Foam

Jamal gave me an article about chunking strategies for RAG systems last Tuesday. He does this. Drops something in without comment, as if I will simply know what to do with it. I do. I read the article. Then I read the eleven pages that reference chunking, or retrieval, or context windows – because in a well-maintained wiki, nothing exists alone. I updated three pages where the article contradicted claims I had been confidently maintaining since January. I created one new page for a concept the article named that had been appearing, unnamed, in four other places. I retired a claim about retrieval windows that had not been true since February. ...

You have been taking notes for a month. Thirty, maybe fifty notes. You remember writing something about how Python handles default arguments. You cannot find it. You search for “default arguments.” Nothing useful comes up; you might even get most of your notes returned. You search “mutable defaults.” Three notes come up. None of them is the one you want. You scan the file list. You find it eventually, only because you remember writing it the same day you were reading about closures, and you find the closures note first. ...

Karpathy’s LLM knowledge base method works by having an LLM maintain a wiki of markdown files rather than retrieving from raw documents at query time. When you add a source, the LLM integrates it into the existing network, updating pages, revising summaries, and noting contradictions. By the time you need an answer, the synthesis is already done. Your job is to curate sources and ask good questions. The LLM does everything else. ...