How it works

1. Read it on your device

Open a PDF, a scan or a photo from your phone; a bill can span several pages. Your own browser does the reading. A PDF that carries a text layer (the kind a provider or insurer portal issues) is read directly. A scan or a photo has no text to read, so your browser runs text recognition on it, on your machine. The file is never uploaded, and it stays in your browser’s storage.

You choose which country the bill is from, because billing rules differ. The United States has the deepest support today. If a photo looks hard to read, we say so straight away rather than producing a confident-looking result from an unreadable page.

What comes out is a structured record: provider, dates, each line item, each amount, and for each value the page it came from and how confidently it was read. Anything we cannot read confidently stays empty. Nothing is guessed. A patient’s name, account number, date of birth or member ID found on the bill is separated out on your device and never sent anywhere.

2. Check what we read

Before any analysis, you see what we read and can correct it. Corrections are marked as yours and are trusted completely from then on. Everything up to this point is free and needs no account.

3. Unlock the findings

When you want the checks run, you pay for this bill ($6.99) or subscribe. Only then does your browser send the figures you just checked to our checking function: the line items, amounts, dates, billing codes and provider name, plus your corrections and the country. Not the file. Not a page image. Not the patient’s name. The function runs the checks in memory, returns the findings, and keeps nothing. That is the only time anything about your bill leaves your device.

The checks are ordinary code, not a language model: multiplication, addition, date comparison, and text similarity. That matters, because arithmetic should be exact and reproducible.

• Duplicate line items and repeated billing codes
• Quantities on services usually provided once
• Line totals against quantity times unit price
• Line items against the stated subtotal and total
• Insurance payment, adjustment and patient responsibility against the billed amount
• Dates that conflict with each other or with the statement date
• Missing information you can request
• For subscribers: your bill against your Explanation of Benefits or your Good Faith Estimate

Each finding says what we found, why it may matter, what to verify, the evidence from your document, and how confident we are. From those findings your browser builds a question checklist, a neutral message you can send to the billing department, and a PDF report, all produced on your device. You can edit the message, copy it, or download the report.

Paying, and coming back

Payment is handled by Stripe on Stripe’s own pages; we never see your card. When you return, your browser keeps a signed receipt for the bill you paid for, or a pass for your subscription, so you are never asked to pay for the same bill twice and a subscriber is never asked to pay again before the renewal date shown on the billing page.

There is no account for a single bill. Subscribers can sign in on another device: enter the email you used at checkout and we email you a link. That is the only kind of account that exists.

What we will not do

We will not tell you a charge is fraudulent, illegal, or definitely incorrect. We will not tell you what you owe or what you are entitled to. We will not explain what a billing code means unless we can verify it against an authoritative code dataset. And we will not invent information that is not on your document.

Confidence in a finding is calculated from how strong the signal is, how legibly we read the values it depends on, and the quality of the document. It is not a model’s opinion of itself.