Sort a whole reference list by what actually holds up.
Free to start. You judge the sources. CoChat sorts them so you know where to look.
What is citation triage?
Citation triage is the fast sort of a reference list by verification signal: which sources resolve to a real paper and support their claim, which are disputed, and which do not exist. CoChat runs the check against CrossRef, reads the full source, and surfaces the result, so you know where to focus before you read.
When a reference list runs to dozens of sources, checking each one by hand is the slowest, least rewarding part of the work. Skip it and a fabricated or misattributed reference slips through. Do it manually and you burn the time you needed for the sources that actually matter.
Triage removes the tradeoff. CoChat does the mechanical sort in one pass and marks the references that deserve a closer look. You still read the sources that count and make every call. You just start already knowing which handful of the list needs scrutiny.
Sort first. Scrutinize what matters.
A triaged list is the start of the work, not the end.
The sources that hold up are ready to use. They flow into the review you are writing and the citations you build, and you can keep them in your Library, organized and already checked.
What was a wall of unsorted references becomes a shortlist you can trust and build on, so the next reference list you face starts from confidence, not doubt.
Hand-checking vs. triage
Checking by hand
CoChat citation triage
First pass
Read every reference, check each DOI yourself
Every source checked and sorted in one pass
Finding the weak sources
Buried until you happen upon them
Surfaced to the top: disputed and non-existent flagged
Where your time goes
Split between DOI-chasing and the work
Concentrated on the sources that matter and the flagged few
Fabricated references
May slip through under deadline
Flagged before you rely on them
A list too long to check
Gets skimmed, or skipped
Sorted in one pass, whatever its size
Who judges
| You |
Still you. CoChat sorts; you decide
CoChat sorts on a real signal, not a guess.
Fabricated and misattributed references are no longer rare. A large-scale audit of 111 million references estimated 146,932 hallucinated, non-existent citations in scientific papers in 2025 alone. Hallucination is not confined to weak models either: rates run from 14.23% to 94.93% across 13 state-of-the-art models in a recent cross-model audit.
At that scale, hand-checking does not hold. Nobody can verify every DOI in every long reference list and still have attention left for the sources that carry the argument. Triage runs the same CrossRef verification and full-source reading that power the rest of CoChat, so the signal you sort on is real, not a guess.
146,932
estimated hallucinated citations in scientific papers in 2025
14–95%
citation hallucination rate across 13 leading models
2 checks
existence against CrossRef, claim against the full source
You decide
CoChat sorts and flags; you judge the sources
Same capability, different context
Same capability, different context. Click to see how citation triage works across fields and roles.
See the complete citation workflow
Management is the lifecycle. These are the focused capabilities inside it, each answering a narrower question.
Management is the lifecycle. Verification and triage are the deep dives inside it. You get all three in one workspace.
Frequently asked questions
1. What is citation triage?
2. Does CoChat decide which sources are good?
3. How does CoChat know a reference is fabricated?
4. How big a reference list can it handle?
5. Do I still read the sources myself?
Give your time to the sources, not the DOI-checking.
CoChat triages any reference list, checking each source against CrossRef and its full text, so you see the weak sources at a glance and spend your attention where it counts.
Free to start. You judge the sources. CoChat sorts them.

