Invoice in, bill out, people on the exceptions
I built this for my own client work, on accounts where the AP team was drowning in data entry. An invoice comes in, OCR reads it, the system matches it to the PO and receipt, the bill gets created, and a person only looks at what actually needs looking at. It runs as a SuiteApp inside NetSuite. If it would fit your environment, get in touch.
- Not days to process a batch
- Minutes
- No manual data entry
- OCR-driven
- 3-way match precision
- Line-level
The same routine, hundreds of times a month
Every invoice triggered the same steps: open the PDF, find the vendor, create the bill, type the line items, match to the PO, check the receipt, attach the document. Fifteen to thirty minutes each, a couple hundred a month. That is most of a person's week spent typing.
The work that mattered (vendor relationships, cash flow, the genuinely odd invoices) kept getting crowded out by the work a machine should handle. So I built the part that is mechanical and left the judgment to people. Cost per invoice stops climbing with volume, and the team gets their attention back.
The pieces that do the work
Intelligent vendor matching
Exact name, fuzzy matching, and tax ID lookup. Confidence scoring tracks how sure each match is.
3-way matching engine
Validates invoice against PO against receipt, at the line-item level, with configurable tolerances.
Automated bill creation
Creates the bill directly in NetSuite, attaches the source PDF, and keeps a full audit trail.
Exception handling
A smart queue for invoices that need a human. Detailed diagnostics, manual override when it is warranted.
From PDF to posted bill
Read → Match → PostInvoice arrives
OCR extracts the data from the PDF automatically. No typing.
Vendor matched
Intelligent matching finds the right vendor by name, tax ID, or fuzzy logic.
3-way match
Invoice is validated against the PO and the receipt. Variances get flagged.
Bill created
The bill is created with the PDF attached, ready for approval.
The situations that call for it
Cost per invoice is creeping up as volume grows
When the only way to keep pace is to hire more people to type, the math stops working. This puts machines on the data entry and people on the exceptions.
The AP team is doing data entry, not AP
A team hired for vendor relationships and cash flow should not be transcribing PDFs. This hands the repetitive part to the system.
A 3-way match needs to be enforced, not hoped for
When invoice, PO, and receipt have to line up before a bill posts, doing it by eye does not hold. The engine checks every line.
Double payments have slipped through before
Duplicate detection at capture time is the cheapest place to catch them, long before a payment run.
For the people deploying it into an account
- SuiteScript 2.1, modern NetSuite architecture
- RESTlet API that integrates with any OCR provider
- Map/Reduce processing that scales to high volume
- Custom records for a complete audit trail and status tracking
- Duplicate detection that prevents double payments
- Full source code, yours to keep
- Invoice data extraction via OCR
- Vendor identification and matching
- PO lookup and validation
- Receipt matching (3-way match)
- Variance tolerance checking
- Bill creation with line items
- PDF attachment to the bill record
- Exception flagging for review
- Processing status tracking
- Complete audit-trail logging
Works with Azure Form Recognizer, AWS Textract, Google Document AI, or your existing OCR.
If this is how you want AP handled
There is no checkout here. I built AP Copilot for my own client work. If it fits your environment, get in touch and we can talk about putting it in your account.