About PaperPull

Built from a problem in research.

PaperPull is a tool for turning scientific papers into structured datasets. It is designed to make research data extraction faster, more transparent, and easier to reproduce.

Why I built it

From one meta-analysis to a tool for other researchers.

PaperPull began with a problem I encountered while conducting a meta-analysis: extracting results from scientific papers was slow, repetitive, and surprisingly easy to get wrong.

Modern AI models seemed well suited to this work. They could read papers and extract structured information far more quickly than I could, while applying the same instructions consistently across every study.

I started building PaperPull to make that process easier for my own research—and then decided it should be available to other researchers too.

Open and inspectable

Research software should not be a black box.

PaperPull is being prepared for release as an open-source project. Researchers will be able to inspect how it works, verify its methods, suggest improvements, or run their own version.

GitHub repositoryComing soon
How access works

No PaperPull subscription or model markup.

PaperPull currently gives researchers a way to use modern AI models without adding another subscription between them and the provider.

01

Start with a free trial

New users can try PaperPull before connecting a provider.

02

Use your own OpenRouter account

After the trial, connect OpenRouter and choose the model that fits your work.

03

Pay the provider directly

PaperPull does not add a markup to the model costs you incur.

Contact

Questions, feedback, and contributions are welcome.

tobiasgedwards99 [at] gmail [dot] com
Try PaperPull