About
Research runs on data that’s stuck in PDFs
Every review, audit and meta-analysis starts with someone copying numbers out of papers by hand. We’re building PDFSifter, AI that does the copying and shows its sources, so that time goes into the research instead.
Our story
Why we started
[The moment it started: the project, the pile of papers, and what went wrong copying them by hand.]
[What you tried first, and why nothing showed where a value came from, or coped with scans and forms.]
[What PDFSifter does differently, and who it’s for today.]
What we believe
Four rules we build by
Show the source
A value without its quote and page is a guess. Every value PDFSifter extracts points back to where it came from.
Measure, don’t promise
AI accuracy is measured on your own examples, and anything uncertain is flagged for a person.
Your data stays yours
Files stored in the EU, AI providers that keep nothing, and PDFs deleted on your schedule.
Price it fairly
You pay by the page, and a retry only counts the documents that failed.
Team
The people behind PDFSifter
[Name]
[Role]
[One line on background and what they work on.]
[Name]
[Role]
[One line on background and what they work on.]
Want to help?
We’re always happy to hear from researchers and engineers who care about this problem.
Get in touch →
See it on your own papers
Your first 200 pages every month are free.