How to add document OCR to MRPeasy
Short answer
Run incoming documents through OCR and field extraction, match the result to open MRPeasy purchase orders and items, and create receipts or attach files through the REST API only when confidence is high, sending the rest to a review screen.
Receiving in MRPeasy depends on paper. A packing slip arrives with the truck, a supplier invoice comes as a PDF, a certificate of conformance rides along with a batch of material. Somebody has to read it, find the purchase order, check quantities and key it in.
OCR can read the page, but reading is the easy part. The value comes from linking the extracted data to the right MRPeasy purchase order and lines, and knowing when not to trust it.
This page is about that linking, the thresholds and the review step, specific to MRPeasy's purchase and receiving data.
What teams use it for
Packing slip to receipt
Read the slip, match the purchase order number and propose received quantities per line for the goods receipt.
Supplier invoice check
Extract invoice number, totals and lines and compare against the received purchase order before it goes to accounting.
Certificate of conformance filing
Read lot, heat or batch numbers from a certificate and attach the file to the received lot or item record.
Delivery note photos from the dock
A phone photo of a slip is processed and the receiver confirms a pre-filled receipt.
Customer drawing and spec intake
Read the title block of a drawing PDF, capture part number and revision, and link the file to the item or order.
Handwritten count sheets
Digitise stock count sheets and propose adjustments for review, with the line image shown next to each number.
MRPeasy integration surface
- MRPeasy REST API for reading purchase orders, items and partners and for receiving or updating records; check which receipt and file endpoints your API version supports.
- Purchase order numbers, item codes and vendor records are the keys for matching; lot or batch tracking depends on how the account is configured.
- File attachment on records may be available through the API or only in the interface; confirm and plan a fallback of storing documents externally and linking by id.
- No OCR inside MRPeasy as a feature to depend on; extraction is done outside and results are posted in.
- Hosted SaaS, so documents are processed in your pipeline and only structured results or files are sent in.
How it works
- 1
Define document types
List packing slips, invoices, certificates and drawings, and the fields you need from each. Start with one type, usually packing slips.
- 2
Set up capture channels
A scanner folder, a mailbox and a phone upload page cover most shops. Normalise everything to PDF or image with consistent resolution.
- 3
Run OCR and extraction
Use a layout aware OCR engine or an open-weight vision model to return fields with confidence scores and the page position of each value.
- 4
Match to MRPeasy
Look up the purchase order by number and vendor, then match lines by item code or vendor part number, quantity and unit.
- 5
Apply thresholds
Set per field minimum confidence. Fully matched and high confidence documents become drafts; anything else goes to the review queue.
- 6
Review screen
Show the image with highlighted fields next to the proposed MRPeasy record so a person can accept or correct in seconds.
- 7
Post and file
Create the receipt or note through the API, store the original document with the record id, and log the extraction result.
Example flows
Dock receiving
Trigger: A receiver photographs a packing slip on a phone.
- Image is uploaded and the PO number is read.
- Open lines on that purchase order are fetched from MRPeasy.
- Quantities from the slip are proposed against each line.
- Receiver confirms and the receipt is posted.
Result: Receiving takes a confirmation instead of typing, and the slip image is stored.
Invoice against receipt
Trigger: A supplier invoice PDF lands in the accounts inbox.
- OCR extracts invoice number, total and lines.
- System finds the purchase order and compares received to invoiced quantities.
- Differences are flagged.
- Clean invoices go forward, flagged ones to a buyer.
Result: Accounting sees only the invoices that need attention.
Keeping data on-prem
Invoices and certificates carry pricing and customer information, so keep OCR on your side when that matters. Run the OCR engine and a local vision or language model on your own server or private cloud, store originals on your file share, and send only the structured result to MRPeasy's API. This keeps documents out of third-party model services while still using a hosted ERP.
Security checklist
- Encrypt stored documents and restrict the review screen to staff who already see that data in MRPeasy.
- Keep API credentials in a secrets store and use a dedicated integration user.
- Log extraction versions and reviewer edits so a wrong receipt can be traced.
- Do not let OCR output create financial postings without review.
- Retain originals for the period your auditors and customers require.
Ways to get it done
Cloud OCR service plus scripts
Effort: 2 to 3 weeks
Fast to start for clean typed documents; check data handling terms.
Local OCR and vision model
Effort: 4 to 8 weeks
Better for confidential documents and varied layouts.
Partner build such as Netray
Effort: 4 to 8 weeks
Suitable when you want the review screen, matching rules and private deployment handled together.
Pitfalls
- !Vendor part numbers differ from your item codes; build a cross reference or matching fails.
- !Units of measure differ between slip and purchase order, such as box versus each.
- !Handwriting and stamps over text cut confidence sharply; do not hide this behind averages.
- !Posting partial receipts from a slip that lists backorders can double count.
- !Skipping the review screen early on destroys trust after the first wrong receipt.
Frequently asked questions
Does MRPeasy include OCR?
You should not count on it. Extraction from scanned or PDF documents is normally done by an external service or model, with the resulting data sent to MRPeasy through its API or entered through a review screen.
How accurate is OCR on packing slips?
It depends heavily on document quality and layout. Clean printed slips from regular suppliers can be very good; phone photos and handwriting are weaker. Measure on a sample of your own documents and set thresholds from that.
Can it match lines to purchase orders automatically?
Yes where the purchase order number and item codes or vendor part numbers are readable. Quantity and unit checks catch many errors, and anything that fails a check should go to a person.
Where should documents be stored?
Keep originals in storage you control and link them by purchase order or receipt id. If your MRPeasy API supports attachments you can add them too, but check size limits and retention.
Is a vision model better than classic OCR?
For varied layouts and messy scans, often yes. For fixed form documents, classic OCR with templates can be cheaper and more predictable. Many pipelines use both and compare confidence.
Want Document OCR working on your MRPeasy?
A free 30 minute assessment: we map the integration path for your version and setup, and where it should run.
Get a free assessment