This is the build list for a dispatch office in 2026. Five processes that can be automated without a new system, in the order we set them up at customers. Per process: which data it needs, what it delivers in operations, how long the build takes.
The order is not a ranking by savings. It follows the data: each process produces the data the next one needs.
Needs: a mailbox where orders arrive, 20 example files, access to the TMS (API or import). Delivers: orders sit checked in the system, the dispatcher only sees the review cases. At a wholesaler with 50 orders a day, from three and a half hours to under one, error rate below one percent. Build: two to four weeks.
Why first: everything after it needs clean orders in the TMS. And capture shows in the first week how good the customer master really is. (Usually worse than assumed. That is not an insult, it is the same everywhere.)
Needs: a portal login for the agent, the rules for which order gets confirmed how, and the match against the TMS from step 1. Delivers: confirmations and quantity changes run without a manual login. At a contract logistics provider, 900 to 5,300 orders per run, for eight weeks now. Build: three weeks.
Why second: without clean orders in the TMS the agent cannot check anything in the portal, only confirm. Confirming without checking is exactly what the human did before, under time pressure.
Needs: access to the telematics (API or screen), the tours with time windows from the TMS, dock addresses. Delivers: a notice 30 minutes before the dock to customer and dispatch, zero calls to the driver. At a forwarder with 40 trucks, about 45 minutes per dispatcher per day. Build: three weeks. Details in the arrival detection case study.
Why here: the time windows come from the orders. If those are wrong, the service reports the arrival at the wrong dock.
Needs: the invoice mailbox, the orders from the TMS, a threshold for deviations. Delivers: the invoice is matched to the order and checked against the agreement before it is booked. In the first weeks the check regularly finds line items that appear in no order. Build: two to three weeks. The build guide is in Processing freight invoices automatically.
Why only now: the check against the order is the value, and for that, orders and tours have to be right. From 2027 Germany's e-invoicing mandate joins in, and the service reads structured invoices instead of PDFs.
Needs: tours and arrival times from step 3, consignee addresses, a template per customer. Delivers: the consignee gets plate number, shipment and time before delivery, without anyone writing a mail. Build: one to two weeks once step 3 runs.
Why last: the advance notice is the simplest process and the one that depends most on the others. Without an arrival time it is an empty form.
The chatbot for customer inquiries. Not because it would never make sense, but because it needs the answers from the five processes above. A chatbot that does not know where the truck is refers to the hotline. First the five, then, if call volume justifies it, the bot.
Strategy& and BVL surveyed 115 logistics providers in November 2025: 76 percent automate processes, 10 percent scale. The difference rarely lies in the technology. It lies in whether the processes build on each other or stand side by side. The five above share orders, tours and addresses, and each writes a protocol the next one can read.
A caveat: the order applies to forwarders with regular customers and their own TMS. A pure charter operation without regular customers starts with the freight invoice, because that is where the money is.
Which of the five do you already have, which is missing? Tell me the number. I will show you on our example what the next one looks like after four weeks. Ask for a process check: 30 minutes, no sales pitch.
