Menu
Home About Us Case StudiesLocationsPricingGlossaryOur Founder Blog Careers FAQ Contact
Solutions
View all Solutions → AI Customer Support AI Voice Receptionist AI Sales Automation Business Process Automation AI Document Processing CRM & ERP Solutions Digital Transformation Business Intelligence Cloud Solutions Managed AI Services
Industries
View all Industries → Construction & Trades Manufacturing Logistics & Transport Healthcare Professional Services Retail & Wholesale Hospitality Property & Real Estate Education Government & Nonprofits
Book a Free Discovery Call

End-to-End Invoice Processing Automation for a NZ Logistics Company

In Short

A New Zealand logistics company’s finance team was manually keying 800+ supplier invoices a month — a three-day cycle with frequent errors and growing compliance risk. Avtrix built an AI document-processing pipeline that reads, validates and posts invoices into Xero automatically. Manual effort fell by 95%, the end-to-end cycle dropped from three days to two hours, and every invoice now has a complete audit trail.

The Client and the Challenge

The client runs freight and transport operations with hundreds of suppliers — fuel, subcontracted carriers, tyres, servicing, tolls and port charges. Invoices arrived by email as PDFs, scans and occasional photos, in as many layouts as there were suppliers.

Each month the finance team opened, read, coded and keyed more than 800 of them into Xero, matched them against purchase orders and dockets, and chased the missing ones. The monthly cycle took three working days of the team’s time; duplicate payments and wrong GST coding were common; and the audit trail was a shared inbox. As the fleet grew, the problem was growing with it.

What We Built

Avtrix designed an automated document-processing pipeline that removes the keying entirely and leaves people to handle only the exceptions.

  • Capture. A dedicated inbox and supplier portal collect every invoice as it arrives.
  • Read. OCR combined with a document-AI model extracts supplier, invoice number, date, line items, GST and totals from any layout — no per-supplier templates.
  • Validate. A custom workflow engine checks each invoice against the purchase order, the delivery docket, the supplier master and duplicate-detection rules, and flags GST anomalies.
  • Post. Clean invoices are coded and posted to Xero via the API with the source document attached; exceptions (typically 5–10%) go to a person with the reason highlighted.
  • Audit. Every step — what was read, what was checked, who approved — is logged against the invoice.

Python Azure Xero API OCR / Document AI Custom workflow engine

How We Delivered It

01

Discovery

Mapped the current process end to end and sampled 300 real invoices across all supplier types to understand the variety.

02

Scope & fixed price

Agreed the validation rules, the exception categories and the success metric — cycle time and manual hours — at a fixed price.

03

Build & test on real invoices

The pipeline ran in parallel with the manual process for a full month; every disagreement was reviewed and the rules tightened.

04

Cut-over & support

Switched to automated posting once accuracy was proven, with the finance team trained on the exception queue and 60 days of support.

Results Delivered

1

95% reduction in manual effort

The team touches only the exceptions. Roughly 50 hours a month of keying and checking became a few hours of review.

2

Cycle time from 3 days to 2 hours

Invoices are in Xero, coded and matched, the same day they arrive — which also unlocked early-payment terms with key suppliers.

3

100% audit-trail compliance

Every invoice carries its source document, extracted data, validation results and approver, ready for the accountant and the auditor.

4

Fewer errors, fewer duplicates

Duplicate detection and GST checks now run on every invoice, not just the ones someone happened to notice.

What This Means for Your Business

If your team is re-keying documents — supplier invoices, delivery dockets, application forms, purchase orders — you are paying for software in wages. Industry benchmarks put a manually processed invoice at US$12–30 against US$2–5 automated; we set out the maths in the true cost of manual data entry. Document automation is usually a four-to-eight-week, fixed-price project, and it pays for itself within months for any business handling a few hundred documents a month.

Frequently Asked Questions

Client details are anonymised for confidentiality. Figures are as reported by the client after go-live. Delivered by Avtrix Software Solutions Limited, Taupō — serving businesses across New Zealand.

← All Case Studies Discuss Your Project
START YOUR PROJECT

Ready to Achieve Similar Results?

Book a free discovery call and let's explore what we can build for your business — and what results you can realistically expect.