Ask the docs and watch retrieval, scoring, and grounding happen live.
99.98% uptime · 30d
Queries today
0
Avg latency
0ms
Recall@5
0%
Chunks
0
Retrieval map
query → nearest chunks
live
query vector
top-3 · 0.94
Query volume
queries / min
live
Librarian
RAG knowledge assistant · retrieval over a bundled doc set
Live · in-browser
Ask the docs
Try an example
Retrieved context
Ask a question and Librarian will retrieve the most relevant chunks, rank them, and answer only from what it finds.
Retrieval, scoring, citations, and refusal all run live in your browser over 12 indexed chunks — TF-IDF cosine with an exact-term boost. No server, no API key.