AI can take over 80% of a dispatcher’s work, but only 5% at the dock. Which jobs are the most at risk?

Digital work first, physical work last

In his keynote “AI in Last Mile”, Kumar showed how strongly AI affects different roles in last-mile operations. The estimates were:

• customer success: assistive, around 90%,
• dispatchers: assistive, around 80%,
• network planners: assistive, around 70%,
• finance auditors and facility managers: assistive, around 40%,
• drivers: assistive, around 20%,
• dock supervisors: assistive, around 5%,
• data analysts: work that can be largely autonomous.

“The world is divided into two parts: physical distribution and digital distribution,” Kumar said. Where work is digital, AI agents will have the biggest impact. Where work is physical, such as at the loading dock, the effect is much smaller.

He also referred to Gartner research showing that 40% of staff capacity is currently spent on low-value, repetitive work, 45% on medium-value work and only 15% on high-value work. Automating the first group, such as data entry, paperwork and clearance audits, would free people for decisions that really matter.

Shippers want control, and AI is getting more expensive

Kumar opened with several Gartner predictions. One is that half of shippers will directly orchestrate transport assets globally. That does not mean owning trucks, but controlling transport through technology. For carriers, the message is that if they do not invest, their customers will start asking why.

He also warned that the “price of intelligence” is set to rise by around 40%. Today’s AI subscriptions do not reflect the real infrastructure costs, and providers will eventually pass them on. That is why every AI investment should be linked to a measurable business outcome.

Multi-agent systems, in which one agent plans routes, another handles exceptions and a third contacts customers, are used in only about 3% of workflows today. Kumar expects that share to grow tenfold within around four years. “We are late, but we are not too late,” he said.

A four-step playbook

Kumar’s playbook for AI in the last mile is built on four pillars:

1. Find the sweet spot. Do not use AI agents for simple, predictable problems that a basic workflow can solve. At the same time, do not hand the most critical decisions to agents. Kumar compared that to letting a robot perform surgery unsupervised.
2. Build or buy carefully. Packaged agents connected through APIs are usually cheaper. Building your own stack means decisions on hosting, computing power and GDPR, including whether data stays inside the EU.
3. Keep humans in the loop. People should check medium-value work and spend most of their time on critical decisions.
4. Own your data. “You can outsource operations, but you will have to control the data,” Kumar said.

Negotiate data, not just rates

For companies working with third-party carriers, gig workers or delivery service partners, data control is becoming as important as price. Kumar advised shippers to negotiate not only rates per drop or per mile, but also data formats and technology roadmaps.

Without good data, AI simply makes mistakes faster. He gave examples familiar to any last-mile operator: an agent routing deliveries to a wrong postcode, to an address that is flooded for months, or calling a customer 100 times before 6 pm when that customer never answers before then.

Cost control is also about paying correctly. “I’m not asking to pay less, I’m asking to pay right,” Kumar said, pointing to models in which carriers are paid per job rather than per route.

The cost pressure behind it

The keynote came as last-mile costs keep rising. In an interview with trans.info at the same event, FarEye’s Suryansh Jalan said cost inflation in the European last mile is about 11%, while carriers raise prices by only 5%. A recent survey also showed that last-mile delivery costs are rising faster than customer expectations.

According to Kumar, by 2035 the market leaders in every industry will be the companies that start investing in AI and data now.

Last Mile Leaders is an invite-only event for decision-makers in last-mile delivery, hosted by FarEye. The 2026 edition took place in Amsterdam from 13 to 15 October.