Method · how it worksResearca · 2026

How Researca
actually reads.

No black box. Here is the exact path from a question you type to an answer you can trust — and why each step exists.

Full text
not abstracts
Per-claim
citations
Open access
+ your PDFs
≈ 30s
question → answer
§ 01The pipeline · six steps
01

Understand

Your plain-language question is parsed into the concepts and constraints that actually define the search — not just keywords.

02

Gather

Researca pulls candidate papers from open sources (arXiv, open-access journals) and any PDFs you've added to your private library.

03

Read in full

Each candidate's full text is chunked and embedded, so ranking and synthesis reason over methods and results — not abstracts.

04

Rank by relevance

Papers are scored against your specific question. A precise, recent, on-topic paper outranks a famous but tangential one.

05

Synthesise

The top papers are composed into one answer, written to be grounded in the source text, never overclaiming beyond it.

06

Cite & verify

Every claim carries a citation linked to its exact source. Open the paper and check it in one click.

§ 02The guardrails

Never invents citations

If a claim has no source in the read papers, it isn't made.

Never hides the source

Every statement is one click from the paper it came from.

Never overclaims

Synthesis stays inside what the text supports — caveats included.

See it run on
your question.

[Begin researching]