Search Across everything

Search your whole library at once.

One question searches every document you own — results ranked and cited per source, not file by file.

Search product illustration
1 query, all docs
40ms p95 retrieval
documents
THE PROBLEM [1/3]

Per-file search doesn't scale.

Scattered sources

Answers live across dozens of files with no single place to look.

Keyword blindness

Exact-match search misses the document that means the same thing.

No ranking

Even when you find matches, nothing tells you which one actually answers.

HOW IT WORKS [2/3]

Workspace-level retrieval.

Search flow diagram
01

Group your files

Drop related documents into a shared workspace.

02

Ask once

Query the entire set in a single natural-language question.

03

Get ranked results

Hybrid retrieval and reranking return the best passages, cited.

HOW IT'S BUILT [3/3]

Hybrid search, tuned for documents

Vector + keyword

Finds meaning and exact terms in one pass.

Cross-encoder reranking

Orders candidates by true relevance.

Per-source citations

Every result carries its document and page.

Try search on your next document

Upload a PDF free — no card required.