I will extract data from invoices and pdfs into google sheets or quickbooks
Automation engineer: n8n workflows, document extraction, inbox triage
About this Gig
Someone on your team retypes invoices, receipts, or forms into a spreadsheet or QuickBooks. I replace that with a pipeline: drop the PDF in a folder or forward the email, and the data lands where it belongs, structured and checked.
I built a document extraction product (Mousing) and run this exact pipeline for my own bookkeeping. Extraction uses a vision model, not template matching, so a new vendor layout does not break it.
Basic: one document type (invoices OR receipts OR one form) to a Google Sheet or CSV.
Standard: same, but into QuickBooks Online, Xero, or Airtable, with validation rules (totals add up, required fields present) and a review queue for anything uncertain.
Premium: up to three document types, an email inbox or Drive folder as the trigger, an exceptions sheet a human can fix in place, and 30 days of tuning on your real documents.
You get the workflow, the field mapping, a walkthrough video, and a handoff doc.
Accuracy is honest, not marketing: expect 90%+ on clean PDFs, lower on phone photos of crumpled receipts. Uncertain fields are flagged, never guessed silently.
Not a fit: energy or manufacturing companies (contractual), and medical records.
Technology:
Excel
•
Google Sheets
•
Python
•
Zapier
•
PowerShell
•
Other
My Portfolio
FAQ
Is my data safe?
Documents are processed inside your own accounts (your Drive, your inbox, your QuickBooks). I do not keep copies after delivery. If you need fully on-premises processing with no cloud model, ask before ordering; it is possible at a higher price.
Does it work on scanned images and phone photos?
Yes. Clean scans extract about as well as native PDFs. Phone photos of crumpled or angled receipts are harder; accuracy drops and more fields get flagged for review. Send 10 samples with your order and I will tell you what to expect before I start.
Can it handle vendor layouts it has never seen?
Yes, that is the point of using a vision model instead of templates. A new supplier's invoice format does not need a new rule. Genuinely odd documents (handwritten, multi-page tables, non-English) may need a rule or two; I flag that up front.
Does extraction cost extra?
Documents are read by Mousing (mousing.ai), my own extraction service, on an account you own. The free plan covers 50 pages a month, enough for Basic and for testing. Above that, Mousing Pro is 79 USD a month for 1,000 pages, billed by Mousing. Have your own extraction provider? I can use it instead

