Skip to content

Transport orders from email,
straight into your TMS.

Transport order entry automation reads booking requests from shippers in any format, extracts the collection, delivery, goods and reference details, checks them against your customer master data and creates draft orders in your transport management system. A planner approves clean orders in one click and fixes the rest.

What does order entry automation do?

A haulier or distributor receives orders in every format its customers like to use. One customer types the order into the body of an email. Another attaches a PDF from its ERP. A third sends a spreadsheet with forty lines at 4pm for collection the next morning. Someone in the office reads each one and types it into the transport management system (TMS).

Order entry automation takes over the reading and typing. It does five things:

  1. Classify each incoming message: new order, amendment, cancellation, query or something else.
  2. Extract the order fields from the email body, the PDF or each row of the spreadsheet.
  3. Validate the result against your customer master data and your own rules.
  4. Create a draft order in the TMS, or a file the TMS can import.
  5. Route anything that fails a check to a planner, with the reason shown next to the original document.

This page is about the transport-specific version. For general document reading, see document processing. For the wider logistics picture, see logistics.

Is transport order entry a job for an agent?

Fit check

Tick what is true of your transport order entry work.

Tick the statements that apply and the suggestion appears here: an AI agent, ordinary automation, or a person with software support.

Which fields are extracted, and how are they checked?

Extraction is the easy half. Modern AI models read varied layouts well. The value is in the checks, because a wrong delivery address costs a wasted trip, not a typo.

Order fields and the check each one needs
FieldWhere it usually comes fromCheck against
Customer accountSender address, letterhead, account numberCustomer master: the sender is a known contact for that account
Collection pointFree text or a site codeThe customer’s known collection sites; a new site goes to a person
Delivery address and EircodeFree text, often incompleteAddress lookup; Eircode format; the consignee list for that customer
Dates and windows“Tomorrow AM”, “by Friday”, a date columnYour cut-off times, delivery days for that area, bank holidays
GoodsPallets, cartons, weight, dimensionsWeight per pallet is plausible; total fits the service ordered
ReferencesPO number, booking reference, delivery noteNot a duplicate of an open order; matches the customer’s reference format
Special needsTail-lift, booking-in slot, temperature, dangerous goodsEach flag maps to a service code; dangerous goods always go to a person

Most of these checks are plain rules, not AI. The model reads the document. Your code decides whether the result is acceptable. Keep that split: a rule that says “weight per pallet over 1,200 kg needs review” is predictable, testable and easy to change.

What goes to a person?

A good setup sends a small, clear queue to the planners. Each item shows the original document, the extracted fields and the reason it stopped. Typical reasons:

  • A field the model is not confident about, such as a smudged weight on a scanned docket.
  • An address or collection site that is not in the customer’s master data.
  • A possible duplicate: the same order sent as a PDF and again in the email body, or resent with “URGENT” in the subject.
  • An amendment or cancellation that refers to an order the system cannot find.
  • Dangerous goods, temperature control or any service the customer has not used before.
  • A request outside your cut-off, or a delivery window you cannot meet.

Amendments need special care. “Change the delivery on PO 4471 to Thursday” must find the right order, check that it has not been planned or dispatched, and change it. If the order is already on a vehicle, the change goes to the planner, never straight into the TMS.

Set the review rule per customer at the start. For the first two weeks, a planner checks every draft. After that, clean drafts from customers with a good record can go through on one click, and the queue holds only exceptions.

How much time could transport order entry give back?

Estimate

Put your own numbers in.

The starting values are an illustrative example, not a benchmark or a neu figure. Use your own volume, the time each one takes today and the share you expect the system to handle without a person.

Hours back each month56
Value of that time€1,680
Over a year€20,160

Time saved is only part of the case: response times, missed work and errors matter too. The ROI guide covers how to set a baseline, and the cost guide what the build and running costs include.

What does this look like in numbers?

Take a hypothetical distributor with 60 orders a day from email. Today, each takes about 4 minutes to read and type, plus corrections later. That is 240 minutes, or 4 hours of office time, every working day.

After the first month, assume 70% of orders arrive as clean drafts that a planner approves in about 30 seconds each. The other 30% need a fix that takes about 2 minutes. That is 42 × 0.5 + 18 × 2 = 57 minutes a day. The saving is about 3 hours a day, before you count fewer wrong deliveries.

These figures are illustrative. Measure your own order times and error rate for two weeks before the build, as described in the ROI guide. The calculator on this page lets you try your own numbers.

Where should you start?

  1. Pick your two or three highest-volume email customers whose formats rarely change.
  2. Collect 50 to 100 real past orders from each, with the order that was finally entered in the TMS. That is your test set.
  3. Agree the target route into the TMS with your supplier: API, import file or EDI.
  4. Build extraction and checks, and measure accuracy on the test set before anyone uses it live.
  5. Run in shadow mode for a week: the system creates drafts, planners still type orders, and you compare.
  6. Switch to one-click approval, then add customers one at a time.

Most businesses start this as workflow automation. If you also want a single screen for planners across email, TMS and tracking, that becomes an operations platform.

Common questions

Can it handle orders in spreadsheets with many lines?

Yes. Spreadsheets are easier than PDFs because the columns are already structured. The system maps each customer’s column names to your fields once, then reads every row. The usual problems are merged cells, totals rows and hidden columns. The checks catch these and send the file to a person rather than creating forty wrong orders.

What happens when a customer changes their PDF layout?

An AI model usually reads the new layout without changes, because it reads the content, not fixed positions. Confidence scores may drop for a few days. Track accuracy per customer, so a sudden drop shows up as an alert instead of wrong orders. Template-based tools, by contrast, need the template rebuilt.

Our TMS has no API. Is this still possible?

Usually, yes. Most transport systems can import a structured file from a folder, and that is often enough. The automation writes the file, the TMS imports it on a schedule, and a check reads the import log for rejected lines. Screen automation is possible but fragile, so we use it only when nothing else works.

Should we ask customers to use a booking portal instead?

Where customers will use one, a portal gives clean data and no extraction risk. In practice many shippers keep emailing, especially smaller ones and those that order from their own ERP. Automating email intake lets you accept orders the way customers already send them, and a portal can still sit alongside it.

Sources

  1. GS1: EANCOM IFTMIN transport instruction message (based on UN/EDIFACT)
  2. GS1: EANCOM IFTSTA multimodal status report message

We last checked these sources on 19 November 2026. Rules, fees and prices change. Check the source before you act.

Bring us the task.

Tell us what happens today, which tools are involved and where it goes wrong. On a free 30-minute review, we’ll work out whether an agent, an automation or neither is the right fix, and what a sensible first release would be.

Start a project.

Two people on a sofa with laptops, planning a project beside a whiteboard of sticky notes

Book a call or send us a message. We read everything.

tash@neu.ie
Working with teams in Ireland and worldwide.

Get in touch

Free 30-minute AI review.