Scan & Sort Business Correspondence
AI classifies every letter & notice
Last updated:
Scan your mail stack as one PDF and upload it. The AI splits each letter, reads type, sender, and date, and names the file, e.g. Notice_City_Utilities_2025-03-15.pdf. No separator sheets needed.
To scan and sort business correspondence, scan the stack as one PDF and upload it. The AI detects where each letter begins, keeps follow-on pages together, and names each file by document type, sender, and date, for example Notice_City_Utilities_2025-03-15.pdf. You get a ZIP of single files plus a CSV, optionally in folders by month or by sender.
Each type of letter carries its own deadline
Tax letters, late notices, and cancellation letters sit in the same stack as ads and bank statements.
Deadlines buried in the stack
An IRS letter, a late-payment notice, or a lease termination usually gives you a few weeks to respond. The clock runs even while the envelope sits unopened.
Retention rules nobody remembers
How long you keep invoices and business letters depends on your country, typically between three and ten years. Without the document type in the filename, you cannot tell later what can go.
One scan, 23 pages, one filename
Scan_Mail_Stack_23_Pages.pdf says nothing about what is inside. Finding the insurance letter means opening the file and paging through.
Four numbers behind letter filing
Two come from tax authorities, two from how Docusplit works.
General period for records that support a US tax return; longer in special cases
IRS, How long should I keep records?From the end of the last company financial year they relate to
GOV.UK, Company and accounting recordsOne PDF per upload, no page limit per file; your plan sets the page quota
Docusplit product dataProgress is shown live; a 30-page stack takes about 5 minutes
Docusplit product dataHandling letters by hand and with Docusplit
What changes in your filing once the AI reads type, sender, and date.
| Criterion | Manual | With Docusplit |
|---|---|---|
| Identify the letter type | Open and read each letter | The AI reads each page and writes the type into the filename and the CSV |
| Apply retention rules | Look up each letter | Filter the CSV by document type and set one rule per type |
| Find letters with deadlines | Page through the stack | Filter the CSV for Notice, Reminder, or Termination |
| Filename | scan_047.pdf, renamed by hand | Notice_City_Utilities_2025-03-15.pdf |
| Overview of the batch | Keep your own list | CSV with filename, type, sender, and date |
From mail stack to named files in three steps
One upload, then the AI reads each page.
Scan the stack as one PDF
Notices, insurance letters, reminders, and contracts can run through the feeder mixed. 200 to 300 dpi is enough.
The AI reads each page
It detects where a new letter starts, keeps follow-on pages together, and reads document type, sender, and date.
Download the ZIP
Each letter is named Type_Sender_Date.pdf. The CSV lists type, sender, and date for the whole batch; optionally the ZIP is sorted into folders by month or by sender.
The document type is in the filename before you open the letter
Three things you stop doing by hand after the upload.
Letter type identified
The AI tells a reminder from a notice, a policy, a contract, or an invoice and puts the type at the front of the filename.
Deadlines easy to find
Filter the CSV by document type to see the reminders, notices, and termination letters in a batch at a glance.
Files deleted after processing
Files travel over TLS, are kept only while processing, and are then deleted from our servers. AI analysis runs through the OpenAI API under a data processing agreement; your mail is never used for training.
Digitizing mail starts with one scan of the whole stack
Most offices still receive paper letters every day, among them invoices, insurance letters, letters from agencies, contracts, and advertising. They end up on a pile and stay there until a deadline looms or someone needs a document.
Digitizing the mail means scanning the stack right after it arrives and letting the AI do the sorting. It treats each letter as a separate document, reads the type (invoice, notice, contract, reminder), the sender, and the date, and names the file accordingly. The archive that results is searchable by filename alone.
For letters with deadlines, such as tax assessments, late-payment notices, or cancellation dates, the date in the filename shows how much time is left. Note that this is the date printed on the letter; legal deadlines usually run from the day it was delivered, so jot the arrival date down when it matters.
A sheet-fed scanner handles a daily stack in a few minutes; for a handful of letters, a scanner app on your phone is enough.
How the AI sorts letters by sender and date
After scanning a mail stack you have one PDF with 20, 50, or 100 pages, and the documents are mixed. The AI reads each page with a vision model. It detects where a new letter begins, reads document type, sender, and date, and keeps multi-page letters together, so a three-page insurance contract stays one file.
You choose the folder layout at download time. Without folders, the files sit flat in the ZIP. By month, each month gets a folder such as 2025-03/, and letters without a readable date go to Unknown/. By sender, each sender gets a folder. There are no folders by document type; instead the type sits at the front of the filename (Notice_City_Utilities_2025-03-15.pdf), so sorting by name groups all notices together.
The CSV that comes with the ZIP lists filename, type, sender, date, and a confidence value for each letter. Open it in Excel or Google Sheets to filter for reminders or notices, or use it as the index when you move the files into your document management system by hand; there is no direct connector.
What letter digitization costs with Docusplit
The free trial covers 30 pages once; you need a free account, not a credit card. After that, Starter is €9.99 a month for 100 pages, Pro €29.99 for 300 pages, and Max €99.99 for 1,000 pages, which works out to about €0.10 per page on any monthly plan and €0.08 on yearly billing (20% off). Your quota resets on your billing date, and you can split as many PDFs as you like within it.
A rough way to judge if it pays off, using our own assumptions rather than a study, is this. Say one letter takes you two minutes to open, name, and file by hand. Fifty letters a month are then about 100 minutes of filing. The app plans on around 10 seconds per page in document mode, so the same 50 pages run through in roughly 8 minutes while you do something else, and the naming comes out in one pattern.
What you do not get is a replacement for an archive system or an automatic upload into one. You get a ZIP with named files and a CSV, and you decide where they go.
A sheet-fed scanner or a phone app is all the hardware you need
All Docusplit needs is a PDF. Which device produces it depends on your mail volume.
Sheet-fed document scanners with an automatic feeder (ADF) suit regular use. The ScanSnap iX1600, for example, scans 40 pages per minute in color at 300 dpi (80 images in duplex), and the Brother ADS-4700W plays in the same league, so a day's stack becomes one multi-page PDF in a minute or two. Set them to 200 to 300 dpi, color or grayscale.
Flatbed scanners are fine for bound documents or fragile originals but slow for larger stacks. Multifunction printers with a scan function are an affordable option for occasional use; check that they can save multi-page PDFs.
Scanner apps like Adobe Scan, Microsoft Lens, or SwiftScan on your phone are enough for a few letters. Even lighting, a straight angle, and the app's auto-crop give the AI a clean page to read. Very faint or skewed scans are where recognition suffers, so a higher scan quality means fewer files you have to rename by hand.
A filing system for letters that the ZIP already follows
A filing cabinet works until it holds a few hundred letters and someone needs the one insurance renewal from two years ago. A digital archive replaces the cabinet with folders and filenames you can search, and it grows without a second cabinet.
The filename does most of the work. Docusplit names letters Type_Sender_Date.pdf with an ISO date, for example Lease_Northwind_Ltd_2025-01-15.pdf, so files sort chronologically and the type is readable without opening them. On top of that you pick one of three layouts for the ZIP: no folders, folders by month (2025-03/), or folders by sender. The naming pattern itself is fixed; you cannot define your own, and there are no folders by type or combined layouts.
If your existing archive is organized by category (Invoices, Contracts, Insurance), sort the ZIP by sender or keep it flat and move the files into your categories; the type at the front of each filename makes that a quick pass. Families usually get by with a handful of top-level folders, freelancers with one folder per client, property managers with one per building.
A simple routine keeps it current. Scan new mail once a week, upload it, and back up the archive once a month.
Letter types the AI reads
The document type comes from the page content, not from a template.
- Tax assessments and letters from government agencies
- Insurance policies, premium changes, and claim letters
- Payment reminders and collection letters
- Leases, employment contracts, and service agreements
- Invoices, named by invoice number and company
- Multi-page letters stay one file
Three stacks that arrive like this
The more mixed the inbox, the more the type in the filename helps.
Accounting firm, morning mail
Twenty letters for eight clients in one scan. After the upload each letter is its own file, and the sender in the filename shows which client file it belongs to.
Property manager's tenant files
Termination letters, repair requests, and utility invoices for several buildings. Sorted by sender, each tenant's and each utility's letters sit in a folder of their own.
Insurance folder at home
Policies, premium changes, and claim letters from three insurers. With the date in the name, the folder sorts itself chronologically.
FAQ about letter digitization
Do I need to scan each letter individually?▼
No. Put the whole stack in the feeder and create a single PDF. The AI detects where each letter begins and ends, so a 23-page scan comes back as separate files.
Which letter types does the AI recognize?▼
Letters, invoices, notices, reminders, insurance documents, and contracts, among others. In document mode the AI first checks if a page is an invoice; invoices are named InvoiceNumber_Company.pdf, everything else Type_Sender_Date.pdf.
Can I choose the folder structure?▼
Yes, at download time. You can keep the files flat, sort them by month (2025-03/, with Unknown/ for letters without a readable date), or by sender. Folders by document type are not an option; the type sits at the front of the filename instead.
How many letters can I process at once?▼
One PDF can be up to 50 MB, and there is no page limit per file. Your plan sets the page quota, 30 pages once for the free trial, then 100 pages a month on Starter, 300 on Pro, and 1,000 on Max.
Which scanners work for digitizing letters?▼
Anything that produces a PDF. For regular use, a sheet-fed scanner such as the ScanSnap iX1600 or Brother ADS-4700W at 200 to 300 dpi is the comfortable option. For a few letters, a scanner app like Adobe Scan or Microsoft Lens on your phone is enough.
What happens if the AI splits a letter wrong or cannot read something?▼
Handwriting, very faint scans, blank back pages, or attachments without a letterhead can be assigned incorrectly. If the AI cannot read type, sender, and date, the file is named Document_Page_7.pdf, it goes into the Unknown/ folder when sorting by month, and the CSV shows 'Not detected'. Check those files first and rename them by hand.
Related guides
Read on for how content-based filenames come about and how one upload splits and names a batch scan.
Split & Rename PDF
Upload one batch scan of mixed invoices, letters and receipts. The AI splits it into separate documents and renames each file by content. Sorted ZIP with CSV.
Learn moreAuto Rename PDF Files
The AI reads type, sender and date on each scanned page and names the file after it, e.g. Contract_Northwind-Ltd_2025-01-15.pdf. 30 pages free, no credit card.
Learn moreSources & References
- IRS, How long should I keep records? — Keep records for 3 years in the general case; 6 years for unreported income over 25%, 7 years for bad-debt or worthless-securities claims, employment tax records at least 4 years.
- GOV.UK, Company and accounting records — Records must be kept for 6 years from the end of the last company financial year they relate to, longer in the listed cases.
- Ricoh, ScanSnap iX1600 specifications — Scanning speed 40 ppm / 80 ipm for A4 documents in color at 300 dpi.
- OpenAI, Your data (API) — Data sent to the OpenAI API is not used to train models (since March 1, 2023); abuse-monitoring logs are retained for up to 30 days by default.
Every letter as its own named file
Upload one mail stack and check the result for type, sender, and date.
Try free now30 free pages with a free account, no credit card
Last updated: