check_circleKey results with statistics exactly as printed
check_circleEvery item cites its section or page
check_circleUnstated fields marked "not stated", never guessed
check_circleCross-paper methods matrix in XLSX and CSV
How it works
1Upload paper PDFsOne paper per file — a single PDF or a whole journal-club batch, up to 30 papers per run.
2Read page by page, then structuredEvery page is transcribed verbatim — numbers and table rows exactly as printed — then each paper is distilled into structured notes with per-item page citations.
3Download notes and matrixOne notes DOCX per paper plus a combined methods-and-results matrix in XLSX and CSV for cross-paper comparison.
Who uses this tool
The same file, different jobs. Tap a prompt to try it.
GRADUATE STUDENTS
A literature review that cites itself
Every extracted claim carries its section and page, so writing up means checking, not hunting.
JOURNAL CLUBS
Ten papers, one matrix
Batch the week's readings and compare designs, samples, and results side by side.
SYSTEMATIC REVIEWS
Extraction you can audit
Statistics are copied as printed and unstated fields say "not stated" — nothing is silently filled in.
META-ANALYSIS
Effect sizes as printed
Ns, CIs, and p-values come out character for character, ready to verify against the PDF.
CLINICIANS
Evidence at a glance
Design, sample, and outcomes per paper, with the authors' own limitations stated plainly.
R&D TEAMS
Scan a field fast
Turn a folder of PDFs into comparable notes and a matrix your whole team can filter.
SECURITY
Built for sensitive documents
Bank statements, medical files, case records. Security isn't a feature we added — it's the foundation.
lock
Encrypted in transit
Encrypted in transit over TLS 1.2+, and stored in access-controlled, encrypted object storage. Files are protected the moment they leave your browser.
admin_panel_settings
Your files stay yours
Workspaces are isolated per account. Role-based access shows teammates only what they need.
auto_delete
Deleted, not stored
Files are deleted after processing. Everything runs on our own hardware and is never sent to an outside AI service, so your data is never used to train models.
dnsProcessed on our own hardwareauto_deleteDeleted after processinglockEncrypted in transit
Questions
No. Statistics are copied character for character as printed — effect sizes, confidence intervals, p-values, Ns, coefficients. If a value isn't in the paper, the field says "not stated" rather than an estimate.
Every extracted item carries a source naming the section or page it came from, e.g. "Methods, p. 4" or "Table 2, p. 7", so you can verify against the PDF before citing.
That paper's notes and matrix row are marked "extraction failed — review manually" instead of being silently invented, and the rest of the batch still completes.
Processed on our own servers, never used to train any model, deleted 7 days after your job completes.