How Papira works — and what to expect
Read this before you buy. It saves us both time, and it tells you upfront whether Papira is the right fit for your papers.
What Papira does
Papira reads up to three research papers and builds a Coverage Map: the problems they tackle, the approaches they propose, the benchmarks they evaluate against, and the gaps between them.
It is a research positioning tool. It is not a summarizer, a citation manager, or a writing assistant.
What kind of papers Papira works best on
Papira is built for empirical research papers — the kind where you can clearly identify a problem, an approach, and evaluation with benchmarks or metrics.
- A problem the paper is trying to solve
- An approach the authors propose
- An evaluation with benchmarks, datasets, or metrics
- Pure theory papers (no empirical evaluation)
- Position papers, opinion pieces, or essays
- Literature reviews and surveys
- Humanities papers with non-comparative methods
- Papers without explicit benchmarks or quantitative results
This covers most of computer science, machine learning, NLP, systems, databases, robotics, much of computational biology, and quantitative subfields of the social sciences.
If you upload papers outside the supported types, Papira will still produce output — but the Coverage Map will be sparse and the gaps it identifies may not be meaningful. This is a scope limitation, not a defect, and is not grounds for a refund.
What you get
- A Coverage Map across your uploaded papers
- Identified gaps and contribution opportunities
- Up to 3 papers, 20 pages each, per analysis
What we cannot guarantee
Papira uses large language models to read and structure your papers. LLMs are probabilistic. That means:
- Two runs on the same paper may produce slightly different extractions
- Edge cases, unusual paper structures, or ambiguous writing can produce extractions you disagree with
- The "right" answer for a research gap is partly subjective
By purchasing, you accept that LLM-driven analysis is interpretive, not deterministic.
Refunds
We refund in exactly two situations:
Technical failure
If Papira fails to produce a Coverage Map at all — the analysis crashes, the output is empty, or the system errors before delivering a result — we refund in full, automatically. You don't need to ask. If you don't see the refund within 7 days, email us at help@papiraflow.com and we'll sort it out.
Charged but no service rendered
Duplicate charges, billing errors, or paying for an analysis you never received. Email us, we fix it.
We do not refund for:
- Disagreement with the extracted problems, approaches, or gaps (this is interpretive output)
- Quality concerns when papers fall outside the supported types described above
- Wanting different output than what was produced
- Buyer's remorse after viewing the result
We chose this policy deliberately. The cost of running an analysis is real and largely incurred before you see the output. A "satisfaction-based" refund policy at our price point would make the product unsustainable, which would hurt every researcher who depends on it.
How to contact us
Pre-purchase questions, technical failures, billing, or feedback.
Papira is run by a small team. We aim to reply within 1–2 business days.
Frequently asked questions
Can I upload more than 3 papers?
Not in a single analysis. Run multiple analyses if you need broader coverage.
Can I upload papers longer than 20 pages?
The limit is 20 pages per paper. Papers longer than 20 pages are not accepted — trim or split your PDF before uploading.
What languages does Papira support?
English only, for now.
Can I share my Coverage Map with collaborators?
Yes. The output is yours.
How is this different from ChatGPT or Claude reading my papers?
General-purpose chatbots don't structure output into a comparable matrix, don't surface gaps systematically, and don't extract benchmarks and scores in a queryable form. Papira does one job and does it consistently.
Will the price stay at $3.99 / $9.99?
The $3.99 first-analysis price is a launch welcome offer. The $9.99 standard price is a launch offer (regular price $19.99) and will increase. Anyone who buys at the current price is grandfathered for that purchase. Future analyses use the then-current price.
I'm a student with a tight budget.
We don't have student pricing or special arrangements right now — everyone pays the same listed prices. We may revisit student-friendly options later; they're not part of the launch.
The fine print
By purchasing an analysis, you confirm that you have read and understood:
- The supported paper types described above
- The probabilistic nature of LLM-driven extraction
- The refund policy described above
This is a one-time purchase, not a subscription. There is nothing to cancel.
Ready to map your field?
First analysis $3.99. No subscription. Results in minutes.
Build your research landscape