Multi-Document RAG: A Folder of Unrelated PDFs Is One Long Document with a Nested Outline

When handling a collection of unrelated documents for RAG, the most effective strategy is to treat the entire folder as a single large document with a nested outline. This approach begins by creating a top-level index containing a single, carefully crafted summary line for each file, written specifically for an AI router to select the correct document. Once a file is chosen, retrieval then drills down into that document's own table of contents to pinpoint the exact section relevant to a user's query. This two-level routing system efficiently narrows down information without needing a complex, field-based index, which is impossible when documents share no common attributes.