Seed it with one paper you already trust
Search for a paper by its exact title, add it to your library, and run a search from it. You get a list of recommendations down one side and a map beside it, where the horizontal axis is time, oldest to newest, and the vertical axis is how heavily a paper has been cited. Your own papers show as filled dots; recommendations show as hollow ones.
Why this one: Connected Papers builds one graph per origin paper and hands it to you finished. ResearchRabbit instead keeps your seeds in a library you keep adding to, and re-runs the recommendations against the whole growing set — so the first seed you pick shapes every round after it, in a way it does not in a tool where you simply generate another graph.
Tweak: Add two or three seeds rather than one. The recommendations tighten noticeably once the tool can see what your papers have in common.
Walk backwards, then forwards, from the same paper
With a paper selected, switch between Earlier Work and Later Work. Earlier Work is that paper's own reference list — what it was built on. Later Work is everything published since that cites it. One paper, two opposite directions in time.
Why this one: The ranking behind these two views is arithmetic, not taste: the vendor's help centre says recommendations are calculated from the most influential articles in the local citation network, "where influence comes from both the number and quality of citations." That makes a well-cited older paper's Later Work the fastest honest way to find out whether its conclusion actually survived — something a similarity-based map like Connected Papers, which does not separate direction at all, cannot tell you.
Tweak: An empty Later Work usually means the paper is too recent to have been cited yet, not that it was ignored.
Use Similar Work when the citation trail goes cold
Switch to Similar Work. Unlike Earlier and Later Work, this view does not appear to follow citations at all — a reviewer who examined the rebuilt tool describes it as matching papers on their title and abstract, though the vendor does not spell out the method. Either way it can reach work in a neighbouring field that never cited yours and never will.
Why this one: Knowing which of the three views follows citations and which does not is the difference between using ResearchRabbit and clicking around it. Litmaps and Connected Papers both fold similarity into a single blended score; ResearchRabbit keeps the routes on separate buttons and makes you pick one — more work, and more control over what you are actually asking for.
Tweak: If Earlier and Later Work keep returning the same clique of papers, that is the signal to switch to Similar Work.
Pull your Zotero library in
Under Profile, then Account Settings, then Preferences, link your Zotero account — it uses a one-time permission screen rather than a key you copy. Then use the Zotero button on the Library page to choose a Zotero collection, pick which articles come across, and say which ResearchRabbit collection they land in. It is a manual import you re-run when you want it, not a live sync, and it runs one way. To get papers back out to Zotero you export a citation file from ResearchRabbit and import it on the Zotero side.
Why this one: The vendor says a two-way sync is in development and has not given a shipping date; until it ships, whichever side you last edited is the side that is correct, and nothing tells you which that was.
Tweak: Import one narrow Zotero collection rather than your whole library — the help centre warns that large Zotero libraries take a while to load.
Save a collection and share the link
Group what you have found into a collection, then share it, either with named collaborators or as a public link. Collections and subcollections, notes, stickers and colour labels are all on the free plan, and the pricing page puts no cap on how many you make.
Why this one: Someone opening your shared ResearchRabbit collection sees not just the list but the route — which papers you seeded and what came back. That is a different artefact from emailing a bibliography, and unlike the 50-seed search ceiling, it is the part of the product the free plan does not meter at all.