I was tired of manual bookkeeping, so I wrote an offline receipt parser with Ollama

Description and use

LedgerLocal is an offline receipt parser that uses a local Ollama model to extract dates, vendors, VAT, and categories from PDFs, eliminating manual bookkeeping and cloud privacy concerns.

Problem

Freelancers and small businesses manually parse PDFs for totals, VAT, and dates, incurring time and privacy risks.

Outcome

LedgerLocal processed 15 freelancer receipts in under 4 minutes, achieving 100% accuracy on tax rates and totals for digital invoices, and was well received as a potential €25 standalone app.

Workflow

• Upload PDF or text file
• pypdf extracts raw text
• Ollama local inference with phi4-mini:latest
• Model returns JSON with accounting fields
• Pydantic validates and stores in ledger.db
• User reviews/edit via UI or CLI
• Export to CSV or spreadsheet

Evidence

I looked at commercial receipt scanners. They all have the same two problems: They want $25 to $40 a month...

Source

The pipeline is simple: Text Extraction, Local Inference, Structured Output, Validation & Storage, UI

Source

It parsed 15 invoices (PDFs and text receipts) in under 4 minutes.

Source

I would never pay a monthly subscription for this, but if this was packaged as a standalone desktop app for €25, I'd buy it immediately...

Source