Case
A supplier invoice from an email → a draft in the books
The system reads the PDF or the Excel file, and the operator gets a prepared document instead of typing one.
≈ an approximate estimate of my actual process
Before
Invoices and goods receipts arrived from suppliers as PDF and Excel files.
Someone opened the file, created a document in the accounting system and moved the items and totals across by hand.
Altogether that took about an hour or two a day, and errors often surfaced only during reconciliation.
How I automated it
The system reads the mailbox and works out what kind of email it is: an invoice, a goods receipt, a payment, or not a document at all.
Known PDF and Excel formats are parsed by the rules of that particular supplier. From them a draft is created in the accounting system, with the counterparty, the items and the totals.
The operator checks the result and posts the document.
If the format is unfamiliar or the figures do not add up, the document goes for review instead of being posted automatically.
After
The draft appears shortly after the email. Instead of typing, there is a finished document to check.
What it gave
Nobody moves a document across line by line any more — they only check the result.
Documents reach the books on the day they arrive, so balances with suppliers are updated sooner.
What would be different in your company
The parsing is set up around your suppliers and your documents: for some companies that is invoices and goods receipts, for others price lists, completion certificates and customs papers.
What the system posts by itself and what it leaves for a person to check is your rule: you can start with an operator looking at every draft, and relax it later.
Show me a similar processTechnical details
Python, IMAP, PDF and XLSX parsing, email classification