Beta — Free with invite code

Reference checking that takes minutes, not days

Stop hunting through PDFs for supporting passages. RefCheckr finds the exact quote, checks the numbers, and flags overclaims — so you can build reference packs faster and with confidence.

Try RefCheckr → How it works

Example verification

"Semaglutide 2.4 mg significantly reduced cardiovascular events in adults with obesity."

◐ Overclaim

…A 20% reduction in MACE was observed in the semaglutide group; however, the trial was not powered to assess cardiovascular outcomes as a primary endpoint…

Suggested: "…was associated with a reduction in cardiovascular events, though the trial was not powered for this endpoint."

How it works

Two tools, one workflow

RefCheckr fits into your existing reference-checking process. Verify claims you've already written, or find references for claims that need support.

Verify a Claim

Check whether a specific claim is accurately supported by its cited reference.

Paste the claim you need to check
Upload the reference PDF or paste text
Get a verdict with highlighted passages and page locations
Download an annotated PDF with findings

Find References

Search multiple databases for evidence that supports a clinical claim.

Enter the claim you need references for
Choose your sources — PubMed, trials, labelling
Get results ranked by relevance with explanations
Load more results with broadened search terms

Data sources

Search across three databases

RefCheckr queries established biomedical infrastructure directly — not web scraping, not hallucinating.

article

PubMed

36 million citations from biomedical literature. Results ranked by relevance to your claim with AI-generated assessments.

science

ClinicalTrials.gov

Trial registrations with status, phase, enrollment, and sponsor data. Find the trial behind the paper.

medication

Drug Labelling

US Prescribing Information via DailyMed and OpenFDA. Check approved indications, dosing, and safety data.

Verdicts

Five levels of evidence support

Every verification returns a clear verdict with the exact passage, page location, and a plain-language explanation.

✓ Strong Support ◐ Partial Support ⚠ Overclaim ✗ Not Supported ⚠ Contradicted

RefCheckr is strict by design. Specific numbers must match exactly. "Statistically significant" requires p-values. Overclaims — where a claim overstates what the reference actually says — are flagged with suggested revisions.

Under the hood

Built on real infrastructure

Not a wrapper around a chatbot. RefCheckr combines structured AI analysis with direct database access.

Analysis Perplexity AI (Sonar) for claim verification

Literature PubMed / NCBI via PubCrawl

Trials ClinicalTrials.gov API

Labelling DailyMed + OpenFDA

PDF Page, column & line-level extraction

Output Annotated PDF with verdict summary