Papera › Use Cases
Built for the office worker drowning in invoices, the student photographing every lecture, and the developer who wants their AI to bring its own keys.
Linh · Office
I used to type vendor names and totals by hand. Papera gives me a CSV before my coffee is cold.
Pain solved. Manual data entry, lost paper trails, fear of cloud uploads from a corporate device.
Minh · Student
I search “covariance matrix” and the right page from week six opens. No more thumb-scrolling 200 photos.
Pain solved. Photographed boards that were impossible to find, blurry photocopies, no offline summarisation tool.
Tuấn · Developer
Apple Foundation by default, OpenAI for hard reasoning, my local model for volume. One app, three providers, my keys.
Pain solved. Vendor lock-in, no structured JSON output, no per-template prompts.
Linh · Office
Accounts payable at a 40-person agency in District 1. Receives 10–15 supplier invoices a day, half on email PDF, half on physical paper.
Yesterday’s deliveries — coffee supplier, courier, two printer cartridges, an office cleaner. Used to be 25 minutes of typing.
Tap shutter five times. VisionKit auto-detects edges and stitches each into its own document. No tap-to-confirm needed — the corner pulse is enough.
Papera recognised the layout (header, totals row, tax line) and assigned the Invoice badge with the matching Saigon Coffee Roasters vendor pre-filled.
The Foundation Model returns typed fields — vendor, invoice number, issued/due dates, line items, total — in 0.8s on her iPhone 15 Pro. She corrects one OCR mistake on a handwritten amount; the model relearns the layout for the next one from this vendor.
One Shortcut Linh built once: take Papera JSON → POST to QuickBooks bills endpoint. Five new bills appear in her ledger before her stand-up tea is half-finished.
Minh · Student
Second-year ML undergrad at HCMUT. Photographs every whiteboard, every handout. Has 312 photos in his camera roll labelled “lecture”.
The professor wrote it out across two boards in week six, in green marker. He took photos. He has no idea which photos.
VI · EN OCR runs on every scan and handwriting capture. Even the green marker on a glare-bright whiteboard. Search returns 4 documents — three from week six, one from week eleven.
Papera’s search jumps to page 3 of the week-six handout. The match is highlighted in yellow. The cross-reference to the week-eleven slides is right there in the same result list.
Saved as a recipe. He runs it across his “Linear Algebra · Lecture” folder and gets a markdown table of every formula from the semester. He pastes it into Anki.
Tuấn · Developer
Indie iOS developer in Đà Nẵng. Builds invoicing tools. Wants on-device for privacy, cloud only when reasoning gets hard.
Apple Foundation Models is on by default. He pastes his OpenAI key for hard reasoning, his Anthropic key for long-context contracts, his local Ollama endpoint for volume work. Each key is validated and stored in iOS Keychain.
For typed invoice extract → Apple Foundation, on-device, free, instant. For a 60-page master services agreement → Claude with 200K context. For a batch of 200 receipts → his local Llama. The same Papera UI; the model is swappable per recipe.
He saves “Extract reimbursable line items into [{date,merchant,amount,category}]” as a recipe. App Intent triggers it. Output ships via Shortcuts → his Cloudflare Worker → his accounting DB.
Across all three users
~95h
Linh’s annual time saved. Replacing manual invoice typing with batch scan + extract.
312 → 1
Minh’s lecture photos. One searchable library replacing a chronological camera roll.
3
Tuấn’s providers. Foundation Models · Claude · local Llama — switchable per recipe.
0
Documents that left a device. Across every workflow, by default.