involved. It doesn’t have a flashy…
read →Most AI learning sessions don't compound because the model never knows your actual starting point. Here's the four-part structure that fixes it, on ChatGPT,…
read the story →
