LedgerLens: A Local Gemma Invoice Workspace My Friend Actually Used

Description and use

LedgerLens is a local-first invoice workspace that uses Gemma 3 via Ollama and Tesseract OCR to extract, review, and store invoice data on the user’s machine, eliminating cloud data leakage and per-page costs.

Problem

Freelancers and small businesses need a fast, private way to extract vendor, date, tax, totals, and line items from invoice PDFs without sending data to cloud AI services.

Outcome

The author’s friend Raj adopted LedgerLens and stopped using other services. The app extracts vendor KFC, invoice number, date, INR 1,092.90, category Meals & Dining, and line items from a Swiggy/KFC merge PDF, demonstrating accurate local extraction.

Workflow

• Drop PDF/PNG/JPEG/WEBP onto the app
• Local OCR with Tesseract extracts text
• Gemma 3 (4B) runs on Ollama to parse JSON‑schema fields and line items
• UI shows stages: accepted → OCR → Gemma → review
• User reviews side‑by‑side source and extracted fields
• Approve & Save writes expense row and vendor profile to SQLite
• Data is available in dashboard, reports, CSV export, and automation rules

Evidence

You drop a PDF or photo on your machine. Gemma 3 (4B) running in Ollama on 127.0.0.1 extracts vendor, dates, tax, totals, and line items.

Source

The first real document I threw at it was a Swiggy/KFC merge PDF. Gemma returned vendor KFC (Swiggy Limited), invoice number, date, INR 1,092.90, category Meals & Dining, and the line items (Epic Savers Chicken Bucket, packing charges).

Source

Ledger lives under %LOCALAPPDATA%\LedgerLens on Windows (or ~/.ledgerlens).

Source