Pick a scenario
Same-type documents in one stream
Use this scenario when a file contains multiple documents of the same type — for example, a stack of invoices from one vendor for a billing period. Each invoice has its own number and may carry page numbers; use that data to find boundaries.- Add an Extraction Rules activity to extract invoice numbers and page numbers.
- Add a Classify activity if the first page of each document differs visually from the others.
- Use the Splitter Script activity to compare extracted values across pages and decide where each new document starts.
Separate documents and store annexes
Use this scenario when documents are accompanied by explanatory pages (annexes) that must be retained but not extracted from.- Use a Classify activity to label each page as a primary-document page or an annex.
- Optionally add an Extraction Rules activity to flag pages with no extractable data — those are likely annexes.
- Use the Splitter Script activity to attach each annex to its parent document or output it as a separate document.
Separate documents and determine their type
Use this scenario when the stream contains documents of different types — for example, a loan application packet with identity documents, income statements, bank statements, and utility bills.- Use a Classify activity to label each page with its document type.
- Add an Extraction Rules activity to extract any data that signals the start of a new document.
- Use the Splitter Script activity to define rules for both segmenting and labeling each output document.
Reorder and clean up pages
Use this scenario when pages arrive out of order or include blank or garbage pages from a haphazard scan. Reordering is only possible if the pages carry an ordering signal — for example, printed page numbers.- Add a field to extract page numbers (or any ordering signal).
- Add a field that detects whether the page contains any text — pages with none can be treated as blank or garbage.
- Use the Splitter Script activity to reorder the pages and route blank/garbage pages into a separate output document.
Build the Document Splitter skill
1
Create a Document Splitter skill
Open Advanced Designer and click Create Splitter Skill on the start page.
2
Upload documents
On the Documents tab, upload your files. Each document set should contain the files for a single business transaction. The source files are converted into separate pages — every activity except the Splitter Script activity processes each page individually.
3
Add classification and extraction activities
Configure the processing flow to extract the data needed to identify document boundaries and types. Add a Classify activity when the stream contains multiple document types, or when first pages differ visually from the rest. Add fields and other activities as needed to capture data that helps separate documents of the same type or label document classes.
4
Configure the Splitter Script activity
Add document types on the Splitter Script Properties pane, and then write the script that turns the flow of pages into a set of documents. The script has access to every page in the transaction and can read data produced by other activities to decide which pages start a new document.
5
Test and publish
Click Test Skill Using Selected Documents to evaluate the results. When the results are good enough, publish the skill.