Ask questions. Find evidence. Understand research.
MedRAG-X retrieves the most relevant passages from documents you provide, then uses a large language model to generate answers grounded in that evidence — with every claim linked back to its source.
PDF, TXT, or Markdown you provide
Embedded chunks matched to your question
Top passages ranked by similarity
LLM answers only from that evidence
Every claim traceable to its passage
Real retrieval, not canned answers
Every demo question runs the full pipeline: document chunking, embeddings, vector search, and grounded generation. No hardcoded responses anywhere.
Honest about uncertainty
When your documents don't contain the answer, MedRAG-X says so instead of inventing facts or fabricating citations.
Open the hood
A built-in Technical Details view shows the retrieved chunks, similarity scores, and the exact context sent to the model.
Healthcare disclaimer
MedRAG-X is an educational research assistant for exploring provided documents. It does not provide medical diagnosis or medical advice. Do not upload real patient records or sensitive personal information.