Built in the
open, for
researchers.
Researca started because reading the literature is slow, lonely, and easy to get wrong. It's built in the open by a 15-year-old in Chandigarh who got tired of tools that summarise abstracts and invent citations.
Cite everything
Every claim Researca makes links to the exact paper it came from. If it can't cite it, it doesn't say it.
Read, don't skim
Abstracts lie by omission. Researca reads the full text, because that's where the methods, caveats and real results live.
Relevance over fame
A perfect 2018 paper beats a famous 2012 one. Ranking is about your question, not citation counts.
Earn the upgrade
The free tier is genuinely useful. You only pay when Researca is already saving you hours.
The itch
Reading for a project meant 40 open tabs and no synthesis. Existing tools summarised abstracts and hallucinated citations.
First version
A working search → synthesis pipeline. A real ML researcher tested it, rated relevance 2/5, and named the fix: read full papers, real relevance, clean citations.
The rebuild
Full-text reading, a proper embedding + ranking pipeline, and citations wired to their sources. The critic agreed to test it again.
The agent
A research agent that plans, reads, and surfaces contradictions and gaps across the literature — the analysis researchers actually want.
“A tool researchers love is downstream of a genuinely good product. The foundation — full text, honest ranking, real citations — comes first. The cathedral comes after.”