Split and named by content in one step

Split a PDF automatically and name each file by its content

Last updated:

Upload one PDF straight from the scanner. The AI reads each page, finds where a new document starts, splits the file, and names each file by what is on it, for example INV-2025-0281_Acme-Corp.pdf.

30 free pagesNo credit cardDeleted after processing

With Docusplit you upload a multi-page PDF, the AI finds where a new document starts, splits the file, and names each file by its content (invoices as InvoiceNumber_Company, other documents as DocumentType_Sender_Date); you get back a ZIP with the single PDFs and a CSV overview.

The scanner hands you one file, your bookkeeping needs thirty

In between sits hand work, because a page splitter has no idea where an invoice ends.

Cutting it up page by page

You open the scan, find where the next invoice starts, mark its pages, and save them as a file. Thirty documents later the morning is gone, and somewhere around the tenth one a first page slipped into the wrong file.

Paying for a license and still cutting by hand

An Acrobat subscription splits by page count, file size, or bookmarks, not by document. You still find the boundaries yourself, and the output files are called page-1.pdf.

Continuation pages land in the wrong document

A three-page invoice turns into three files, and the attachment to a contract sticks to the next reminder letter. You usually notice when your accountant asks.

Splitting by hand or with the AI, line by line

Worked example: assume one minute per page by hand, so a 50-page scan takes about 50 minutes. The app budgets about 6 seconds per page in Invoice mode and about 10 in Document mode, so 5 to 8 minutes, plus your check of the CSV.

CriterionBy handWith the AI
Where a document startsYou look at each page and decide yourselfThe AI reads the letterhead, page counter, and invoice number
Time for 50 pagesAbout 50 minutes if you allow one minute per pageAbout 5 to 8 minutes at 6 to 10 seconds per page, plus your check
Multi-page documentsContinuation pages and attachments assigned by handStay together, recognized by page counter or a missing new letterhead
File namingBy hand, file by fileAutomatic, by content, e.g. INV-2025-0281_Acme-Corp.pdf
CheckingWhile saving each fileCSV overview with confidence; check files named Document_Page_N.pdf first
CostYour working time30 pages free, then from €9.99 a month for 100 pages

Splitting by content compared with iLovePDF, PDF24, and Acrobat

Page splitters are the better choice for many jobs, often free and offline. The difference is naming by content without templates. As of August 2026; prices on each vendor's site.

ToolSplits by document contentNames files by contentSplits by page numberPrice
DocusplitInvoiceNumber_Company or Type_Sender_Date, no templates30 free pages, then from €9.99 a month
iLovePDFSmart range mode since May 2026, with category presets or a promptYes, according to the vendor (Smart range mode)Free allowance, Premium on the vendor's site
PDF24Free, ad-supported, also offline as a Windows app
Adobe AcrobatSubscription, see the vendor's site

Three steps from scan to named files

You pick a mode, the AI does the rest.

1

Upload the PDF

Drop a scanned or digital PDF of up to 50 MB into the upload. Before that, pick Invoice mode for a stack that is all invoices or Document mode for mixed mail.

2

The AI reads each page

It looks for letterheads, page counters, and invoice numbers, and starts a new file wherever a new document begins. Continuation pages stay with their document. Expect roughly 6 to 10 seconds per page, with progress shown live.

3

Download the ZIP

You get one PDF per document, named by invoice number and company or by document type, sender, and date, plus a CSV overview. Optionally sorted into folders by month or by sender.

The AI splits by content, not by page count

That is what sets it apart from a page splitter, and it decides if you have to go back in afterward.

Boundaries read from the page

A new letterhead, a page counter like 'Page 1 of 3', or a different invoice number are the signals the AI reads on each page. It splits where a document really starts, and you never type page numbers.

Continuation pages stay together

A three-page invoice comes out as one file, not three. The AI recognizes continuation pages by page counters, a missing new letterhead, or text that carries on mid-sentence, with no separator sheet and no barcode.

Deleted after processing

Files are transferred over TLS, kept only while processing, and then deleted from our servers. AI analysis runs through the OpenAI API under a data processing agreement and is never used for training.

How the AI spots a document boundary

The AI receives each page as an image and reads it the way you would on screen. A new letterhead with a different sender, logo, or footer is the strongest sign that a document starts here. So is a salutation, a date of its own, or a subject line at the top of the page.

Page counters such as 'Page 2 of 3' tell it that a page belongs to the previous document. In Invoice mode it is also told which invoice number and company appeared on the previous page. If the same number shows up again or no new header appears, the page counts as a continuation; a new number with a header of its own opens the next invoice, even when the sender is the same.

When the layout breaks without any of those clues (a different typeface, a different grid, a table instead of a letter), the AI treats that as a boundary. When the text instead carries on mid-sentence or mid-table, or the page holds nothing but totals, a signature, and a footer, it stays with the previous document. It does not read amounts, line items, or bank details, only what it needs for the split and the file name.

→ How invoice-only stacks are split

01
Letterhead and sender changes
02
Page counters like 'Page 2 of 3'
03
Invoice number of the previous page

Multi-page documents stay together

A page splitter knows pages, not documents. Tell it 'every 2 pages' and a three-page invoice turns into two files, neither of them complete. Separator sheets or barcodes only help if someone put them between the documents before scanning.

The AI decides page by page whether it opens a new document or continues the previous one. So a one-page receipt, a four-page invoice with an attachment, and a twelve-page contract can sit in the same scan. Each comes out as one file, with the pages in the order they had in the scan.

You can trace that in the CSV overview inside the ZIP. It lists, for each file, the page of the original PDF on which the document starts, so you can check an assignment afterward.

1
Different lengths in the same scan
2
No separator sheets or barcodes
3
Start page of each file listed in the CSV

What the AI can't do reliably

Handwriting, faded thermal receipts, and scans below about 200 dpi read poorly. Blank back sides from duplex scanning, a delivery note right behind its invoice, or an attachment without a letterhead can end up in the wrong file. A page on which the AI finds neither a sender nor a number is named Document_Page_7.pdf, or Invoice_Page_7.pdf in Invoice mode, using the page number from the original file.

That is why a CSV overview ships with each ZIP. Unrecognized fields show as 'Not detected', and the confidence column is a rough guide to which files to open first. With folders by month, documents without a readable date go to 'Unknown'.

You cannot enter page numbers, move split markers, or preview the result before download. If you only need pages 3 to 5 of a contract or want to halve A3 double-page scans, a page splitter such as PDF24 is the better tool. The output PDFs also get no text layer, the pages are copied unchanged, and the XML inside an e-invoice PDF is not carried over.

→ How the file names are built

Fallback name Document_Page_N.pdf
'Not detected' and confidence in the CSV
'Unknown' folder when no date is readable

What arrives in the ZIP

One PDF per document, named files, and an overview you can open in a spreadsheet.

  • Document boundaries read from page content, no separator sheets or barcodes
  • Continuation pages stay with their document ('Page 2 of 3', no new letterhead)
  • File names built from invoice number and company, or document type, sender, and date
  • Two modes chosen before upload, for invoice-only stacks or mixed mail
  • Reads scans and digital PDFs, tested in English, German, French, Spanish, Italian, and Dutch
  • ZIP with single PDFs, a CSV overview, and optional folders by month or sender
  • Confidence column in the CSV showing which files to check first
Uploaded
Scan_2025-03-15_47-pages.pdf
In the ZIP
INV-2025-0281_Acme-Corp.pdf

Three stacks that typically show up

Anywhere a document feeder packs mixed mail into a single file.

The receipts binder at year end

A binder of invoices, insurance letters, and tax assessments, fed through the scanner in one pass. The single file becomes separate documents with the date in the name, in folders by month if you want.

The day's incoming mail at the office

Supplier invoices, late-payment reminders, and a contract amendment in the same scan. Each invoice comes out as InvoiceNumber_Company.pdf, the amendment as Contract_Sender_Date.pdf.

A client's stack at the accounting firm

The client sends a year of receipts as one 80-page PDF. Split and named, the files go into QuickBooks, Xero, or Dext one by one, with the CSV overview for checking.

Frequently asked questions about automatic PDF splitting

How does the AI know where a document ends?▼

It reads each page as an image and looks at the letterhead, the sender, page counters like 'Page 2 of 3', and in Invoice mode the invoice number of the previous page. A page with no header, date, or subject line of its own that carries on mid-sentence or mid-table counts as a continuation page; a new sender or a layout break opens a new document.

Does it work with scanned PDFs?▼

Yes. The AI works from the page image, so it makes no difference if the PDF came from a scanner or from software. Good results need about 200 dpi; handwriting and faded thermal receipts read poorly. You can only upload PDFs, not image files.

How big can the PDF be, and how long does it take?▼

Up to 50 MB per file, with no limit on pages per PDF; only your page allowance counts. Expect about 6 seconds per page in Invoice mode and about 10 in Document mode, so a 30-page scan takes a few minutes. You see the progress while it runs.

What happens when the AI splits wrong or can't read something?▼

A page with no readable sender or number is named Document_Page_7.pdf, or Invoice_Page_7.pdf in Invoice mode, so it is easy to find. In the CSV overview, unrecognized fields show as 'Not detected', and the confidence column tells you which files to open first. If pages were wrongly merged, you split them afterward with a page splitter; the pages in the ZIP are copied unchanged, so nothing is lost.

Can I pull out specific pages by page number?▼

No. The AI decides from the content where a document starts; you do not enter page numbers or halve double-page scans. For 'pages 3 to 5 of the contract', a page splitter such as PDF24 or the split tool in iLovePDF is the right choice.

Is it free, and do I need an account?▼

The first 30 pages are free, and you do not need a credit card. A free account with an email address is required, because processing is tied to your page allowance. After that, 100 pages a month cost €9.99 (Starter), 300 pages €29.99 (Pro), and 1,000 pages €99.99 (Max); yearly billing is 20% less, you can cancel anytime, and there is a 14-day money-back guarantee. It all runs in the browser, with nothing to install.

Is my PDF safe when I split it online?▼

Files are transferred over TLS, kept only while processing and then deleted from our servers. AI analysis runs through the OpenAI API (data processing agreement with OpenAI Ireland, Standard Contractual Clauses, never used for training; OpenAI retains API data for up to 30 days for abuse monitoring). See our privacy policy for details.

Docusplit compared

How does Docusplit stack up against other tools? See the head-to-head comparisons.

Sources & further reading

  1. iLovePDF blog, Smart range mode for Split PDF (May 8, 2026) — The vendor describes the AI mode of its split tool, which splits by document category, preset, or a custom prompt and delivers named output files.
  2. PDF24 Tools, Split PDF — Tool page listing the modes pages per PDF, even/odd, halve pages, and scissors; free and ad-supported, servers in Germany, files deleted after one hour according to the vendor.
  3. Adobe Community, Split PDF and auto-rename each split document — Thread in which Acrobat experts explain that the Split feature cannot split and name files by content; the suggested routes are bookmarks or custom JavaScript.

Try it on your next batch scan

The first 30 pages are free so you can check the result on real documents. Splitting and naming happen in one step.

Upload a PDF now

30 free pages · no credit card · files deleted after processing

Last updated: