Every transport operator I have sat down with has the same tale about AI for transport and trucking. There is a magnetic board on the dispatch office wall, and exactly one person in the building who can read it properly.
It gets sold as autonomous convoys and route optimisation engines. For a twelve-truck operation running out of Hamilton, that is not where the money is. The money is in the four hours a day your dispatcher spends on the phone, the compliance paperwork someone does on a Sunday night, and the customers who ring three times before lunch asking where their freight is.
This post covers the three places AI actually earns its keep in a transport business: dispatch, compliance, and customer updates. What to automate first, what to leave alone, and how to start without turning the operation upside down for six months.
Key Takeaways
- AI for transport and trucking pays off fastest on three jobs: allocating loads, chasing compliance paperwork, and answering where-is-my-truck calls.
- Most operators lose more margin to admin drag and missed calls than to fuel price movements, and admin is the cheaper of the two to fix.
- A dispatcher fielding 60 to 100 calls a day is a single point of failure. Capture what they know first, automate the routine calls second.
- Voice AI answers every inbound call, so after-hours enquiries and subcontractor callbacks stop dying in a full voicemail box.
- Compliance is a data problem before it is a paperwork problem. Once logbook, RUC and CoF dates sit in one place, the reminders write themselves.
- Proactive delivery updates cut inbound status calls sharply, because customers ring when they are uncertain, not because they enjoy ringing.
- Start with one recurring task, not a platform migration. Roughly 95% of AI initiatives fail because they start tool-first instead of process-first.
- Your dormant customer list is usually worth more than the next ad campaign, and reactivating it costs nothing but the messages.
What AI for Transport and Trucking Actually Looks Like on a Tuesday
Forget the demos for a minute. Here is the real day.
Your dispatcher gets in at 5:30 am. Two drivers have texted overnight: one is sick, one had a warning light come on outside Bulls. There are eleven jobs booked for today, three of them time-critical, and one customer who has already left a voicemail asking to move a pickup to the afternoon. The magnetic board gets rearranged. Nothing gets written down about why.
By 9 am your dispatcher has taken about thirty calls. Maybe six of them needed a human. The rest were customers asking where a truck is, a subcontractor confirming a rate, and a driver checking a site access code that lives in someone’s head.
Meanwhile, the numbers you actually run the business on are spread across four or five places. Jobs in the transport management system. Invoices in Xero. Fuel in the card portal. Road user charges in a government login. Driver hours in paper logbooks or a phone app that half the team forgets to close off. Nobody has a single view of the week, including you.
The real constraint is not trucks or drivers
Most operators I speak to think the bottleneck is capacity. Not enough trucks, not enough drivers, not enough hours. Sometimes that is true. More often, the business is running at maybe 70% of what its existing assets could do, because the coordination layer sits inside two or three people’s heads and those people are saturated.
That is the trap. Your business works. It works because you are working it. If your dispatcher takes a week off, the whole operation drops a gear, and everyone in the building knows it.
This is not a motivation problem or a hiring problem. It is a systems problem, and it is the same shape in a 6-truck bulk cartage business as it is in a 40-unit line-haul operation. The symptoms differ. The cause is identical: no part of the business can think without a specific person present.
What AI is genuinely good at here
AI is good at the work that is repetitive, rule-shaped, and communication-heavy. Transport is full of it. Confirming bookings. Sending ETAs. Chasing proof of delivery. Watching a date and reminding someone before it lapses. Answering the same eight questions in a friendly voice at 7 pm on a Sunday.
AI is not good at deciding which customer you burn when two loads clash, and both are urgent. That is judgment, and it stays with you. The point of automating the first list is that you get more time and a clearer head for the second.
If you have already tried a few tools and found they helped for a fortnight and then became another tab, you are not alone. That pattern has a specific cause, and I have written about why AI projects stall in small businesses separately.
Dispatch: The First Place AI for Transport and Trucking Pays for Itself
Dispatch is where the compounding happens, because every minute recovered there flows into everything downstream.
Job intake without the phone tag
Most transport businesses take bookings three ways: email, phone, and the customer who texts your mobile directly because they have had your number for nine years. All three end up as a manual entry somewhere, usually retyped by hand.
An AI intake layer reads the email, pulls out the pickup, delivery, weight, timeframe and special requirements, and creates the job in your system. When something is missing, it emails the customer back and asks for it, in a sentence that sounds like a person wrote it. Your dispatcher opens their morning to jobs that are already complete rather than a mailbox of half-formed requests.
The same applies to new enquiries. Research from InsideSales via Harvard Business Review puts the first-responder advantage at around 78%: the business that responds first wins the majority of the time. In freight, where a shipper is often ringing three carriers, that number is not academic. If a new enquiry arrives at 4:50 pm and nobody replies until 8:30 am, you have already lost most of those.
Allocation and the running order
Here is where operators expect magic and get disappointed. AI will not replace your dispatcher’s read on which driver handles a fussy site and which one you never send to that particular yard in Onehunga.
What it will do is prepare the decision. Given yesterday’s completions, today’s bookings, driver availability, licence classes, vehicle types and known time windows, an AI layer can produce a proposed running order with the conflicts flagged. Your dispatcher then spends ten minutes adjusting instead of ninety minutes building.
That distinction matters. Assisted beats automated for anything with judgment in it. The system does 80%, your person steers the rest, and the quality of the decision goes up rather than down because they are not making it at 5:40am while a phone is ringing.
Driver communication
Drivers should not have to ring the office to find out what is next. A well-built setup pushes the job, the address, the site contact, the access notes and any special conditions straight to the driver’s phone, and takes back a confirmation, a delay note, or a photo without anyone typing.
When a driver flags a delay, the system does the next three things automatically: updates the job, notifies the affected customer, and tells dispatch which downstream jobs are now at risk. Today, all three of those happen only if your dispatcher remembers, and remembering is the part that fails under load.
Subcontractor coordination
If you sub out overflow work, you already know the admin cost. Ringing around to find who has a unit free, confirming the rate, chasing their paperwork, matching their invoice to your job.
An AI layer handles the outbound part: messages your subbie list with the job details, collects the responses, and presents you with who is available and at what rate. You decide. It confirms, sends the details, and files the paperwork against the job. What used to be forty minutes of phone tag becomes a decision you make from your phone in the yard.
Compliance Without the Sunday Night Paperwork Session
Compliance is where transport operators get quietly bled. Not usually by fines, though those hurt. By hours. By the Sunday evening someone spends reconciling logbooks, and the Tuesday morning someone loses to a Certificate of Fitness that expired because it lived in a diary nobody checked.
Work time and logbooks
New Zealand’s work time rules are not complicated in principle. A maximum of 13 hours of work time before a continuous 10-hour rest break, and a cumulative work day capped at 70 hours before a 24-hour break. The full detail sits with NZTA Waka Kotahi, and it is worth reading properly rather than relying on what someone told you in 2019.
The complication is not the rules. It is visibility. In most operations, the office finds out a driver is close to a limit when the driver mentions it, which is often too late to reroute a load without cost.
Once driver hours are flowing into one place, an AI layer can watch them continuously. It tells your dispatcher on Wednesday morning that two drivers will hit their cumulative limit by Friday afternoon, before Friday’s loads have been allocated. It flags logbook entries that look incomplete so they get fixed while the driver still remembers the day. It builds the audit trail as a by-product of normal operation rather than as a separate task nobody wants.
That is the shift worth understanding. Compliance stops being a job someone does and becomes something the system maintains while people do their actual work.
The dates that lapse
Every transport business runs on a calendar of expiries. Certificates of Fitness. Transport Service Licence. Driver licence classes and endorsements, including dangerous goods. HPMV permits. Insurance renewals. Scheduled servicing intervals. Road user charge purchases against hubodometer readings.
Right now that calendar probably lives in three places: a spreadsheet, someone’s Outlook reminders, and your memory. Put those dates into one place, and the reminder problem solves itself. Not a notification that fires the day something expires, but a sequence: a heads-up six weeks out with the booking options, a nudge at two weeks, and an escalation to you if nothing has moved at one week.
This is unglamorous, and it is exactly the sort of thing that keeps a truck earning instead of parked.
Health and safety and the incident trail
Under the Health and Safety at Work Act, the paperwork burden is real, and the enforcement is real. Toolbox talks, hazard registers, incident reports, vehicle pre-start checks.
Most operators do the safety work properly and record it badly. A driver reports a near miss over the phone, someone means to write it up, and it never makes it into the register. AI closes that gap by taking the report in whatever form it arrives- a voice note, a text, a photo- and turning it into a structured record with the right fields populated, then flagging anything that needs your attention. The driver does thirty seconds of talking instead of ten minutes of forms, so the reports actually get made.
Customer Updates and the “Where’s My Truck” Problem
Here is the number that surprises operators when they measure it: somewhere between a third and half of all inbound calls to a transport office are status enquiries. Where is my freight? Has it been picked up? What time is it landing?
Every one of those calls is a customer telling you they are uncertain. They are not ringing for the pleasure of it.
Get in front of the call
The fix is not a better phone system. It is removing the reason for the call.
Automated milestone updates do most of the work. A message when the job is booked and confirmed. A message when the driver is loaded and rolling, with an ETA. A message when there is a delay, sent before the customer notices. A message on delivery with the proof of delivery attached.
Four messages, none of which require a person. The customer who used to ring twice now rings zero times, and rates you higher on service, because certainty is most of what they wanted. Operators who put this in typically see status calls drop by half or more within a month, which hands your dispatcher back an hour or two a day.
Answer every call, including the ones after 5 pm
You will never get to zero inbound calls, and you should not want to. New business comes down that line.
The problem is that the calls arrive when everyone is busy or gone. A voice AI receptionist answers on the first ring, at any hour, and handles the routine end properly. It identifies the caller against your customer records, gives a job status, takes a new booking with all the required details, or routes a genuine problem to a human with the context already captured.
I have seen what this does in other industries where the phone is the front door. A dental practice I worked with, Dr Claire’s, was missing roughly 47% of inbound calls with two receptionists on the desk. After the voice AI went in, missed calls went to zero and booked appointments rose 44%. The mechanics are the same in transport. The call gets answered, the enquiry gets captured, and nothing depends on somebody being free at that exact moment.
I have covered the specifics of after-hours call answering in more detail if that is the piece you are feeling most.
Proof of delivery and invoicing
The gap between delivery and invoice is a cash flow problem that most transport businesses treat as an admin problem.
A driver delivers at 3 pm. The docket comes back to the office on Thursday when they next pass through. Someone matches it to the job on Friday. Invoicing runs Monday. That is five days of avoidable lag, multiplied across every job you run.
With the driver capturing the proof of delivery on their phone at the point of delivery, the job closes itself, the invoice generates against the agreed rate, and the customer gets both the confirmation and the invoice the same day. AI handles the exceptions: the delivery where the quantity differs from the booking, the one where the signature is missing, the one where a wait time charge applies. Those go to a person. Everything clean goes straight through.
The customers who stopped calling
One more thing worth checking. Every transport business has a list of accounts that used to book regularly and quietly stopped. Not because of a problem, usually. They got busy, changed a contact, or someone rang them at the right moment, and you did not.
That list is worth working, and nobody has time to work it. I ran a multi-touch reactivation for a finance broker called James against 319 dormant contacts his team had written off completely. It brought back $49,000. The database was already paid for. The messages cost nothing but the writing.
Transport has exactly the same asset sitting in the customer table of a system nobody opens.
How to Start Without Blowing Up the Operation
The MIT research doing the rounds put the failure rate of AI initiatives at around 95%. That number gets used to argue AI does not work. It argues something narrower: tool-first implementations do not work. Buy the platform, roll it out, hope adoption follows, watch it die.
The order that does work runs the other way.
Capture what is in the heads first
Before a single automation gets built, the system needs to know your business. Not generic freight knowledge. Yours. Which customers get priority when there is a clash. What each site’s access rules are. Which driver goes where. How you price a wait, a two-man job, a rural delivery. What your standard terms are.
This is the step everyone wants to skip, and it is the one that makes everything after it work. An AI with no context gives you generic answers you have to rewrite. An AI that has been briefed like a new operations manager on day one gives you answers you can use, and it never forgets the briefing.
It also solves a second problem for free. Once that knowledge is out of your dispatcher’s head and into a system, they can take a week off. That alone is worth the exercise for most operators.
Get the numbers into one place
Next, connect the data. Jobs, invoices, fuel, RUC, driver hours, vehicle costs. Not a migration, and not a new platform. Your existing tools stay exactly where they are. Small connections pull the numbers into one place so that the full picture exists somewhere other than your memory.
The immediate payoff is that you can ask a plain question and get a real answer. Which runs actually made money last month. Which customer’s freight costs more to move than you charge? Which truck is costing you in maintenance? Most operators find at least one loss-making lane they had been carrying for years.
Then automate one task
One. Not ten. Pick the recurring task with the worst effort-to-time ratio and cross it off permanently.
For most transport operators, that is either the status call problem or the compliance date calendar, because both are pure repetition with a clear trigger. Get one working, feel the difference, then do the next one. That sequencing is the whole method, and it is the same approach I lay out in the piece on AI adoption for SMEs.
Track it as a percentage. What proportion of your recurring tasks does the system now handle? Start at zero. You will feel the change at 20 to 30%. Six months in, 60 to 70% is realistic. Every task crossed off is capacity you never have to buy back.
The cost question comes up here, and the honest framing is comparison rather than a figure. Compare it to what a dispatch coordinator costs you fully loaded, to the value of one truck sitting idle for a week over a lapsed certificate, or to the margin on the jobs you are not winning because nobody answered the phone at 5:10 pm.
Conclusion
Transport is a low-margin, high-coordination business. That combination is exactly what makes it a good fit for technology, and exactly what makes tool-shopping a waste of time.
The wins are not exotic. Bookings that arrive complete. A running order that is drafted before your dispatcher walks in. Compliance dates that chase themselves. Customers who stop ringing because they already know. Invoices that go out the day of delivery instead of the following week.
None of that happens by buying software. It happens when the business finally has a brain of its own: a system powered by AI for transport and trucking that holds what your people know, sees your real numbers, and does the repetitive work so your team can do the work that needs a human. Get that in place, and the individual automations stop being projects and start being obvious.
You built a transport business. You should not be its dispatch system, its compliance officer and its after-hours answering service as well.
If you want to work out where the biggest hour-drain sits in your operation, book a 30-minute Discovery Call. No slide deck. We will map where your week actually goes and which one task is worth automating first. If the answer is that you do not need me yet, I will tell you that too.
Frequently Asked Questions
How can AI help a trucking company?
The three highest-value uses are dispatch support, compliance monitoring and customer communication. AI can read incoming bookings and create jobs, draft a running order for your dispatcher to adjust, watch driver hours and expiry dates, send automatic ETA and delivery updates, and answer inbound calls around the clock. The pattern is consistent: it takes the repetitive coordination work and leaves the judgment calls with your people.
Is AI for transport and trucking worth it for a small operator?
Yes, and often more so than for large fleets. A six- to twenty-truck operation usually runs its coordination through one or two saturated people with no redundancy. Automating status calls and compliance reminders alone recovers hours every week, which matters more when there is no depth on the bench. Big carriers have staff to absorb the admin. Small operators absorb it personally.
Will AI replace my dispatcher?
No, and any vendor promising that has not worked in transport. Dispatch involves relationship knowledge, site quirks and judgment about which customer to disappoint when two loads clash. What AI removes is the volume of routine calls and manual data entry that surrounds those decisions. The dispatcher stays, makes better calls with a clearer head, and stops being a single point of failure.
Can AI handle New Zealand transport compliance requirements?
It handles the tracking and the reminders, not the legal responsibility, which remains yours. AI can monitor work time against the 13-hour and 70-hour limits, flag incomplete logbook entries, watch Certificate of Fitness, licence endorsement and permit expiry dates, and build the audit trail automatically. Check current rules directly with NZTA Waka Kotahi, and treat the system as a monitor rather than an authority.
What should a transport business automate first?
Whichever recurring task has the highest repetition and the lowest judgment content. For most operators, that is customer status updates, because roughly a third to half of inbound calls are people asking where their freight is. Automated milestone messages remove the reason for those calls. Compliance date reminders run a close second because the failure cost, a truck off the road, is high.
Do I need to replace my transport management system?
No. Replacing a working system is the fastest way to a failed project and six months of pain. The approach that works connects to what you already run, pulls the data into one place, and adds the intelligence layer on top. Your team keeps using the tools they know. If your current system genuinely cannot export or connect at all, that is worth reviewing, but it is rarely the case.
How long before we see a result?
The first automation should be live and delivering within weeks, not quarters. Context capture and data connection typically take a few sessions. The first task automation follows straight after, and you notice it immediately because a job you used to do stops arriving. Meaningful change across the operation, in the range of 60 to 70% of recurring tasks handled, is a six-month arc.
About Octavius
Titus Mulquiney is the founder of Octavius AI, where he builds AI brains and AI workforces for founder-led businesses stuck running everything out of their own head. Twenty years in marketing, ex-Sony product manager, ex-GM Zeal NZ. Based in Auckland, working with operators across NZ, Australia, and the US. Connect on LinkedIn.