A UK business with 400 supplier invoices per week moves from full manual AP processing to 94% touchless — month-end close accelerated from day 5 to day 1.
A UK professional services company received approximately 400 supplier invoices per week from over 120 active suppliers. Every invoice was processed manually — a team member reading the PDF, typing the data into Sage, matching against the PO in a separate system, chasing the approver, and logging the payment. The process was slow, error-prone, and consumed 28+ hours of AP staff time every week.
We built an end-to-end AI invoice processing pipeline. Supplier invoices arrive by email. AI reads each invoice using GPT-4V, extracts all fields, validates against the Sage PO database, applies approval routing rules, and posts clean invoices automatically. Only the 6% of invoices with genuine discrepancies reach a human — with the specific issue highlighted and a suggested resolution pre-filled.
Every piece designed to solve a specific part of the problem — integrated into one system that works end-to-end.
The AP team went from spending 28 hours per week on data entry to spending 3 hours reviewing exceptions. The remaining 25 hours were redirected to supplier relationship management, cash flow analysis, and strategic finance work.
With invoices processed same-day rather than batched, the month-end reconciliation that previously took 3 days was reduced to a half-day review on day 1 of the new month.
Unlike template-based OCR, the AI reads invoices in any format — regardless of supplier. 120 different supplier layouts processed without a single template created or maintained.
From the initial audit to live deployment — every stage designed to minimise risk and maximise speed to value.