Logo Lanfrica

Killmanga-AI/AIRauto

Domain:

socioeconomic

Record type:

software
Creator:
Kil
Host:
AI Invoice and Receipt Automation for South Africa # AIRAuto – AI Invoice & Receipt Automation for South Africa > A full-stack AI-powered invoice/receipt automation system tailored to the South African SME market. Process invoices, extract key fields using OCR + AI, validate against SARS VAT requirements, and export to local accounting platforms. --- ## Table of Contents - Overview - Market Context - Key Features - Tech Stack - Project Structure - Quick Start - Configuration - Development - API Endpoints - Database Schema - Testing - Known Issues - Deployment --- ## Overview AIRAuto is an MVP invoice processing system designed specifically for South African SMEs. It combines: - **OCR** (server-side Tesseract) for text extraction from PDFs and images - **AI** (OpenAI GPT with cheap-first strategy) for structured field extraction - **SARS Validation** (South African tax compliance rules) - **SA Accounting Integrations** (Sage, Xero, QuickBooks stubs + functional CSV export) - **Async Processing** (BullMQ/Redis queue for scalability) - **POPIA Compliance** (encryption, audit logging, data retention) ### Why Built for South Africa? SA invoice processing has unique requirements not addressed by global solutions: | Area | Requirement | Implementation | |------|-------------|-----------------| | **Currency** | South African Rand (ZAR, R) | All amounts default to ZAR; R symbol recognized | | **VAT** | 15% standard rate | Auto-calculated & validated; detects inclusive/exclusive | | **SARS Tax Invoice** | Specific mandatory fields | Label, VAT#, sequential #, date, description, qty, value + VAT | | **Invoice Thresholds** | Full > R5,000 / Abridged R50–R5,000 / None ≤ R50 | Auto-classified; enforces field requirements per type | | **Accounting Software** | Sage (dominant), Xero, QuickBooks | Integration stubs ready; CSV fully functional | | **POPIA** | Data protection (ZA privacy law) | Encrypted storage, audit logs, 5-year retention, consent tracking | | **Date Format** | DD/MM/YYYY common | Parser handles multiple …