rag·assistant
PORTFOLIO DEMONSTRATION

Ask the system how it works.

This LLM-powered assistant retrieves relevant passages from a collection of documents before answering, cites the material it used, and abstains when the collection does not support an answer.

The collection

The public demo reads a fixed set of documents about its own architecture, retrieval strategy, evaluation harness, citations, and deployment. Public uploads are disabled.

Questions to try
  • How does hybrid retrieval work in this project?
  • Why does the assistant sometimes refuse to answer?
  • How are citations kept valid after documents change?