Power Automate Desktop can be built into a scan splitter: it has actions to extract page ranges into new PDFs and to read text with OCR, but you design the flow, the page logic, the naming rules and the error handling yourself. ScanRoute Local is a Windows app that does that one job out of the box: it splits a combined scan at Code 128 or QR separator sheets, blank pages or every N pages, reads one zone with Windows OCR, and builds each file name and folder from a template, with a preview before anything is written. If you already run Power Automate flows and enjoy building them, it is a capable toolkit; if you just want named PDFs in the right folders, ScanRoute Local is the shorter path.
Split combined scans into named PDFs, filed in your folders, on this PC.
Get it from Microsoft
What does Power Automate Desktop offer for PDFs?
Microsoft’s PDF actions reference says the actions “enable you to extract images, text, and tables from PDF files, and arrange pages to create new documents.” The ones that matter for scan splitting are:
- Extract PDF file pages to new PDF file, which takes a page selection such as 1,3,17-24 and writes a new file. This is the splitting step.
- Extract text from PDF, for a page or range. It reads existing text, so on an image-only scan there may be nothing to read.
- Extract images from PDF, which saves page images as PNG files to a folder.
- Merge PDF files. Microsoft notes that bookmarks in the original files aren’t included in the merged result, and warns against writing the merged file over one of the originals.
How does its OCR work?
The OCR actions reference says Power Automate supports the Windows OCR and Tesseract engines, that both “work locally without connecting to the cloud”, and that Windows OCR is the default and needs the language pack for the language you want to read. The Tesseract engine needs a CPU that supports AVX2.
The Extract text with OCR action lists its sources as Screen, Foreground window and Image on disk. So for a scanned PDF, a typical flow first saves the page as an image, then runs OCR on the image, then parses the text it gets back.
We did not find a barcode reading action in the PDF or OCR references we checked. If your office uses barcode separator sheets, check what your edition and connectors offer before you plan around them.
What would you build in Power Automate Desktop?
A working scan splitter needs every one of these steps, each as actions, loops and variables in a flow:
- Watch or loop over a folder of combined PDFs.
- Decide where each document starts. With no barcode action, that usually means OCR on every page looking for a marker, or a fixed page count.
- Turn each first page into an image and OCR the area holding the reference line.
- Parse the client, date, type or number from the text, and check it.
- Build a file name and folder, create the folder, and handle names that already exist.
- Extract the page range to a new PDF.
- Log what happened, and route failures somewhere a person will see them.
None of this is hard for someone who builds flows, but steps 4, 5 and 7 are where most homemade splitters break: a misread date silently files a document in the wrong place.
Power Automate for desktop also asks you to sign in. Microsoft’s introduction lists a Microsoft account, a work or school account, and an organization premium account, with features that differ by account type.
Preview every proposed name and folder before a single file is written.
Get it from Microsoft
How do they compare side by side?
| Power Automate Desktop | ScanRoute Local | |
|---|---|---|
| What it is | General automation platform | Single-purpose scan splitter and filer |
| Where a document starts | Your logic: OCR markers or page counts | Code 128 or QR separator sheets, blank pages, or every N pages |
| Reading a field | OCR on an image you produce, then your parsing | One zone drawn over the reference line, read with Windows OCR, pattern checks each field |
| File names and folders | Built with variables and actions | Template such as Client, Date, Type and Id placeholders |
| Review before writing | Whatever you build | Preview of every proposed name and folder; nothing written until you commit |
| Failed reads | Whatever you build | Inbox with the zone crop, to type, fix, accept or skip |
| Originals | Depends on your flow | Only read; never changed, moved or deleted; nothing overwritten |
| Record of output | Whatever you log | CSV manifest per commit with SHA-256 of every file |
| Searchable text | Not part of the PDF actions we read | Invisible text layer on upright scanned pages when an OCR language is installed |
| Account | Sign-in required | No account, no telemetry, no network connection |
Where does Power Automate Desktop win?
Be fair to it. It is the better choice when:
- the split is one step in a larger process, such as entering invoice data into another app or sending files on,
- you need logic ScanRoute does not offer, like reading several zones or branching on content,
- your organisation already standardises on Power Automate and has people who maintain flows.
ScanRoute Local reads one zone per document, works with PDF files in folders on this PC, and is not a document-management server or a scanner app. If you need more than that, a flow or a larger system is the right tool.
Where does ScanRoute Local fit better?
ScanRoute is made for small offices and solo practitioners who already scan paperwork to PDF on a copier or desktop scanner: bookkeepers sorting client statements, property managers filing vendor invoices, law offices filing correspondence, and clinic front desks filing supplier letters. The work is the same every day, and the cost of a misfiled document is high. The preview, the inbox for doubtful reads and the manifest are there so a person checks each name before it is committed. OCR and barcodes can misread, which is why ScanRoute asks you to check rather than filing silently.
If you are deciding how to mark document boundaries, our guide to barcode separator sheets explains the options, and our NAPS2 vs ScanRoute comparison covers splitting at scan time instead.
Frequently asked questions
Can I use both?
Yes. Some offices let ScanRoute produce the named PDFs and the manifest, then use a flow to act on the finished files.
Does ScanRoute need an OCR language installed?
It reads text with the Windows OCR language installed on this PC. Without one, splitting still works and the fields wait for you to type them.
Does either tool upload my documents?
The OCR engines in Power Automate Desktop run locally, per Microsoft’s reference. ScanRoute Local makes no network connection at all.
Will ScanRoute change my original scans?
No. Originals are only read, never changed, moved or deleted, and no file is overwritten.
Try ScanRoute Local
ScanRoute Local is a one-time purchase in the Microsoft Store with a free trial, so you can run it on a real stack of your own scans first. More detail is on the ScanRoute Local product page.
Try it on today’s scan pile and check every name before you commit.
Get it from Microsoft
Power Automate is a trademark of Microsoft. Automata Labs is not affiliated with Microsoft.
Last updated 22 October 2026. Published by Automata Labs, a software company in Canada.