Almost everything written about AI in road transport is about the vehicle. Autonomous braking, telematics, route optimisation that promises to shave miles off a manifest. That work matters. It also has very little to do with where the unpaid hours in a haulage business actually go, which is the office behind the cab.
We assess businesses that have almost nothing in common with each other, and the drag lands in the same handful of places every time. Industry makes surprisingly little difference to which ones. In a transport office it shows up as compliance, subcontractor invoicing and driver debriefs, none of which turns a wheel, and all of which has to happen before a load gets booked or an invoice goes out. So it slides to the end of the day, then to the end of the week.
Transport is also behind on using AI for any of it. The government’s AI adoption research puts adoption in the sector at around 10%, against 43% in information and communication and an average of 16% across firms with five or more staff.
Compliance paperwork
The assimilated drivers’ hours rules apply where the maximum weight of the vehicle or vehicle combination is more than 3.5 tonnes. Daily driving is capped at nine hours, extendable to 10 twice a week, with ceilings of 56 hours in a week and 90 across a fortnight. Somebody has to check tachograph downloads against all of that while keeping track of Driver CPC and licence renewals, and read infringement reports closely enough to tell a pattern from a bad Tuesday.
Miss the pattern and it stops being an admin problem. A Traffic Commissioner can call a public inquiry, suspend or revoke the operator’s licence and disqualify the transport manager named on it.
AI earns its keep here on the reading. It will turn an infringement report into something a transport manager can check in two minutes rather than 20, and it will flag the Driver CPC expiring next month while there is still time to do something about it. Deciding whether an infringement is minor or whether it escalates is a different job, and the named transport manager carries the legal exposure if that call goes wrong. Any tool implying otherwise is wrong, and a Traffic Commissioner reads the file closely enough to find that out.
Subcontractor invoicing
A mixed fleet burns real hours matching signed proof of delivery (POD) to job numbers and rate confirmations before anything can be invoiced, then chasing whoever has not sent theirs in. The cash is stuck behind the paperwork rather than behind the load.
Most of that matching follows rules a machine can hold: whether the POD exists, whether it matches the load, and which handful need a person to look at them. AI can run the check continuously instead of on a Friday afternoon, and send the chase without waiting for anyone to notice the gap.
It cannot settle a rate dispute, judge whether a subcontractor’s excuse for a late POD is fair or decide what to do about the driver who always holds up the invoice run. Those need somebody who knows the relationship and has to keep working with them next month.
Driver debriefs
Debriefs throw off a steady stream of notes: a near miss, a fault worth logging before the next inspection. Most of it gets written because somebody has to write it, and nobody goes back and reads a month of it looking for a pattern.
That is a good use of AI. Pull the recurring themes out of scattered notes and a transport manager can see that the same roundabout keeps producing near misses, or that one route always runs long against its planned time.
The debrief conversation itself is another matter. You are reading a driver’s tone as much as their account, and working out whether a “minor” incident is worth pulling at. That needs a person in the room.
What it takes to get there
All of it depends on groundwork most offices already half have: tachograph data landing somewhere consistent, and a transport management system export that is not three separate spreadsheets. Then pick one job to start with instead of a platform that promises to fix everything at once. Skip that and the AI becomes one more thing the transport manager has to check, on a list that is long enough already.
David Bevan, founder, HoursBack
