LIVE LEXICAL BASELINE · SYNTHETIC DATA

Enterprise RAG explorer

Ask a question about fictional company documents. Inspect the extractive answer, citation, retrieved chunks, ranking scores, access level and no-answer reason. This static online demo runs lexical retrieval only; local Python also supports real embedding-based semantic and hybrid modes.

Question

Answer and sources

Retrieved chunks and scores

All documents are synthetic. The scope switch illustrates filtering, not authentication. Answers are exact sentence extracts, not generated text. A keyword match is not proof of semantic relevance. Do not use this demo as a production knowledge or security system.