How to automate email processing for QAD Adaptive ERP
Short answer
Read shared inboxes, classify each email, extract the structured data and create QAD transactions through QXtend or web services, with a review queue for anything uncertain. The highest payoff for QAD users is customer purchase orders to sales orders and supplier acknowledgements to PO schedule dates.
Even in QAD environments with EDI, a large share of business arrives as email: customer POs as PDF or Excel from smaller accounts, change requests in the body of a message, supplier order acknowledgements, price quotes, shipping notices and invoices. Someone retypes them into sales order maintenance, PO maintenance or voucher entry.
That rekeying costs time and introduces errors. In regulated or automotive supply chains it also loses the link between the original customer message and what was entered, which matters when a customer disputes a quantity or date.
Email automation here means a pipeline, not magic. Messages are classified, data is extracted and validated against QAD master data, valid items post through standard QAD logic and everything else lands in front of a person with the source email beside the proposed transaction.
What teams use it for
Customer PO to sales order
Read PDF or Excel POs, match customer, ship-to and item cross references, validate price against the customer price list and create a draft sales order for review.
Customer change requests
Detect emails asking to move a date or change quantity on an existing order, identify the order line and prepare a change for the CSR to confirm.
Supplier acknowledgement capture
Match acknowledgement emails to open PO lines and update promised dates on the purchase order after buyer approval.
Supplier invoice intake
Route invoice attachments to OCR and matching against PO and receipts, then prepare supplier invoice (voucher) entries with exceptions flagged.
RFQ responses
Parse supplier quote emails into price, lead time and MOQ and store them for comparison against the requisition or RFQ.
Shipping notices and certificates
Link advance ship notices and certificates of conformance from suppliers to the PO and the receiving record.
QAD Adaptive ERP integration surface
- Sales order, PO and voucher maintenance can be driven by QXtend documents or QAD web services, which run the standard business logic rather than writing tables.
- Reference data for matching: customer and ship-to (cm_mstr and address tables), item master (pt_mstr), customer item cross references, supplier master (vd_mstr), open PO lines (pod_det) and open sales order lines (sod_det), all available read-only through SQL-92.
- QAD price lists and customer-specific pricing rules for validating extracted prices before a draft order is posted. Use the QAD pricing service where available instead of reproducing rules.
- Attachments: QAD can hold document references against records, but implementations vary. Many teams store the source email in a document system and keep the link in a reference field.
- Mail side: Microsoft Graph or IMAP for shared mailboxes, plus a queue and database for message state, extracted fields and review decisions.
- Existing EDI in QAD (for example via an EDI hub) should remain the path for customers that send EDI. Email automation covers the rest.
How it works
- 1
Inventory the email streams
Sample a month of the sales, purchasing and AP mailboxes. Count by type, sender and format. Choose two high-volume streams first, usually customer POs and supplier acknowledgements.
- 2
Connect the mailbox safely
Use a service account with access only to shared mailboxes. Process, label and move messages rather than deleting. Keep the original email stored with an immutable ID.
- 3
Classify and extract
Classify the email type, then extract with a mix of rules and a language model. For attachments, use document OCR first. Capture fields with confidence scores.
- 4
Validate against QAD
Check customer, ship-to, item, unit of measure, price and dates against master data. Unknown items or price mismatches become exceptions, not guesses.
- 5
Create drafts through QAD logic
Submit via QXtend or services to create the order or update, in a state the CSR or buyer can review. Auto-post only for narrow, low-risk cases after trust is established.
- 6
Review queue with source side by side
Show the email, extracted values, QAD match and proposed transaction. Reviewer corrections feed back into matching rules and cross references.
- 7
Measure and widen
Track straight-through rate, correction rate and cycle time. Add streams only when corrections for the existing ones are low.
Example flows
Customer PO by email
Trigger: A PDF purchase order arrives in the customer service inbox.
- The pipeline classifies it as a PO and runs OCR and extraction.
- It matches customer, ship-to and customer part numbers to QAD items.
- Prices are checked against the QAD price list; one line is different.
- A draft sales order is created, with the price difference flagged.
- The CSR reviews, resolves the price and releases the order.
Result: Entry time drops to a review, and the source PO is linked to the order for later disputes.
Supplier acknowledgement
Trigger: A supplier replies to a PO with a new delivery date.
- The email is matched to the PO number in the subject or body.
- The model extracts line numbers and promised dates.
- The buyer sees old and new dates and approves.
- The date is updated through QAD logic and the MRP picture refreshes.
Result: Planners work from supplier commitments rather than stale dates.
Unknown sender with an invoice
Trigger: An invoice arrives from an address that is not on the supplier master.
- The pipeline cannot match the sender and does not post.
- It puts the item in the AP exception queue with the likely supplier candidates.
- AP confirms the supplier and, if needed, a master data change is requested.
Result: Fraud and mis-posting risks are reduced because uncertain items never auto-post.
Keeping data on-prem
Email contents include commercial terms and sometimes controlled technical data, so many QAD customers need the extraction models local. Run OCR and a language model in your own data center or private cloud on GPUs, connect to the mailbox through your own mail server or Graph with tenant restrictions, and keep the QAD calls on the internal network. Store emails and extracted data in your own database with encryption and retention. For customers who must keep technical data in-region, this keeps the whole pipeline inside the boundary.
Security checklist
- Use a dedicated service account with mailbox access limited to named shared mailboxes.
- Scan attachments for malware before parsing, and sandbox document conversion.
- Never auto-post transactions for unknown senders, new bank details or unusual amounts. Require review.
- Keep the original email and extraction result as evidence linked to the QAD document.
- Apply role-based access to the review queue so AP, sales and purchasing see only their streams.
Ways to get it done
QAD partner or AP automation vendor connector
Effort: 4 to 10 weeks including onboarding
A specific stream such as supplier invoices, where mature products exist and a connector to QAD is supported for your version.
In-house pipeline with QXtend
Effort: 2 to 4 months for two streams
You have developers and want control over matching rules and review screens.
Partner build with Netray
Effort: Scoped after discovery; first stream often in weeks
You want a private language model and OCR, a review UI and QAD posting, with your QAD team owning master data and business rules.
Pitfalls
- !Trying to auto-post from day one. Start with drafts and measure the correction rate.
- !Poor customer part number cross references, which cause most item matching failures.
- !Ignoring revisions and duplicates, so a customer's resend creates a second order.
- !Extracting prices without checking QAD pricing rules, which hides quote or contract differences.
- !Not keeping the source email linked to the transaction, which loses the audit trail.
Frequently asked questions
Can we create QAD sales orders from emailed customer POs automatically?
Yes, as drafts with validation first. The pipeline extracts the PO, matches customer and items against QAD, checks price and posts through QXtend or web services. Reviewers handle exceptions. Full auto-posting is only sensible for stable customers with low correction rates.
Does this replace EDI?
No. Keep EDI for trading partners that support it because it is structured and reliable. Email automation covers customers and suppliers that will not or cannot do EDI, which is often a long tail of small accounts.
How do we avoid duplicate orders?
Use the customer PO number, revision and sender as a key, and check existing sales orders before creating a draft. If the PO already exists, treat the message as a change request or a duplicate and route it accordingly.
What accuracy should we expect?
It depends on document quality and how clean your cross references are. Expect structured PDFs from regular customers to work well and scans or free-text change requests to need review. Measure correction rate per sender and expand automation only where it is low.
Can this run without sending emails to a public AI service?
Yes. OCR and extraction can run on open-weight models in your own infrastructure. This is the usual choice when emails contain pricing or controlled drawings. The trade-off is that you manage the GPU and model updates, or have a partner do so.
More for QAD Adaptive ERP
Want Email automation working on your QAD Adaptive 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