LLM Wiki vs RAG

Use RAG when the answer is the product

RAG is a good fit when users need fast question answering over a changing corpus and do not need to own the intermediate knowledge structure.

• Customer support over a document set
• Internal search across policies or code references
• Chat interfaces where the response is the primary artifact

Use an LLM Wiki when the knowledge asset matters

An LLM Wiki is better when you want Markdown pages, backlinks, indexes, open questions, and a source-backed archive that remains useful outside a chat box.

• Research that must survive across projects
• Learning notes you want to revisit and edit
• Market research that needs evidence, entities, and synthesis pages

They can work together

The choice is not religious. A wiki can be the maintained knowledge layer, and a RAG system can retrieve from it later. The wiki keeps the corpus legible before retrieval begins.

The practical test

If the output should become a file you can edit, link, publish, or review, use the LLM Wiki pattern. If the output only needs to answer the current question, RAG may be enough.