AI Invoice Processing & Data Extraction

I need an end-to-end AI solution that can ingest incoming invoices, pull out key information with at least 95 % accuracy, rename each file intelligently, and store everything in an organised folder structure (or another storage layer you recommend). Input formats are varied—PDFs, JPEG/PNG photos taken on phones, and scanned documents—so the pipeline has to be format-agnostic. My initial thought is a machine-learning-driven extractor that combines OCR with layout analysis, but I’m happy to hear alternative strategies if they boost reliability or speed. Here is the data I must capture on every invoice: • Supplier name • ABN • Invoice number • PO number • Invoice date • Total amount and GST • Line-item stock details (quantity, description, unit price or total value) After extraction the system should: • Auto-validate obvious errors (e.g., missing ABN or mismatched totals) • Rename the original file using a consistent pattern like “Supplier_Invoice#_Date.pdf” • File it into a logical directory or feed an API/DB for downstream accounting I’ll supply a representative invoice set for training and final acceptance testing. The finished deliverable should include source code, model weights (if applicable), a brief setup guide, and a simple way for me to add new supplier layouts without heavy re-coding.