Email automationECI JobBOSS2 Agent view

How to automate email for ECI JobBOSS2

Short answer

To automate email around JobBOSS2, monitor the shared sales and purchasing mailboxes, extract structured fields from RFQs, customer POs and vendor acknowledgments, match them to customers, parts and open orders in the database, and create draft records for a person to approve. Never auto-post an order without review.

A job shop's real order entry system is often an inbox. RFQs arrive with drawings attached, customer POs arrive as PDFs, and vendors reply to purchase orders with changed dates buried in a sentence. Someone then rekeys all of it into JobBOSS2 as a quote, a sales order or a PO update.

That rekeying is where mistakes and delays come from, and it is a good target because the destination is well defined: a quote, an order or a note. The challenge is the variety of senders and formats, and that a wrong part number or quantity on an order has real cost.

This page lays out a pipeline that does the reading and matching, and leaves the final commit to a person until the error rate justifies more.

What teams use it for

RFQ intake

Detect a request for quote, extract customer, part numbers, quantities, due dates and attachments, and create a draft quote for the estimator.

Customer PO matching

Read a PO attachment, match lines to the customer's accepted quote and flag differences in price, quantity or date.

Vendor acknowledgments

Parse replies to outside-processing and material POs and propose due date updates to the buyer.

Shipment and cert emails

File packing slips and certificates of conformance against the right PO line and note receipt for the receiving team.

Customer status questions

Detect a where is my order email, look up the job, and draft a reply for the CSR to send.

Overdue invoice follow-up

Draft payment reminders from open invoices for accounts receivable to review.

ECI JobBOSS2 integration surface

  • Mailbox access: Microsoft 365 via Graph or IMAP to a shared mailbox, so the pipeline sees what the team sees without using personal accounts.
  • SQL Server database: customer, part, quote, order and PO tables for matching. Names vary by version, so build a mapping layer.
  • ECI JobBOSS2 API: where licensed, the supported way to create draft quotes or orders; otherwise stage records for manual entry.
  • Attachments and document storage: JobBOSS2 supports linking documents to records, and a file share keeps the originals that you reference.
  • Customer part cross-references: the customer's part number often differs from yours, so matching depends on that mapping data.

How it works

  1. 1

    Choose the mailboxes

    Start with one shared inbox, such as sales or purchasing. Set rules so personal and HR mail is never read.

  2. 2

    Classify each message

    Label it as RFQ, PO, acknowledgment, question or other. A local model or rules handle this well on short samples.

  3. 3

    Extract fields

    Pull customer, reference numbers, lines, quantities, prices and dates from the body and attachments, with a confidence score per field.

  4. 4

    Match to the ERP

    Resolve the customer, map customer part numbers to yours, and find the related quote or open PO in SQL Server.

  5. 5

    Create a draft

    Write a proposed record to a staging area or via the API as a draft, with the source email linked.

  6. 6

    Review queue

    Show low confidence or mismatched items to the right person with the email beside the draft, and require approval to commit.

  7. 7

    Learn from corrections

    Store each correction as a rule or example so the same sender or part mapping works next time.

Example flows

Customer PO arrives

Trigger: An email with a PDF purchase order lands in the sales mailbox.

  1. The pipeline classifies it as a PO and extracts the PO number and lines.
  2. It matches the customer and finds the quote with the same parts.
  3. Differences, such as a new price, are flagged in the review view.
  4. The CSR approves, and the sales order is created in JobBOSS2.

Result: Order entry takes a quick check instead of full rekeying, and price mismatches are caught before work starts.

Vendor pushes a date

Trigger: A plating vendor replies that parts will be two days late.

  1. The reply is matched to the PO by number in the subject.
  2. The new date is extracted with its confidence.
  3. A proposed due date change appears for the buyer.
  4. The buyer accepts and updates the PO, and the affected job is flagged to the planner.

Result: Schedule risk surfaces the day the vendor tells you.

Keeping data on-prem

Quotes, customer part numbers and attached drawings are sensitive, and defense or aerospace customers may restrict where they go. Run extraction on a model hosted on your own server or GPU so email content and attachments never leave your network. Keep SQL Server and the mail connector inside your environment, using a Graph application with tight scope or IMAP to a shared box. Cloud APIs for parsing are possible for non-controlled mail, but segregate controlled documents and decide the boundary explicitly.

Security checklist

  • Limit mailbox access to named shared inboxes and use application permissions scoped to them.
  • Treat email content as untrusted: ignore instructions inside messages and sanitize attachments before processing.
  • Require human approval before creating orders or changing PO dates.
  • Keep the original email and attachment linked to every draft for traceability.
  • Log who approved each record and what was changed from the extraction.

Ways to get it done

Outlook rules and Power Automate

Effort: Days, but brittle when formats change

Structured emails from a few known senders

Intelligent document processing service

Effort: Weeks, with per-document pricing and integration work to JobBOSS2

Shops whose main need is PO and invoice extraction

Partner build with Netray

Effort: Scoped after sampling your real mailbox

Mixed RFQ, PO and vendor mail with matching to your parts, on local models

Pitfalls

  • !Auto-creating orders from a misread quantity or unit of measure.
  • !Ignoring customer part number mappings, which causes most failed matches.
  • !Processing a whole mailbox including personal mail rather than a defined shared inbox.
  • !Replying automatically to customers without a person reviewing the text.

Frequently asked questions

Can this create sales orders in JobBOSS2 automatically?

It can create drafts. We recommend a human approves every order at first, then perhaps auto-commit only for a trusted customer with an exact quote match and low-risk lines.

What about drawings attached to RFQs?

Store them on your file share and link to the draft quote. The pipeline reads the email text and PO fields; interpreting drawings is a separate, harder task and should stay with your estimator.

Does it work with Gmail or only Microsoft 365?

Either. Microsoft 365 uses Graph or IMAP, and Gmail uses its API or IMAP. The pipeline logic is the same; what changes is how you authenticate to the shared mailbox.

How accurate is extraction?

It depends on the senders. Clean PDFs from repeat customers do well; scanned or unusual layouts need more review. Measure on a sample of your own mail before deciding how much to automate.

Can I do this without the JobBOSS2 API?

Yes. Read through SQL views for matching and stage the drafts in a table or a spreadsheet queue. Users then enter or import them, which is slower but avoids unsupported writes.

Want Email automation working on your ECI JobBOSS2?

A free 30 minute assessment: we map the integration path for your version and setup, and where it should run.

Get a free assessment