InvoicePilot: a private AI invoice sorter for freelancers that runs offline with Gemma 4

Description and use

InvoicePilot lets freelancers process invoices offline, preserving privacy and eliminating subscriptions, with validation and human approval.

Problem

Freelancers must manually sift through emails, downloads, and photos of receipts to extract data into spreadsheets, while existing AI scanners risk violating NDAs and incur subscriptions.

Outcome

InvoicePilot enables freelancers to process invoices offline, preserving privacy, eliminating subscription costs, and providing a validated, human‑approved CSV export.

Workflow

• Drop PDF/PNG/JPG into the interface
• OCR or PDF text layer extracts raw text
• Local Gemma 4 E2B model parses vendor, dates, line items, tax, total
• Schema validation (Zod) checks output format
• Bounded repair attempts correct validation errors
• Deterministic math check verifies subtotal+tax-discount=total
• Human review approves or rejects record
• Approved records exported to CSV and visualized

Evidence

Every freelancer I know has the same quarterly ritual: dig through email attachments, download folders and photos of paper receipts, then retype everything into a spreadsheet.

Source

Everything runs on localhost. No document text, price or client name leaves the machine.

Source

Small on-device models can invent a zero tax line when a tax breakdown exists, confuse MM/DD with DD/MM dates, or return totals that don’t add up.

Source

A deterministic math check: the backend verifies subtotal + tax - discount = total itself, in integer minor units, and warns on any mismatch

Source