distils.

Open source Apache 2.0

Scientific papers, distilled
into reusable knowledge.

Turn research PDFs into standalone facts, equations, and visual findings for RAG, GraphRAG, and scientific knowledge graphs.

Read the docs View on GitHub Explore the examples

From paper to records.

How it works

Read each page in context

PDF pagesread in order
Read one page
Earlier facts + glossarycarried into the next page

Extract standalone knowledge

Structured JSON

Facts
Explicit subjects & essential conditions
Equations
The mathematics and its meaning
Visual findings
What figures and tables show
Page references
GROBID citation markers & internal references
Extract standalone knowledge to search across papers and connect them through shared entities.

Build on the knowledge.

Explore use cases

RAG & semantic search

Index standalone scientific statements for retrieval-augmented generation (RAG), semantic search, and question answering across papers.

Knowledge graphs & GraphRAG

Use extracted entities to connect papers through shared methods, datasets, and concepts. Add relationships and graph retrieval to build a GraphRAG application.

Citation graphs & literature discovery

Explore each page’s scientific records alongside GROBID citation and internal-reference markers. Resolve their targets to connect papers, pages, and referenced figures, tables, or equations.

Scientific hypergraphs

Explore relations that keep a finding’s methods, datasets, metrics, and conditions together. Requires additional entity linking and relation extraction.