Guides › SortedScan workflows
SortedScan workflows
A workflow turns a copier button into a filing rule: how a stack is split into documents, what is read off each one, how the file is named and where it goes. Set one up, point a scanner folder at it, and scans file themselves.
1Where it goes
Open SortedScan › Workflows › New workflow. Give it a name that will make sense on the copier, like Invoices to SharePoint, and pick a Destination: SharePoint, OneDrive, Teams, an on-site computer or a webhook. Destinations are shared with SortedPost, so one you've already set up is there to choose.
For common paperwork, start from a template: Invoices, Receipts, Contracts or Identity documents. One click fills in the fields the AI should read (supplier, invoice number, dates, net, VAT and total for an invoice, say), a filing path and a file name that use them, and switches OCR and AI on. Everything stays editable afterwards.
Hover over (or tap) the ⓘ next to it in the app for a fuller explanation.
2How a stack is split
| Split on separator sheets | A printed sheet between documents starts a new one. Reliable and cheap; the sheet is removed from the scan. |
|---|---|
| Split on barcode cover sheets | A page with a barcode starts a new document, and the barcode's value can fill an index field. For client packs and case files. |
| AI decides where each document starts | Reads the text and finds the boundaries itself, from page numbers, letterheads and sentences that run on. Separator sheets are still honoured. Needs OCR on. |
| Every few pages (fixed count) | Every N pages become a document, for stacks of same-length forms such as two-page timesheets. A separator sheet still ends a document early, so a form with a page missing doesn't put the rest out of step. |
| One document per scan | The whole stack is one file. For single documents and things that are already sorted. |
3What to read off each document
Index fields are the values captured for each document: a client name, an invoice number, a date. Each has a label, a type (text, date, number or a list to choose from) and a source:
| Read from a box on the page | Upload a filled-in example of the form and draw a box around the value. Every scan is read in the same spot. Options for first line only and keep only a pattern, such as the invoice number out of a longer line. No AI, so it's exact and fast. |
|---|---|
| From barcode | The value of the barcode on the cover sheet or first page. |
| AI extracts it | Ask for the value by name and the AI finds it in the text, wherever it sits. For documents that vary in layout. |
| Fixed value | The same value for everything this workflow files. |
| Typed at review | Left blank for a person to fill in before filing. Only useful when Check by a person is set to every scan. |
Tick Required on a field that must be filled: a scan missing it is held for someone to complete, even in a workflow that otherwise files straight away.
A Format check on a field is a pattern the whole value must match, for example INV-\d+ for invoice numbers or \d{6} for a six-digit account. A value that doesn't match is flagged, and held for a person when the workflow checks only when unsure (step 6). Letters are matched regardless of case.
Reference lookup goes a step further. Upload a reference table, say your client list, pick the field that matches a column, and other fields fill from that row. A client code read off a form can then supply the client's name and folder.
4Name and path
File name and Filing path are templates built from tokens. Date tokens are always available: {year}, {month}, {day}, {date}, {datetime}. Every index field adds its own token, shown next to the field as you add it. So a path of {client}/{year} and a file name of Invoice {invoice_no} file an invoice as Acme Ltd/2026/Invoice 10423.pdf. The form shows a worked example as you type. Leave the file name blank for the workflow name plus date and time.
5Processing
| Make searchable (OCR) | Reads the text so files can be searched in SharePoint or OneDrive. Off files the scan exactly as the copier sent it, which is the fastest option and fine for copiers that OCR on the device. |
|---|---|
| AI index fields | Lets "AI extracts it" fields fill. Off leaves them blank. |
| Drop blank pages | Removes empty reverse sides. Photos and drawings are kept. |
| OCR accuracy, Straighten pages, Compress after OCR | Fine tuning. Fast and the defaults suit normal office print; Best is for faint or tiny type. |
| Clean up pages | Removes specks, dark scanner edges and punch-hole marks. Only the outer 20 mm of each page is checked for dark areas, so logos, photos and dark table headings are left alone. Every page is reprocessed, so it's slower; use it for old or much-copied paper. |
| Redact | Blacks out card numbers, bank details (IBANs, sort codes and account numbers), National Insurance numbers, email addresses, phone numbers, or anything matching your own patterns, one per line. The area becomes solid black and the hidden text under it is removed, so it can't be copied out, and the page is read again so the rest stays searchable. Only the filed copy is redacted: the original scan stays in Sorted Suite for your retention period, like every scan. |
| File format | Searchable PDF/A, or multi-page TIFF for document systems that want it. |
| Also file an editable Word copy | A .docx with the same name next to each PDF, holding the scanned text for editing and copying. Layout, tables and pictures aren't rebuilt, so the PDF stays the record. Not available with password protection, since an open Word file would give away what the password protects. |
| Metadata file | A .json or .xml with the same name next to each document, for document systems that import from a folder. It holds each field's value, where it came from (AI, box, barcode, fixed, typed or lookup) and whether it passed its checks, plus the page count, filing path, scanner and scan time. Optionally it can also carry the document's text and who reviewed it and when. |
| Watermark | Text stamped on every page of the filed PDF, small at the foot or large across the page. Tokens work as in the file name, so {client} · Scanned {date} stamps each file with its own client and date. |
| Password-protect PDFs | Filed PDFs need a password to open (AES-256). Admins can replace the password but can't read it back. Anyone who opens a file can still print and copy from it. |
The metadata file and Word copy go to SharePoint, OneDrive and folders, including on-site computers running Sorted Connector 0.5.0 or later. Webhooks already carry the fields in the message.
6Before filing
Check by a person decides which scans wait in Scan jobs before they're filed:
| Never | Scans file straight away. Only one with a required field still blank is held. |
|---|---|
| Only when unsure | Scans file straight away unless the AI flagged a value it wasn't sure of, or a value failed its format check. The doubtful fields are highlighted for whoever checks it. The usual choice with AI fields. |
| Every scan | Everything waits for someone to check and fill in the fields first. |
Active off pauses the workflow.
7Point a scanner at it
On Scanners, give the copier a folder named after the workflow and choose the workflow under Sends to. On the copier, that folder name goes in the Directory field of a second destination using the same login, with its own one-touch key. Copiers without a Directory field use the folder's own scan-to-email address instead. One copier can feed as many workflows as it has buttons.
From a phone
Not everything comes through a copier. Open Capture in the app on a phone, choose the workflow, and tap Take photo for each page. Reorder or remove pages, then Send: the photos are turned upright, made into one document and filed by the workflow exactly like a scan. Lay the page flat in good light, fill the screen with it and keep your shadow off it.
Day to day
SortedScan's home page lists recent scans: what was filed where, and anything To review. A held scan shows its fields ready to fill or correct, with any the AI was unsure of, or that failed their format check, highlighted. Then Approve & file. A scan that failed, because its destination was unreachable, shows why and can be retried once the destination is fixed. The Dashboard has a SortedScan section: how much was filed and by which workflow, scanner and day of the week, how scans arrived (copier, email, upload or phone), pages over the period, anything waiting, and failures.
A workflow you no longer need is best paused (Active off), which keeps its history. It can only be deleted while no scanner points at it and nothing has been filed through it.