How should a freight broker, fleet or 3PL choose a partner to build AI agents?
Pick one place where loads run late or shippers ring your desk, such as time lost at the dock, check calls or late-delivery news, and ask each candidate how they would change it inside the TMS and tracking you already run. A good partner tells you which work an agent takes on and which stays with your dispatchers, keeps rates, free time, detention and carrier checks in code rather than a model's guess, never lets an agent push a driver past legal hours, and measures each release against on-time loads and calls avoided rather than features shipped. Then ask who owns the code and data afterwards: it should be you. Unthinkable works this way, and it also builds logistics and mobility products, so it knows the software from both sides. Its fleet work includes an IoT route-planning system for Asian Paints that cut in-transit damage by 85% on a four-route pilot (as reported).
How can AI reduce truck detention?
By catching waits early. ATRI found drivers are detained at 39% of stops, 135 million hours in 2023. An agent watches dock arrival from geofences and check calls, runs the free-time clock from each contract, and warns the shipper while there is still time to load the truck. When a delay affects delivery, it asks customer service to move the appointment. Negotiating with facilities and deciding on lanes stay with your team.
How do AI agents help stop cargo theft and double brokering?
By checking identity at booking and again before pickup. Verisk CargoNet put 2025 cargo theft losses at about $725 million, up 60%. Code compares each carrier's authority, insurance and contact details with official and vetting-service records, then matches the driver, truck and tracking device to the contracted carrier. A mismatch stops the booking or holds the pickup, and a manager decides what happens. The agent never accuses anyone.
Can AI replace carrier check calls?
Mostly, where tracking data is available. The agent reads ELD or app tracking first and texts the driver only when data is stale or a stop runs long. It understands free-text replies, voice notes and photos, updates the load in your TMS and starts detention clocks from recorded times. Negotiation, driver welfare and anything unusual go to a person. No independent figure on check-call time exists, so we measure it in a pilot.
Can AI tell shippers about a delay before they call?
Yes. When tracking shows a load running late, the agent works out the new arrival from live position and appointment times by code, then messages the shipper and consignee with the reason and the new time. If the delay will miss an appointment, it asks customer service to rebook the dock. Arrival times come from tracking data, never a guess, and anything unusual goes to a person with the load history.
Will AI replace our dispatchers and customer service team?
No. Agents take the check calls, status updates, paperwork matching and detention clocks, so dispatchers and customer service spend their time on exceptions, carriers and customers who need a conversation. Rates above target, carrier blocks, pickup holds and waived charges always need a person, and every agent action is logged, so you can see which work the agents did and what went to your team.
How do you keep shipment and customer data secure?
By giving each agent only the access its role needs in your TMS, tracking and vetting services. Customer, rate and carrier data stays in your systems and is never used to train models, and every action is logged and replayable for an audit or a claim. Changes to a carrier's bank details always need a person, which also closes a common route for payment fraud.