✂️ Split Multi-Page PDFs into Separate Documents
Document split allows you to break a multi-page PDF into several separate documents and parse each one individually. This is useful when multiple documents — such as invoices, contracts, or statements — arrive bundled together in a single file.
How to configure it
Go to Inbox Settings → General and scroll to the Document split section. Choose one of the four modes below.
1. Split every N pages
Divides the PDF into equal chunks of N pages. For example, setting N to 2 on a 6-page document produces three 2-page documents.
Best for: batches where every document has the same fixed page count.
2. Split by page ranges
You define exactly which pages form each document using a comma-separated list. Use the same syntax as Page processing:
1-3, 4-6— two documents: pages 1–3 and pages 4–61, 3-5, (1)— pages 1, then pages 3–5, then the last page(3)-(1)— last three pages as one document
Pages not covered by any range are dropped. Page numbers are 1-based.
3. Split by keywords
Airparser scans the text of each page and starts a new document whenever it finds a matching keyword. You can add multiple keywords and configure each one independently.
Options for each keyword:
Position — split before the page that contains the keyword (the keyword page starts a new document) or after it (the next page starts a new document).
Regular expression — treat the keyword as a regex pattern for advanced matching.
Case sensitive — by default matching is case-insensitive; enable this to require an exact case match.
Every page is always included in some output document — no pages are ever dropped.
4. Split using AI
The AI automatically detects where each document begins and splits accordingly. This is the most flexible mode — it handles mixed document types and variable page counts without any manual configuration.
You can add optional instructions to guide the AI:
This PDF contains multiple invoices merged into one file. Each invoice starts on a new page with a vendor name and "Invoice #" header. Split into one document per invoice and name each document after the vendor.
The AI also names each resulting document based on its content, so you can identify them at a glance.
No pages are ever dropped — every page is guaranteed to appear in exactly one output document regardless of what the AI detects.
Output document names
Each split document is named automatically using the original filename and the page range it contains. For example, a file named batch.pdf split into pages 1–5 of 20 produces batch pages 1-5 of 20.pdf. When using AI split with named segments, the AI-suggested name is used instead.
Interaction with Page processing
If Page processing is also enabled, it runs first — pages are filtered before the split. The split then operates on the already-filtered PDF.