Split multiple invoices from one PDF

A client emails one PDF. Inside it: a three-page utility bill, four receipts photographed at angles, two supplier invoices, and a bank statement, in whatever order the scanner produced. They scanned the stack in one pass, so it lands as a single file. This is not an edge case, it is the everyday input for a bookkeeper, and it is exactly what untxt. is built to take apart. You upload the file and choose how to process it; for a bundled PDF, that is the Auto-split & sort mode.

At upload you pick one of three modes:

• One document per file: each upload is treated as a single document (PDF, JPG, PNG).
• One document per page: splits a PDF one page at a time.
• Auto-split & sort: the mode for a bundled file.

With Auto-split & sort, untxt. reads the whole PDF, understands where one document ends and the next begins, and groups the pages that belong together. Page 4 stays attached to page 64 if they are the same six-page invoice spread across the file, and anything that does not belong in the books is set aside rather than forced into them. Each document is then classified on its own (invoice, receipt, credit note, bill) and extracted separately, with its own vendor, dates, totals, and line items, keeping tables intact. That mode takes a few extra moments, because the boundaries and document types are understood before anything is extracted, not after. No page-break drawing, no separate upload address per document type, no template that breaks when a client reorders the file.

untxt. keeps a document whole however many pages it runs to, and tells two near-identical invoices apart when they sit directly back to back. So a file where similar invoices run together, or where one document is buried among others, still comes back as separate, correct records. This is the case fixed page-break splitting gets wrong most often.

On cost, the rule is simple: one credit covers a file of up to five pages. Longer files, like a multi-page bank statement or a thick bundled PDF, use additional credits. Flex starts at $0.32 with the first 100 free, so running a real client file through to see how the split lands costs nothing.

untxt. even handles the classic mess of several receipts scanned onto one page: it detects the separate items, splits them apart, and shows them in the review window as "multiple items detected, viewing 1 of 3" so you can step through each one. The review overview lists every document with a confidence score, so anything it is unsure about, a faded total or a handwritten note, is surfaced for you to check rather than buried or guessed at. You confirm or correct it. Each vendor stays its own entry, and everything waits for your review before it reaches your books.

If a tool makes you separate the file first, it has solved the easy half and left you the tedious half. Splitting the ugly batch is the whole job, not a precondition for it. Drop the file in as it arrived, review the handful of exceptions, and send clean records straight to QuickBooks or Xero.