Agentic AI for freight that keeps movingagentic AI
Logistics runs on thousands of small, time-boxed decisions — tendering a load, chasing a check call, fixing a rejected EDI, clearing a border, telling a customer their freight is late. Most of it is still a dispatcher with eleven browser tabs. We deploy agents that do that work, with the safety-critical and trade-compliance calls kept firmly in human hands.
- Dispatch & load matching
- Freight & customs documentation
- Exception & detention handling
- Track-and-trace & customer comms
The margin is in the exceptions
Freight moves fine until something slips — and something always slips. The work that protects service and margin is the unglamorous chasing and reconciling between systems.
A dispatcher's day is interrupt-driven: a 204 tender comes in, a carrier rejects it, a rate has to be re-shopped, a driver runs out of hours, a receiver pushes the appointment, a 214 status never arrives, a customer emails asking where their truck is. None of it is hard judgment — it's retrieval and coordination across a TMS, a load board, three carrier portals, EDI, and email, under a clock.
Cross-border adds a second rulebook on top of the first. Every international move sits inside customs valuation and HTS classification under CBP or CBSA, advance manifest filings (ACE, ACI, ENS), sanctions and denied-party screening under OFAC and BIS, and a growing stack of forced-labor (UFLPA) and traceability obligations. A missed filing window or an unscreened counterparty can strand a container at the border for days.
That is exactly the shape of work agents are good at — high-volume, document-heavy, deadline-driven, bottlenecked on copying and checking rather than on the one decision a licensed person should own. We automate the legwork and keep dispatch authority, hours-of-service, hazmat routing, and customs sign-off with your people.
What the agents actually do
Concrete, bounded jobs across the quote-to-cash freight lifecycle — each one observable, reversible, and bounded by the rules you set.
Dispatch & load matching
Match open loads to capacity on lane, rate, equipment, and service history, auto-tender within contracted bounds, and escalate spot, new-carrier, hazmat, or oversize moves to a dispatcher.
Freight & customs documentation
Read and validate BOLs, commercial invoices, packing lists, and certificates of origin; draft HTS codes and entry data with the supporting basis, then route to a licensed broker for sign-off.
EDI exception handling
Catch rejected 204s and missing 990/214/210/997 acknowledgments, diagnose the mismatch, and repair or re-send — turning a silent EDI failure into a resolved tender instead of a lost load.
Track-and-trace & check calls
Pull positions from telematics and visibility feeds, run automated check calls, reconcile status against the TMS, and surface only the shipments that have actually fallen out of plan.
Delay & disruption response
Watch port, rail, and weather feeds for at-risk freight, compute reroute, expedite, or hold trade-offs, and brief the planner with options — the spend-or-move decision stays with a person.
Trade & sanctions screening
Screen shippers, consignees, and notify parties against OFAC, BIS, and denied-party lists on every cross-border move, halting any hit for compliance review with the matched record attached.
From tender to settlement
The agent compresses the chasing and reconciling; the human keeps dispatch authority and the regulated calls.
Tender
Inbound 204 parsed and matched to capacity on lane, rate, and equipment. Auto-accepted within contracted bounds; anything off-policy routes to a dispatcher.
Move
Agent runs check calls, reconciles 214 status against telematics, and flags only the shipments that have slipped — appointments, dwell, hours-of-service risk.
Clear
On cross-border legs, documents are validated and parties screened; the broker signs the customs entry before anything is filed with CBP or CBSA.
Settle
Agent reconciles the rate confirmation against the invoice and accessorials, drafts detention and OS&D claims, and queues clean settlements for approval.
Agents stay off safety-critical control loops
There is a category of decision in transportation that should never be automated, and we treat it as a hard architectural boundary, not a setting. Hours-of-service and driver fatigue, vehicle control and ADAS, dangerous-goods routing and segregation, and acceptance of brokered-load liability all stay with licensed, accountable people. Agents can prepare the facts and draft the move; they do not pull the trigger on anything that puts a person or a regulator's clock at risk.
Everything an agent does runs through a governed action layer: every tender, status update, and document touch is logged, off-policy moves halt for a named human approver, and every cross-border action is screened and traceable end to end. Compliance and safety are part of the architecture from the first call — not a checkbox bolted on after the demo.
- Hours-of-service & fatigue stay with the driver and dispatcher
- Hazmat routing and liability acceptance are human-only
- Off-policy tenders and customs filings require named sign-off
- Full action lineage for FMCSA, CBP, and audit
A bolt-on visibility dashboard vs. an Automatic.co agent
The difference between another screen that shows you the problem and an agent that works it.
| A visibility dashboard | An Automatic.co agent | |
|---|---|---|
| Exceptions | Lights up a tile for you to chase | Diagnoses, drafts the fix, and chases it |
| EDI failures | Surfaces a rejected 204 | Repairs and re-tenders within bounds |
| Dispatch | You re-key the move | Auto-tenders in-policy, escalates the rest |
| Customs | Hands off to a broker portal | Validates docs, screens parties, routes to sign-off |
| Safety calls | Out of scope | Deliberately out of scope — kept with people |
Frequently asked questions
Will an agent ever dispatch a driver or accept a load on its own?
Only inside the limits you set. Agents can auto-tender to a preferred carrier under a contracted rate and lane, but anything outside those bounds — a spot rate above threshold, a hazmat or oversize load, a new carrier, a same-day reroute — pauses for a human dispatcher. We never put agents on safety-critical control loops: hours-of-service decisions, vehicle control, brokered-load liability acceptance, and dangerous-goods routing stay with licensed people, and the agent's role is to assemble the facts and draft the move.
How do agents stay compliant with customs and cross-border rules?
Documentation agents validate commercial invoices, packing lists, bills of lading, and certificates of origin for consistency and completeness, and draft entry data and HTS classifications with the supporting basis attached. But a licensed customs broker signs anything that touches a CBP or CBSA filing, and counterparties are screened against OFAC, BIS, and denied-party lists before a shipment clears. Misclassification and a missed ACE/ACI filing are your liability, so those steps are human-gated and fully logged for recordkeeping.
Do agents replace our TMS, ELD, or visibility platform?
No — they act on top of them. Agents read and write through your existing TMS, ELD, and rating engines via API and EDI (204/214/990/210/997), and pull positions from your telematics and visibility feeds like project44 or FourKites. The system of record stays the system of record; the agent just does the copying, chasing, and reconciling that people do between those screens today.
How do agents handle a disruption — a port backup or a breakdown?
They detect it early and prepare the response, not execute it blindly. An agent watching carrier feeds, port and rail status, and weather flags an at-risk shipment, calculates the cost and ETA trade-offs of reroute, expedite, or hold, and briefs the planner with options. For service-failure and detention situations it drafts the customer notification and the OS&D or claim packet. The decision to spend money or move freight stays with a person.
Bring your worst exception queue.
One working session to map a single freight workflow — dispatch, EDI, customs, or track-and-trace — and the supervised, audited agent that can run it.