How an international F&B franchise automated invoice processing across 14 supplier invoice formats with KIP.ai
KIP.ai, Your AI Native Finance Department, automates invoice processing from email for an international F&B franchise in Malaysia. KIP.ai splits multi-invoice PDFs, extracts and normalizes header and line-item fields, and maps the data to the ERP's data format. A maker-checker workflow lets the team review, edit and approve data before it is exported as ERP-ready CSV files. The same franchise also uses KIP.ai for multi-channel AR reconciliation. KIP.ai is built by Axrail, an AWS Premier Partner and a team of Applied AI experts.
At a glance
- Customer
- An international F&B franchise
- Industry
- Food & Beverage
- System
- Odoo ERP
- Use case
- Accounts payable (AP)
- KIP.ai workflows used
- Email parsing and invoice storage, Multi-invoice PDF splitting, AI field extraction and normalization, Data mapping to the ERP's data format, Maker-checker review, ERP-ready CSV export
An international F&B franchise operating in Malaysia receives supplier invoices from multiple vendors across its outlet network. Invoices arrive by email in a wide range of formats, creating significant manual workload and accuracy risks for the finance team.
Supplier invoices arriving in many formats had to be processed by hand, creating workload and accuracy risks for the finance team.
The finance team processed supplier invoices arriving by email in many different formats. Inconsistent date formats, variations in invoice ID formatting and complexities in line-item structures such as SST and specific billing categories meant invoices had to be checked and entered by hand. Blurry, low-resolution or stylized invoices made manual processing even less reliable.
KIP.ai reads invoices from email, extracts and normalizes header and line-item fields, and routes extracted data through a maker-checker review before ERP export.
Supplier invoices arrive by email at the franchise’s designated inbox. KIP.ai parses the email attachments and stores them automatically. Multi-invoice PDF files are split into individual invoice documents before processing.
KIP.ai extracts header fields from each invoice: supplier name, address, supplier code, payment terms, currency, total invoice amount, tax type, tax amount, tax rate, payment due date, buyer name, buyer address, invoice number and invoice date. It also extracts line-item fields: item code, description, unit price, unit of measure, quantity, total price and SST line items. Extracted data is normalized and mapped to the ERP’s data format.
Where the invoice format is inconsistent or fields cannot be reliably extracted, KIP.ai generates an exception report and flags the document. The finance team reviews all extracted data through a maker-checker workflow: one team member reviews and edits extracted data and a second approves before the data is exported as ERP-ready CSV files at both invoice level and line-item level.
Reduced manual data entry, structured handling of format variations and a review step before data reaches the ERP.
- Reduced manual data entry: Invoice fields are extracted and normalized automatically, removing the manual entry step for header and line-item data.
- Structured handling of format variations: KIP.ai processes invoices across a wide range of supplier formats, normalizing data before it reaches the review step.
- Controlled hand-off to the ERP: A maker-checker workflow gives the finance team the ability to review, edit and approve extracted data before it is exported as ERP-ready CSV files.
Key takeaways
- Invoices come in from emailKIP.ai reads supplier invoices from the designated email address, parses the attachments and stores them automatically.
- Fields are extracted and normalizedKIP.ai splits multi-invoice PDFs, extracts header fields (invoice number, date, amounts, supplier details, tax) and line-item fields (item code, description, quantity, unit price, SST) and normalizes the data to match the ERP's data format.
- The team reviews before exportA maker-checker workflow lets the team review, edit and approve extracted data. Every exception is flagged for the team to review and decide. Approved data is exported as ERP-ready CSV files.