How to add document OCR to Priority ERP
Short answer
Run documents through an OCR and extraction pipeline, match results against Priority purchase orders, receipts and parts, then post drafts through the API or a load interface. Low-confidence fields go to a reviewer before anything is saved.
Priority users still spend hours on paper and PDF: supplier invoices, packing slips, certificates of conformance, customs documents and delivery proofs. Many arrive scanned, photographed from a phone or in the supplier's own layout, and in some regions in more than one language including right-to-left scripts.
The ERP side is well suited to this because purchase orders, goods receipts and supplier master data give you something to match against. A strong match reduces the work to confirming, not typing, and a weak match is a clear signal to send the document to a person.
This page is about the capture-to-posting design for Priority: where OCR output lands, how to match to documents, and how to store the source file so audits are painless.
What teams use it for
Supplier invoice capture with PO matching
Extract invoice number, date, totals and lines, match to the purchase order and goods receipt, and create a draft supplier invoice for the accountant.
Packing slip to goods receipt
Read the packing slip at the dock, match to open PO lines and prefill a receipt for the warehouse operator to confirm quantities.
Certificate of conformance and material certs
Attach CoCs and mill certs to the receipt or lot, extracting heat or lot number and part, so quality can find them later.
Customs and shipping documents
Pull reference numbers and values from import documents into the related purchase order or shipment record.
Delivery proofs against sales deliveries
Scan signed delivery notes and link them to the delivery document, closing the loop for disputes.
Multilingual invoices
Handle supplier documents in Hebrew, English and other languages with field extraction that does not assume left-to-right layout.
Priority ERP integration surface
- Priority OData REST API for reading purchase orders, receipts, suppliers and parts for matching, and for creating or updating draft documents.
- Priority interfaces for loading supplier invoice and receipt data into the relevant forms with standard validation.
- Document attachments on Priority forms, used to store the original image or PDF with the transaction.
- Supplier part cross-references and unit-of-measure conversions, which matter when invoice units differ from purchase units.
- Tax and currency fields on the supplier invoice form, which vary by country localization of Priority.
How it works
- 1
Define document intake
Choose intake channels: scanner folder, mailbox, phone upload. Keep the original file immutable and assign a document id.
- 2
Run OCR and layout analysis
Use an OCR engine suited to your languages and scan quality, then a layout-aware extraction step for headers, line tables and totals.
- 3
Score confidence
Assign a confidence to each field and an overall score. Totals that fail arithmetic checks drop the score regardless of the model's claim.
- 4
Match to Priority records
Resolve the supplier, find the PO by number or by supplier plus part and date, compare quantities and prices to the PO and receipt, and flag variances.
- 5
Review exceptions
Show the image next to extracted fields with the match candidates, and capture corrections to improve rules for that supplier.
- 6
Post as draft with the source attached
Create the draft through the API or interface, attach the file, and leave final posting and payment approval to normal Priority workflow.
Example flows
Three-way match on a supplier invoice
Trigger: A supplier invoice PDF arrives in the AP inbox.
- OCR extracts header and lines.
- The service finds the PO by number and reads the receipts.
- Quantities and prices are compared within tolerance.
- A clean match creates a draft invoice for one-click confirmation.
- A variance routes the invoice to the buyer with the difference shown.
Result: Accountants handle exceptions and approvals instead of retyping every line.
Dock scan of a packing slip with a CoC
Trigger: A receiving clerk scans two pages at the dock.
- The service separates the packing slip from the certificate.
- The slip is matched to open PO lines and prefilled as a receipt.
- The lot or heat number from the certificate is extracted.
- The clerk confirms quantities and the lot is attached to the receipt.
Result: Receiving is faster and the certificate is traceable from the lot record.
Keeping data on-prem
OCR is a good fit for local processing. Open-source OCR engines and open-weight vision-language models can run on a modest GPU server beside Priority, so invoices, certificates and drawings never leave your network. Keep the original files on your file share or in Priority attachments, run extraction in a service with no outbound internet, and expose only the review screen to users. For scans in mixed languages, test the engine on your real documents before committing, because recognition quality varies widely by script and scan quality.
Security checklist
- Store originals read-only with a hash, and never overwrite them after extraction.
- Segregate duties: the person who approves a payment should not be the one who corrects extracted bank details.
- Flag any change of supplier bank details extracted from an invoice for out-of-band verification.
- Restrict the API user to creating drafts and attachments, not posting or paying.
- Log extraction version, confidence and reviewer changes for each document.
Ways to get it done
Commercial AP capture product
Effort: Weeks of setup and per-document pricing.
Good for standard invoices at volume if it offers a supported Priority connection or can load through your interface.
Open-source OCR with in-house rules
Effort: A few weeks per document family, then ongoing tuning.
Works for a few high-volume suppliers with stable layouts and a capable developer.
Partner build with Netray
Effort: Pilot on one document type in a few weeks, extend after measuring accuracy.
Good for mixed document types, on-prem processing and PO matching tuned to your master data.
Pitfalls
- !Trusting a high OCR confidence on totals without an arithmetic check.
- !Matching only by PO number when suppliers often omit it, with no fallback to supplier, part and date.
- !Ignoring unit-of-measure differences between invoice and purchase order.
- !Skipping tax and currency handling for your country localization of Priority.
- !Allowing the pipeline to auto-update supplier bank details.
Frequently asked questions
Does Priority include OCR?
Some Priority offerings and partner add-ons include document capture, but I would not assume it for your version. Check with your reseller. A separate OCR service that posts drafts through the API or an interface is the common way to get exactly the matching rules you need.
How do I set confidence thresholds?
Run a few hundred real documents, record how often each field is right at different scores, and set thresholds by the cost of an error. Auto-accept only fields where measured accuracy is consistently high, and keep totals under arithmetic checks.
Can OCR handle Hebrew and mixed-language documents?
Often yes, but quality depends on the engine, scan resolution and layout. Test on your real invoices, especially mixed right-to-left and left-to-right tables and stamps, before promising results. Plan a review step for the first months.
Where should the scanned file live?
Keep an immutable original on a controlled file share or as a Priority attachment on the document, and record a hash. The posted transaction should link to it so an auditor can open the source from the invoice.
Can this run fully on-prem?
Yes. OCR and extraction models can run on local hardware, and the matching layer talks to Priority over your network. A single GPU server is often enough for moderate volumes, though sizing depends on document counts and model choice.
Want Document OCR working on your Priority ERP?
A free 30 minute assessment: we map the integration path for your version and setup, and where it should run.
Get a free assessment