Trucking has the lowest AI adoption of any industry. It also has the clearest use case.

Start with the money you already billed and never collected

The American Transportation Research Institute studied driver detention across 2023 data and found two things that sit uncomfortably together.

94.5% of fleets charge detention fees. Fewer than half of those invoices get paid.

Now put your own fleet through it. ATRI puts truckload driver detention at between 117 and 209 hours a year depending on sector; take 173 hours, the truckload figure. A 75-truck fleet running one driver per truck is carrying roughly 12,975 detention hours a year.

The only sourced detention rate card available comes from the Owner-Operator Independent Drivers Association, whose 2023 survey of 253 members puts billed rates at $53 an hour for owner-operators under their own authority, $45 leased-on, $37 for company drivers. Take the middle figure. That is $583,875 of theoretically billable detention on a 75-truck fleet.

Apply ATRI’s own collection reality: 94.5% invoiced, under half paid. Roughly $308,000 a year, billed or billable, that never lands.

Every dollar of that is lost to a clerical failure. An arrival time not captured. A departure not timestamped. An invoice not raised inside the customer’s dispute window. A claim raised without the evidence to survive a phone call. None of it is a driving failure.

Three hundred thousand dollars of work you already did. Establishing whether that is your number, against your own ELD and TMS records, starts at $15,000 and reports in six weeks.

“Time of arrival at either the shipper or consignee starts the clock on detention.”

— Danny Crooks, VP of Corporate Transportation, Averitt Express

The clock is a record. Whoever holds the better record wins the argument.

And the loss you never billed at all

The uncollected invoice is only half of it. ATRI’s 2026 operational cost report puts the direct cost of detention at $5,392 per truck per year, inflation-adjusted to 2025. On 75 trucks, that is $404,400 in fuel, wages, insurance and truck payments burned while parked.

Set that against the same report’s margin table. Truckload carriers ran a 0.4% operating margin in 2025. Flatbed and oversize ran negative. At four tenths of one percent, a carrier would need roughly a hundred million dollars of revenue to earn the operating profit that a 75-truck fleet is currently leaving at loading docks.

That is not a rounding error inside a thin-margin business. In a thin-margin business, that is the business.

“Detention is so common that many industry professionals have accepted it as inevitable without realizing the true extent of its costs.”

— Chad England, CEO, C.R. England

Eight years, one federal recommendation, still no number

Here is how thoroughly unmeasured this is.

In January 2018 the US Department of Transportation’s Inspector General published an audit of driver detention. It found that a fifteen-minute increase in average dwell time raises a carrier’s expected crash rate by 6.2%, and estimated detention was costing truckload drivers between $1.1 billion and $1.3 billion a year in earnings, or $1,281 to $1,534 per driver, on 2013 data.

Its central finding was the absence of data. In the report’s own words: “Accurate industrywide data on driver detention do not currently exist.” The recommendation was to build a plan to collect it.

FMCSA commissioned that study. The information collection was approved in February 2024, the work went to the Virginia Tech Transportation Institute, and completion was expected in July 2025.

As of today the study page is still recruiting carriers to take part.

Eight and a half years after the federal government identified the gap, there is still no industry-wide detention number. Meanwhile every carrier reading this already holds the answer in its own ELD and TMS records, timestamped, load by load, going back years. The data problem is not that it does not exist. It is that nobody has turned it into an invoice.

The industry is pointing AI at the part that works

Look at what carriers say they are buying.

Penske Truck Leasing, which sells AI fleet products, commissioned a survey of 255 transportation and logistics executives in April 2025 and found 70% reporting they had adopted AI solutions. Take that as the ceiling. The interesting part is the use cases:

Route optimization. Fleet planning. Predictive maintenance. The truck and the road. Not one of these surveys names documentation, and no driver coaching either — the industry’s own conception of AI in trucking is aimed squarely at the hours-of-service and the driving, which is the part that already works.

Seventy percent report adoption. Under six percent have bought anything. That distance is the market, and it is the same shape across every industry that runs the physical world.

The same Penske survey found 84% of transportation executives saying their industry lags behind others on AI. They are right, and they are looking in the wrong direction while being right.

The record is also where the fines live

Detention is the largest documentation cost in a carrier, and it is not the only one.

Federal penalties for HOS recordkeeping violations run to $1,584 for each day the violation continues, up to $15,846. That is a fine for the record, not for the driving. A carrier can run a clean fleet, on time, without an accident, and still be paying four figures a day because a log was wrong.

FMCSA’s own paperwork burden estimate for hours-of-service regulations, published in the Federal Register in September 2025, comes to 53.4 million hours a year across 4.45 million drivers and 813,844 carriers. That covers duty-status entry and supporting-document retention only. It does not count exception review, driver coaching memos, or the argument with a shipper about who was where when.

Nobody has measured those. We looked. Every figure in circulation on safety-manager time comes from a compliance software vendor with no method disclosed.

One more gap worth naming. ATRI’s 2025 survey of over 4,200 industry stakeholders ranks detention as the fifth most important issue for drivers. It does not appear in the carriers’ top ten at all. The people who absorb the cost and the people who could fix it are looking at different problems.

Where the first step goes

If the loss is in the record, then the first thing worth fixing is the drafting and the capture, not the routing.

That has two shapes, and both are measurable in weeks.

Dispatch and safety. First-pass hours-of-service exception summaries drafted from your ELD data, raw inspection reports turned into consistent driver coaching memos, FMCSA regulation questions answered during roadside-incident follow-up. This runs alongside Samsara, Motive or whatever your fleet already reports through. It is one department, measured, from $15,000.

The six weeks, concretely: a week pulling your own detention and HOS exception history and timing what safety currently spends on it, two weeks configuring agents against your live ELD feed and your own past memos, three weeks of the safety desk working real exceptions with it, and a written comparison at the end. Your ELD provider stays where it is.

Freight operations. The heavier version: an embedded pod rebuilds the detention workflow against your real TMS and ELD data, auto-flagging detention events against each customer’s contracted free time and generating a ready-to-send invoice packet with the timestamps already attached. That runs against live data, wired into McLeod, Trimble or whatever runs your loads, with a go or no-go gate before anything scales.

The full picture of what AI-native transportation looks like runs from dispatch to delivery. No carrier should start there. Start where the money is already leaking.

The number to go and find

Everything above is somebody else’s data. Here is yours, and it takes one query against systems you already pay for.

Pull last quarter. How many detention events did your fleet record. How many got invoiced. How many of those invoices got paid.

If the answers are close to ATRI’s, a 75-truck operation is leaving around $308,000 on the table every year, in money it already earned, for want of a timestamp and an invoice raised on time.