Skip to main content

Connected Papers

A graph from a seed paper so prior and derivative work stops hiding behind bad keywords.

Visit Website
Alternatives

Hands-on Review and Evidence

Leo Liang · Hands-on

Hands-on score: 4 / 5

Connected Papers Review: A Real Citation Map Before the Account Wall

Anonymous Connected Papers built an Attention Is All You Need citation graph in 2–3 seconds. Prior/Derivative tables, Share, and a 57.5 KB BibTeX download worked without login; Save hit the auth wall. Final rating: 4 out of 5.

Best for

  • Quick survey of a subfield around one known paper
  • Researchers who want Prior/Derivative tables before an account
  • Anonymous BibTeX export of a similarity neighborhood

Not ideal for

  • Saved graph history without signing in
  • Heavy monthly graph volume on the free 5-graph cap (not stress-tested)
  • Anyone needing paid Academic/Business features I did not run
Connected Papers

Evaluation

What to know

Key features

  • Graph from a seed
  • Prior and derivative layout
  • Visual clusters

Pros

  • Anonymous graph for Attention Is All You Need in ~2–3 seconds
  • Prior works and Derivative works tables usable without login
  • Share and ~57.5 KB BibTeX download worked anonymously

Cons

  • Dense graph labels overlap; detail pane is required for reading
  • First-use help overlay blocks the canvas until dismissed
  • Save redirects to auth; free plan lists 5 graphs/month

ToolifyNav verdict

Hands-on (Sep 2026): anonymous Connected Papers built a usable Attention Is All You Need citation graph in about 2–3 seconds; Prior/Derivative tables, Share, and BibTeX download worked without login. Score 4/5. Crowded labels, first-use overlay, and Save-behind-auth kept it from a 5. Full write-up: https://toolifynav.com/blog/connected-papers-review

About

About Connected Papers

Connected Papers is a map. You plant a seed article, and the graph shows prior and derivative work that keyword search on page one will miss.

The workflow is seed → clusters → shortlist unread nodes → deep-read → then freeze the bibliography. The graph is not citation quality. Treating every nearby node as equal, screenshotting the pretty picture for a slide, and never opening the “off topic” PDF are how people fake a literature review.

ResearchRabbit is the longer recommendation feed. Elicit is the extraction table. SciSpace is the single opaque PDF. Scite is whether later papers still clap. Connected Papers is topology.

If you will not open PDFs, you wanted a poster, not a tool.

Pick a seed you have actually read. A graph grown from a paper you only know by title will send you into a confident, wrong neighborhood. One honest seed beats five fashionable ones. Print the shortlist. A graph you only remember visually will evaporate by Thursday. One seed, read first.

Alternatives

Alternatives to Connected Papers

View all alternatives →

Discover More

Weekly digest

Get weekly AI tools + hands-on reviews

A short digest of new tools and hands-on breakdowns — unsubscribe anytime.