How to automate email for Made2Manage
Short answer
Read the shared mailboxes where POs, RFQs and acknowledgments arrive, extract the fields with a model, match them to M2M customers, parts and orders, and put a draft in front of a person. Only reviewed drafts reach the ERP, through a path you control.
In a typical Made2Manage job shop, the order book arrives by email. Customer POs as PDFs, RFQs with drawings attached, supplier acknowledgments with revised dates, and invoices addressed to accounts payable. Somebody opens each message, reads it, and keys it into sales order entry, a quote, a PO release or an AP voucher.
That keying is slow, and errors in part number, quantity or ship date cost real money. Because the ERP has no inbound channel of its own, email automation is one of the highest-return additions you can make.
The principle here is draft, not post. The system reads and proposes; a person confirms. That is what makes it safe on an older ERP without a modern API.
What teams use it for
Customer PO to sales order draft
Extract customer, PO number, line parts, quantities, prices and requested dates, match parts to the item master, and prepare a sales order draft that flags price differences from the quote or last order.
RFQ intake
Read RFQ emails, pull part numbers and quantities, check for existing parts and prior quotes, and open a quote draft for estimating with the attachments filed.
Supplier acknowledgments
Detect confirmations and date changes from vendors, match them to PO releases, and propose date updates for the buyer.
Order change requests
Recognize 'please push line 2 by two weeks' or quantity changes, link them to the open order, and queue the request for the CSR.
Customer status questions
Detect 'where is my order' emails and draft a reply with the real promise date from the order for a person to send or approve.
Certificate and document requests
Route requests for certs of conformance or material certs to quality with the job and lot references already looked up.
Made2Manage integration surface
- Lookups: customer, part, vendor and open order tables are read to match extracted text. Common names include Slcdpm for customers, Inmast for parts, Somast and Soitem for orders, Pomast and Poitem for POs. Confirm names for your version.
- Entry points: sales order entry, quote entry, PO release and AP voucher screens. Without a documented API, drafts are rendered as a review page and keyed or imported by an authorised user.
- Import options: depending on version and license, there may be an import utility or interface for orders and items. Ask your vendor or reseller what is supported.
- Attachments: M2M has document linking features in some versions, otherwise store files in a share keyed by order or quote number.
- Mail access: Microsoft Graph or IMAP to the shared inbox, plus a processed-message ledger so nothing is handled twice.
How it works
- 1
Connect to the shared mailboxes
Use Graph or IMAP with a dedicated service account and mark each message as seen only in your own ledger, not the user's inbox state.
- 2
Classify the email
Decide whether it is a PO, RFQ, acknowledgment, invoice, change request or other. Unknown types go to a human folder.
- 3
Extract fields
Use text extraction plus OCR for PDFs and a model to produce structured JSON with a confidence for each field.
- 4
Match to master data
Resolve customer and part numbers against M2M, including customer part cross-references, and compare price and lead time to history.
- 5
Build a review screen
Show the original email beside the proposed transaction, highlight low-confidence fields and mismatches, and require a click to accept.
- 6
Hand off to M2M
Export an approved draft to the supported import, or present it as a pre-filled entry sheet for the CSR. Record the resulting M2M number back on the email.
Example flows
PO arrives as PDF
Trigger: A customer emails a PO to the orders mailbox.
- Mailbox reader picks it up and classifies it as a PO.
- Extraction yields PO number, three lines and ship dates.
- Two parts match; one customer part number is unknown and flagged.
- CSR resolves the unknown part, accepts the draft.
- Draft is imported or keyed, and the order number is saved on the email record.
Result: Entry time drops to a quick review and the exception is caught instead of buried.
Supplier pushes a date
Trigger: A vendor replies that line 3 will ship ten days later.
- Reply is matched to the PO via PO number in the subject or body.
- System proposes a new promise date on the release.
- It lists jobs that depend on the part.
- Buyer accepts and updates the release.
Result: The buyer sees downstream impact before changing the date.
Keeping data on-prem
Customer POs and RFQs often carry confidential pricing and drawings, so avoid sending them to a public model API. Run extraction with an open-weight model on a local GPU server and OCR with a local engine, reading mail through Graph or IMAP from inside your network. Only the mail connection leaves your site, and you can restrict that to your own tenant. The review queue and ledger can live in a small database beside the M2M server.
Security checklist
- Use a dedicated mailbox service account with least privilege and conditional access.
- Never auto-post transactions without human acceptance, particularly prices.
- Scan attachments for malware before processing and sandbox PDF parsing.
- Keep originals immutable and linked to every created transaction for audit.
- Watch for spoofed senders on payment-detail changes and route them for manual verification.
Ways to get it done
Generic email parsing service
Effort: Days to weeks, brittle on varied PDFs.
Works for consistent templates from a few customers.
Custom pipeline in-house
Effort: Four to eight weeks for a first workflow.
Best when you have a developer and a small number of document types.
Partner build such as Netray
Effort: A scoped pilot on one mailbox first.
Mixed document formats, on-prem models and review UI delivered together.
Pitfalls
- !Customer part numbers rarely match yours; build the cross-reference early.
- !Quantity units and price-per-hundred conventions cause silent mistakes.
- !Threads with forwarded history confuse extractors, so isolate the latest message.
- !Automatic acceptance on high confidence tempts teams; start with review on everything.
Frequently asked questions
Can Made2Manage read email on its own?
Not in the classic product as far as I know. You add a separate service that reads the mailbox and prepares drafts. Ask your vendor whether a newer module covers part of this.
Will it create sales orders automatically?
It can, but the safer pattern is to prepare a draft and have a CSR accept it. After weeks of clean data you might auto-accept narrow cases, such as repeat POs with no price change.
What about handwritten or scanned POs?
OCR handles clean scans reasonably; handwriting and faxes need lower confidence thresholds and more review. See the document OCR approach for details.
Does this work without an API?
Yes. Reads come from the database, and writes go through a supported import, a controlled stored procedure or a pre-filled entry sheet. Direct table inserts are the option to avoid.
Is a cloud model required?
No. Open-weight models can run on your own GPU for classification and extraction, which keeps PO pricing and drawings off third-party services.
Want Email automation working on your Made2Manage?
A free 30 minute assessment: we map the integration path for your version and setup, and where it should run.
Get a free assessment