Here’s my beef! We are asking the wrong question about AI in freight

Early in my career at C.H. Robinson, the largest transportatıon broker of US, brought me from Türkiye to the United States. A large part of my job involved calling drivers, checking whether a truck was available and giving directions over the phone.

That was how freight worked. Two phones, incomplete information and a human being holding the network together.

Since then, we have digitized freight through TMS platforms, marketplaces and visibility systems. We then learned to predict freight: rates, arrival times, capacity, routes and risk.

Now we are crossing another line.

AI is moving from predicting what will happen to acting on that prediction. AI agents can contact carriers, compare rates, run mini-bids, recommend or execute awards, book loads and manage routine exceptions within defined limits.

But here’s my beef:

We are still asking whether AI will replace people or which AI tool we should buy.

Those are not the most important questions.

Freight leaders should be asking:

Which decisions are we prepared to delegate?
What information and authority will an AI agent need?
When must a human intervene?
And when something goes seriously wrong, who remains accountable?

For shippers, AI could shift freight procurement from periodic tenders towards continuous sourcing and dynamic allocation.

For asset owners, it could optimize pricing, load acceptance, routing, vehicle utilization and empty kilometres.

For freight forwarders, the question may be more uncomfortable. If customers currently pay you to collect quotations, answer emails and transfer information between parties, AI will compete directly with that work.

If your real value comes from capacity, expertise, relationships, risk ownership and solving difficult exceptions, AI can make that value more scalable.

We are not yet at the point where AI runs the transportation department. Freight data remains fragmented. Systems are poorly connected. Decision rights and accountability are often unclear.

Even the best AI agent needs open roads.

Over the next 12 months, leaders should select one important freight workflow, connect the necessary systems, establish clear guardrails and measure the financial result.

Freight cost per shipment.

Gross margin per load.

Sourcing time.

Empty kilometres.

Manual interventions.

Service performance.

If the CFO cannot see the impact in the operating results, it is not transformation. It is another technology experiment.

The question is no longer whether AI will enter freight decisions. It already has.

The real question is which decisions we are prepared to give it, and what humans will do once we do.

Gökalp Kobe Ertuğrul
18 September 2026