Disclosure. We make Paperlock, which is one of the apps this page is about. The test at the end works on any of them, including ours.

What "ask your documents" has to mean

Three things have to happen for an app to answer "how much did I spend at the hardware store in August?" from a pile of receipts.

  1. Reading. Each receipt becomes a record: store, date, line items, total. Photos of receipts are hard to read; thermal paper fades, totals sit next to subtotals, and the store name is a logo. Misreads here become wrong answers later, so the record has to be visible and correctable.
  2. Retrieval. The question has to find the right records. "Hardware store" should match receipts from a store the app knows sells hardware, in August, this year. This is where filename search dies and where the plain-language layer earns its keep.
  3. Composing the answer, with sources. The total is the sum of the totals of the receipts retrieved. That arithmetic is trivial once the retrieval is right, and impossible to trust if you cannot see which receipts went into it.

The third point is the one to judge apps on. An answer with the receipts attached can be checked in five seconds. An answer without them is a number.

The failure modes to look for

  • Duplicate receipts. The same receipt photographed twice should count once. Ask a total, then check the sources for repeats.
  • Refunds versus purchases. A refund slip looks like a receipt. A good tool treats it as negative or separates it; a poor one adds it to the total.
  • Date ranges. "Last August" and "this August" are different months. Watch which one the app picked; the sources tell you.
  • Currencies. A holiday receipt in a different currency should not be added to home spending as if it were the same unit.
  • Missing information. A receipt with no readable total should be flagged, not silently skipped or guessed.
  • Not in the vault. Ask about a store you never visited. The correct answer is "I can't find that in your documents," not a plausible number.

The fifteen-minute test

Run this on any app before paying for it, ours included.

  1. Import ten real receipts, including two from the same store, one refund, and one that is blurry. The guide on how to scan receipts covers getting clean images.
  2. Open three of the records and check the store, date and total. Correct one deliberately and confirm the correction sticks.
  3. Ask: "How much did I spend at [store] in [month]?" Then open the sources. Are they the right receipts? Is the refund handled? Is the blurry one included, excluded, or flagged?
  4. Ask: "Which receipts made up that total?" The list should match the sources from step 3.
  5. Ask about a store you never visited. The app should say it cannot find it.
  6. Ask a date question: "When does the warranty on [item] end?" from a receipt or warranty card you imported. Then correct the date and check any reminder moves with it.

If an app passes all six, the pitch is real for your paperwork. If it fails step 3 or 5, no dashboard makes up for it.

Where Paperlock stands

Paperlock is built around steps 1 to 3: it reads each document into a record you can inspect and correct, files it by type, sender, date and amount, and answers questions with the source documents attached. It says when the answer is not in the vault. Expiry dates it reads become reminders (30 and 7 days before, 6 months for passports and IDs) once you turn reminders on, and correcting a date updates the reminder. Documents are encrypted on the phone; pages are sent to our servers only for the time it takes to read or answer, and OpenAI is the processor. The free tier is 25 documents and 5 questions a month, enough for the test above.

What Paperlock does not do: spending dashboards, category budgets, or bank sync. If you want merchant and category totals as charts, LumenDoc advertises exactly that, and a budgeting app does it from transactions. Our lane is the question, the answer, and the paper behind it.

The manual alternative

None of this requires an app. The guide on the best way to store receipts digitally describes a folder-and-spreadsheet system that answers the same questions with more typing, and the purchase record builder does the record-keeping half in your browser. The app removes the typing and adds the questions; the paperwork discipline is the same either way.