Scrutinise the evidence behind any paper.
Papers stop at publication. The evidence doesn't. See how it has moved on a paper's main claims since it was published.
Engine settings
- 01Resolve the DOILook the paper up in Europe PMC, with CrossRef as the fallback.
- 02Read the abstract and extract the main claimsThe claim extraction step lists the empirical claims the abstract makes and marks the main claims, the ones the paper's conclusion rests on. The primary-outcome claim is reviewed automatically; click any other claim to review it instead.
- 03Retrieve the literature that bears on the claimThe claim's intervention, comparator, population and outcome become PubMed queries. Four lanes run at once: papers citing this paper, keyword search since publication, keyword search before publication, and the paper's own references.
- 04Screen each candidate for bearing on the claimEach abstract is read against the claim. A paper bears when it reports its own evidence on the same intervention and outcome. The deciding sentence is quoted verbatim and checked against the abstract.
- 05Four-role panel reviewMethodologist, synthesist, sceptic and moderator weigh the bearing papers and grade the claim from 0 to 100 on the fixed scale. This runs when you request the review and is delivered by email.
Your evidence review is ready.
Enter your email and we will send it to you.
Here is an example of what you will get: the full claim review for "GLP-1 receptor agonists reduce all-cause mortality in type 2 diabetes".
What makes this different
AI can summarise a paper, but a summary cannot tell you how the evidence has moved since it was published.
We have built a systematic evidence protocol that finds the papers bearing on a claim, weighs each one, argues the case against, and grades the result on a fixed scale. It lets researchers see where a paper's claims stand today, what they rest on, and what research would shift them.
How it works
- 01Share a DOIScrutinise reads the abstract and lists the paper's main claims.
- 02Choose a claimPick the one you want checked, or write it in your own words.
- 03Get the claim review by emailEvery paper on the claim, from before publication and since, weighed by a four-role panel.
What you get
A claim review showing where the evidence stands today.
Example: Perkovic et al., N Engl J Med 2024 · GLP-1 receptor agonists reduce all-cause mortality in type 2 diabetes
- 01
Where the claim stands today
One grade from 0 to 100 on a fixed scale, from Refuted to Established.
85Established - 02
The papers it rests on
Every paper on the claim, before and since publication, with what it does to the claim and the sentence that decides it.
SupportsMarso et al. 2016 · RCT, n = 9,340Does not bearLincoff et al. 2023 · people without diabetes - 03
The case against
One reviewer's only job is to argue against the verdict. Its strongest objections sit beside the grade.
"Mortality was exploratory, not multiplicity-protected." - 04
Research gaps that could shift it
The gap that would most move the claim, the study that would close it, and the questions still open.
The study that would close it: a placebo-controlled RCT in type 2 diabetes powered on all-cause mortality as its primary, multiplicity-protected endpoint.
Every claim review is open to correction.
Add a paper that should be there, or flag one that was misread. The claim is re-graded with the change, and accepted corrections go into Priors, the open registry of claims in health and medicine.
See where any paper stands.
Share a DOI, choose a claim, and the claim review is emailed to you.