Use caseFor scientific & medical comms
Catch the wrong figure, finding, or reference before peer, client, or MLR review.
Your writers and editors turn primary papers, datasets, and reference packs into research manuscripts, biotech and R&D whitepapers, clinical summaries, and pre-MLR drafts on tight deadlines. Lenz identifies the factual statements it can check against Crossref, OpenAlex, Semantic Scholar, PubMed, and official regulatory records, and checks that cited DOIs and study links exist and support what’s written.
- Private by defaultstored in the EU
- Zero data retentionon Pro and Scale plans
01What you get back
Three statements in a study summary. One dose result is swapped.
Here is a paragraph summarizing Phase 3 SURMOUNT-1 results, the way Lenz checks it.
Your text
“The Phase 3 SURMOUNT-1 trial enrolled 2,539 adults without diabetes who had a body-mass index of at least 30, or at least 27 with at least one weight-related complication. Participants received once-weekly subcutaneous tirzepatide at doses of 5 mg, 10 mg, or 15 mg, or placebo, for 72 weeks. At week 72, the mean percentage change in body weight was −20.9% with once-weekly 5 mg tirzepatide.”
What Lenz found
- TrueSURMOUNT-1 enrolled 2,539 adults under the BMI criteria10/10 · high confidence
- True5 mg, 10 mg, or 15 mg tirzepatide vs placebo for 72 weeks10/10 · high confidence
- FalseMean weight change at week 72 was −20.9% with 5 mg1/10 · high confidence
The correction
−20.9%−15.0%
At week 72 of SURMOUNT-1, once-weekly 5 mg tirzepatide produced a −15.0% mean weight change; −20.9% was the result for the 15 mg arm.
Suggested rewriteAt week 72 of the Phase 3 SURMOUNT-1 trial, the mean percentage change in body weight was -15.0% with once-weekly 5 mg tirzepatide.
The sources
nejm.orgaccessdata.fda.govdoi.org+19 more
- Your text
Lenz reads your manuscript, whitepaper, or slide draft and picks out every factual statement it can check. Here it found three.
- A verdict for each fact
A verdict from True to False (or No Public Record), with a score from 1 to 10 for how strongly the sources back it. Two held up. One didn’t.
- A suggested correction
When the primary paper or registry establishes the right figure, Lenz suggests the correction and explains why. Your writer or editor decides whether to change it.
- The sources behind it
Every result lists the papers, registries, and documents it was checked against, so co-authors, scientific directors, and MLR reviewers can see for themselves.
A real deep check, run on 19 November 2026. Each statement opens the verification behind it.
Open the full verificationCitations
A study DOI that resolves can still contradict the sentence.
When your draft cites a DOI, a PubMed record, or a technical paper, Lenz resolves it across Crossref, OpenAlex, Semantic Scholar, and PubMed, reads the paper or abstract, and checks whether it supports the sentence.
The DOI is real and points to the right landmark paper, so a link checker passes it. Lenz reads the paper and flags that the sentence attributed the composite endpoint reduction to a single component.
The reference uses a valid Nature Medicine prefix and format, so nothing looks wrong in a bibliography list. Lenz checks the registries and flags that the paper does not exist.
02Use cases
Where scientific and medical teams put Lenz to work.
Verify biomedical and scientific citations and check for fake, made-up, or hallucinated DOIs, PubMed references, and study figures in manuscripts, R&D whitepapers, medical communications, and pre-MLR drafts before publishing or review.
-
Manuscripts, literature reviews & congress posters
Check the numbers and bibliography before author or peer review. Sample sizes, confidence intervals, p-values, and cited DOIs checked across Crossref, OpenAlex, Semantic Scholar, and PubMed before a manuscript, systematic review, or poster goes to co-authors, a steering committee, or a journal.
-
Biotech, life-sciences & R&D whitepapers
Check technical deliverables before client or editorial sign-off. Assay and benchmark comparisons, environmental and materials studies, and technical monographs checked against primary literature, patents, and official datasets before delivery.
-
Medical affairs, CME & pre-MLR materials
Check clinical claims and reference packs before MLR submission. Study populations, dosing regimens, trial endpoints, treatment-guideline summaries, and cited papers checked against PubMed and regulatory records before your editors annotate the submission pack.
03Where you work
Check the draft in the tools your writers and editors already use.
Choose where you work 3 ways to run a check
Review the manuscript in Word.
Upload a .docx, a PDF, or an image in the Workbench on desktop. Download the file with a Word comment from Lenz on each statement and citation that needs attention, and suggested edits written in as tracked changes you can accept or reject.
A suggested edit is not itself verified: read it alongside the primary paper before accepting it.
Week-72 weight reduction
At week 72 of the Phase 3 SURMOUNT-1 trial, the mean percentage change in body weight was −20.9% -15.0% with once-weekly 5 mg tirzepatide.
LenzFalse1/10
Once-weekly 5 mg tirzepatide produced −15.0% mean weight change at week 72.
AcceptReject
Check a section in Claude or ChatGPT.
Connect Lenz to Claude or ChatGPT with one sign-in and ask it to check a manuscript draft, literature review, or slide script right in the thread.
- Extract the statementsSend the draft section; Lenz pulls out the checkable claims and DOIs.
- Triage with a quick checkGet a fast first read and confidence level per statement in ~15 seconds.
- Run a deep check on key linesEstablish the verdict with primary papers, registries, and the reasoning in about a minute or two.
Check the endpoints, doses, and DOIs in this summary with Lenz.
Lenz · deep check · 3 statements
Lenz checked three statements. One weight-reduction figure swaps the 5 mg and 15 mg arms:
- False5 mg tirzepatide produced −15.0% mean weight change at week 72, not −20.9%.
- True ×22,539 adults enrolled under stated BMI criteria; 5/10/15 mg vs placebo over 72 weeks.
In Google Docs, or against your reference pack.
Install the Google Docs add-on to review findings in a sidebar beside your text and apply suggested corrections in place, or wire an automated check step in n8n or Zapier.
When your team works from an approved reference pack, source dossier, or claim matrix, pass those documents through the API (supplied_sources) so Lenz checks the draft against them alongside Crossref, OpenAlex, and PubMed.
# one command, then a sign-in
$ claude mcp add --scope user --transport http lenz https://lenz.io/mcp
# check a manuscript draft from your workspace
> check the claims and citations in drafts/surmount1-summary.md
{
"statement": "At week 72, mean weight change was -20.9% with 5 mg tirzepatide.",
"verdict": "False",
"lenz_score": 1,
"sources": ["nejm.org", "accessdata.fda.gov"]
}
04The record
Route the draft for review with its primary sources attached.
Every statement comes back with its own record: the claim as Lenz read it, the verdict and its score, the dated sources, and the case made for and against it before peer, author, or MLR review.
1–2min
Deep check
The record comes from a deep check. Lenz researches primary publications, biomedical registries, and official records, has models argue both sides, and returns the verdict with those sources and the reasoning, in about a minute or two per statement.
~15sec
Quick check
A panel of models returns a verdict and a confidence level in about fifteen seconds, for triaging a long manuscript or technical report. When the deep check disagrees, rely on the deep check.
At week 72 of the Phase 3 SURMOUNT-1 trial, the mean percentage change in body weight was −20.9% with once-weekly 5 mg tirzepatide.
At week 72 of the Phase 3 SURMOUNT-1 trial, the mean percentage change in body weight was -15.0% with once-weekly 5 mg tirzepatide.
- nejm.orgPrimary Phase 3 SURMOUNT-1 report
- accessdata.fda.govFDA prescribing information & review
- doi.org10.1056/NEJMoa2206038
- +19 more
Illustration · record fields abridgedlenz.io/c/…
05Scope
What Lenz checks, and what stays with your reviewers.
Checks
The factual statements and cited references in your draft, against scientific registries and public sources.
Sample sizes, p-values, confidence intervals, benchmark figures, trial enrollment criteria, dosing arms, primary and secondary endpoints, DOIs, Crossref, OpenAlex, Semantic Scholar, and PubMed records—plus your own reference pack or source dossier when supplied through the API.
Doesn’t check
Not a substitute for peer review, scientific sign-off, or client MLR approval.
Lenz does not evaluate experimental methodology, judge fair balance, write or validate Important Safety Information (ISI), approve promotional claims under FDA OPDP, EMA, or ABPI rules, provide medical advice, or guess whether prose was written by an AI model.
Warranty
Vendors disclaim accuracy. Lenz warrants the result.
On Pro and Scale, qualifying checks carry a contractual accuracy warranty, at no extra charge.
A deep check qualifies when it returns True or False with high panel confidence and enough independent sources behind it. Its certificate is a record you can keep alongside the reference pack.
06Pricing
Try free, no card.
$7.99/ month
500 quick checks or 50 deep checks a month, in any combination.
Or $6.59 a month, billed $79 a year
Choose Plus$99/ month
5,000 quick checks or 500 deep checks a month, in any combination. Qualifying checks carry the accuracy warranty.
Or $83 a month, billed $990 a year
Choose Pro$399/ month
20,000 quick checks or 2,000 deep checks a month, for production volumes. Qualifying checks carry the accuracy warranty.
Or $333 a month, billed $3,990 a year
Choose Scale07Questions
What Lenz checks, and how manuscripts and drafts are kept.
Not answered here? Talk to us
Yes. Upload a .docx in the Workbench on desktop and download the annotated .docx with Word comments on flagged statements and citations and suggested edits written in as tracked changes. Review any suggested edit before accepting it: a suggested edit is not itself verified.
Yes. Lenz resolves DOIs and research papers across Crossref, OpenAlex, Semantic Scholar, and PubMed, and checks whether the retrieved paper or abstract supports the sentence. When a full text is behind a journal paywall and the abstract does not settle the statement, Lenz marks the citation as needing a closer look rather than guessing.
No. Lenz catches swapped figures, misquoted study endpoints, and broken or unsupported references before you circulate the manuscript, whitepaper, or submission pack, so co-authors, scientific directors, and MLR reviewers spend their time on scientific interpretation and compliance rather than hunting down citation errors.
Yes, through the API you can supply up to 50 reference documents, such as primary papers, technical specs, or an approved claim matrix, so Lenz checks the draft against those documents alongside Crossref, PubMed, and public sources. Those checks stay private and do not qualify for the warranty.
When no public record settles a statement, Lenz answers “No Public Record” instead of a score. That doesn’t make the statement false.
Every check is private by default, with account data and uploaded documents stored in the EU. On the Pro and Scale plans you can enable the zero-retention option so submitted text is never written to Lenz’s database.