AI Dispatch for Flatbed Carriers

Flatbed dispatch is not van dispatch with different filters. Van carriers drown in cheap loads and the job is to sort a flood; flatbed carriers see fewer postings, each worth more, each carrying real cost the posted rate hides. A load that pays well on paper can lose money once you account for two hours of tarping, an oversize permit across three states, and the deadhead to reach a steel mill that only ships open-deck.

The dispatch problem, then, is about precision rather than volume. You need to find the handful of loads that fit your trailer and your authority, price them honestly including the work nobody itemizes, and hold your number through a negotiation where the spread between the broker's opening rate and a fair rate is often wide. AI is good at the first two and useful in the third. It is useless at the part that actually moves the freight, and any honest account of where it helps has to say so.

The flatbed load is a worse signal than the van load

When a van dispatcher reads a posting, the line tells most of the story: origin, destination, weight, rate. Equipment is a single category. The numbers on the screen are close to the numbers that hit the bank. Flatbed breaks that assumption in every direction.

"Flatbed" is not one thing. A 48-foot standard deck, a 53-foot step deck, a double drop, an RGN, a Conestoga, and a lowboy are different trucks with different decks, different height clearances, and different jobs. A posting tagged "flatbed" might need a step deck you do not run, or call for a load height that only clears on a drop deck. The board's category does not tell you that. Worse, the rate rarely prices the specialized work. Tarping is two hours of labor and a real safety task in wind or rain, and it is frequently assumed rather than stated. Oversize and overweight loads need permits per state, escorts on some lanes, and travel restrictions that shrink your driving day. None of that is a field on the board, yet all of it determines whether the load makes money.

The consequence is that a flatbed dispatcher does more interpretation per load than a van dispatcher does, on a smaller pool, where each wrong read costs more in absolute dollars because flatbed runs at a higher rate per mile than dry van. With deadhead eating somewhere between 15 and 30 percent of miles industry-wide, and open-deck shippers clustered around mills, yards, and quarries that are not evenly spread, the cost of repositioning to the next good load is a bigger share of the math than it is for van. The job is reading a noisy signal correctly, fast, before someone else books the load.

Surfacing the loads that actually fit

The first thing AI does well for flatbed is cut the pool down to loads you can legally and physically haul, across more than one board at once. Flatbed freight is thinner on any single source, so a carrier who only watches one board sees a fraction of what is moving. Pulling from several boards and then filtering hard on equipment subtype, deck dimensions, weight, and permit status turns a scattered search into a short list of real candidates.

The filtering only matters if it is specific. Excluding RGN-only loads when you run step decks, flagging postings whose stated dimensions imply oversize so you know a permit is coming, separating loads that clearly involve tarping from those that do not, keeping distinct profiles for each truck in a mixed fleet so a step deck and a standard flat get different recommendations. This is the kind of structured matching software is genuinely good at, and it is where the time savings are largest, because every load you rule out before calling is a broker conversation you did not waste on equipment you do not own. Numeo's Load Hub searches across boards and applies these equipment filters so the list a dispatcher opens is already qualified rather than raw.

What the software cannot do is verify the things the posting left out. Brokers fill in load details unevenly, so a clean-looking match can still hide a height problem or an unstated tarp requirement. The filter narrows the field and orders it; the dispatcher still confirms the specifics that decide whether the load is real. That division is the right one. Surfacing and ranking is mechanical work that scales. Confirming a tarp call or a load-height clearance is judgment, and it stays with the person.

Pricing the cost the posted rate hides

Ranking flatbed loads by posted rate per mile is misleading in a way it usually is not for van, because the posted rate and the take-home rate diverge more. The value of AI ranking here is that it can fold the hidden costs into the comparison before the dispatcher spends a call on the wrong load.

Three costs distort flatbed economics and belong in the ranking. Deadhead comes first: a load paying less per mile from your current position can beat a higher-rate load three hundred miles away once empty miles are priced in, and open-deck shippers are geographically lumpy enough that this happens often. Securement and tarping come second: tarping is real labor and real time, and a load that requires it is worth less than an identical untarped load at the same rate. Permits and oversize handling come third: a load needing permits across several states, possibly an escort, and reduced daily miles carries cost and risk that a flat rate-per-mile sort completely ignores. A reasonable way to think about the adjustment looks like this.

A flat industry truck-cost figure of roughly $2.26 per mile (ATRI's 2025 report, on 2024 data) is the floor under all of this, and flatbed typically operates above it on the revenue side, which is exactly why getting the cost side right matters: the margin is there to win or lose. AI that ranks on adjusted economics rather than headline rate points the dispatcher at the loads where the spread is genuinely favorable. It does not decide; it orders the options so the human spends attention where the money is.

Negotiating where the spread is wide

Flatbed negotiation has more room than van negotiation, and that room is the reason rate intelligence pays off here more than almost anywhere else in dispatch. Capacity is tighter on specialized equipment, the carrier pool for a given load is smaller, and brokers often post low and expect a counter. Broker margins on van freight ran around 13.5 percent in DAT's 2023 data, and the specialized segment tends to run higher, which means there is more give in the opening number on open-deck loads than the posting suggests.

Where AI helps is in arming the counter with data rather than feel. Knowing the market range on a lane, what the load moved for historically, and how current capacity compares lets a dispatcher push back with a specific number and a reason instead of a hunch. On a load posted well under the lane's market, that gap is the negotiation, and walking in knowing the gap exists is most of the battle. The same tools can handle first-pass outreach on the secondary loads so the dispatcher spends their own time on the high-value ones where the spread is widest and the relationship matters.

The limit is worth stating plainly. The hardest flatbed negotiations are relationship freight: the steel mill, the lumber yard, the equipment dealer who ships open-deck every week and works through a few brokers who know their requirements. Those are not won with a market chart. They are won by a dispatcher who has earned trust, knows the lane, and can commit a specific truck with confidence. AI is the right tool for the spot and transactional loads and a poor substitute for the relationship calls. The win is that handling the routine outreach frees the dispatcher to spend more time on the handful of brokers who consistently offer premium flatbed freight. Numeo's AI Hub ranks and runs first-pass negotiation under the dispatcher's control, which is the correct place to draw that line.

Where the human stays, permanently

There is a category of flatbed work AI will not touch, and pretending otherwise sells the wrong thing. Securement is physical and consequential: chains, straps, binders, edge protection, the judgment of how a specific load of coils or pipe or machinery wants to be tied down so it does not shift. Tarping is a person on a deck in the weather. Deciding whether a load is safe to move in current conditions, whether a securement plan holds for an irregular shape, whether to take an oversize load on a lane with marginal clearances, is judgment with cargo and liability on the line. Cargo theft ran around $725 million in CargoNet's 2025 figures, and the kind of attention that protects high-value open-deck freight is human attention.

So the honest split is the one that respects the equipment. AI surfaces qualified loads across boards, prices the real cost of securement, permits, and deadhead into how it ranks them, and supports negotiation on the high-value loads where the spread is wide. The dispatcher confirms the load is what the posting claims, makes the tarp and securement calls, holds the relationships that produce repeat premium freight, and decides what actually rolls. Flatbed rewards specialization, and the carriers who do best with AI are the ones who let it remove the searching and the math so they can spend their hours on the parts of flatbed that pay precisely because they are hard. The takeaway is narrow and durable: use AI to find and price the right open-deck load faster, and keep the human on the deck and on the phone where the freight is actually won.