SPECIAL ADVERTISING SECTION
AI IN LOGISTICS
Artificial intelligence is transforming logistics and supply chain management, enabling companies to automate oncelaborious manual tasks, slash error rates, optimize shipping costs and boost both productivity and efficiency.
Logistics is a physical, real-world business where relationships between customers and the companies that move their freight are exceedingly important. However, the opportunity to strengthen and streamline operations with new technology is enormous.
AI“ will change every function [ of the logistics industry ], but it will change each function in different ways and at a different pace,” said Fabio Brocca, chief product officer at Norway-based Xeneta. The business has built the world’ s largest ocean and air freight rate benchmarking and market analytics platform.
“ The biggest step change— a jump in what’ s possible— is making agent tech systems and agent workflows possible and deploying and delivering those to customers in a way that works for them at enterprise-grade reliability and at scale,” said Nisarg Mehta, cofounder and chief technology officer at London-based Raft. The company is the world’ s largest AI-powered logistics platform, automating operations for freight forwarding.
“ That’ s what we’ re building now. How do we give people access to the power of autonomous agents while giving them control and visibility over what these agents are doing? It’ s finding the right balance,” he added.
Impact and influence
AI-driven automation is reshaping core logistics workflows, from order processing and documentation to load planning, scheduling, dispatching and routing shipments. By reducing human inputs, companies get faster, more reliable processing and fewer errors.
Automated workflows also scale far more easily than human-centric ones, enabling logistics providers to handle greater volumes at lower marginal cost while redeploying staff to higher-value exception management.
However, businesses seeking to incorporate AI into their workflows must be careful not to underestimate the change management process required at the organizational level to become comfortable with permitting AI to make decisions on its behalf.
“ That will take time because we are a logistically physical industry; we are moving stuff around and the likelihood that something is not going to go as planned is 100 %,” Brocca said.
“ For us, AI in regard to automation is more around enhancing the work of humans and enhancing procurement versus replacing the whole thing.”
Integrating the latest AI capabilities enables companies across the logistics industry to fundamentally alter how they do business.
“ You can rethink your capacity planning, your processes and how data quality works. You’ re still doing the same process but in a totally different way,” Mehta said.“ The models that people have historically used for some of these things and the mental models that people have, they adapt and shift.
“ That’ s, in many ways, what separates really successful companies from unsuccessful companies. The ones that know how to change their processes and culture and ways of working with what modern technology allows go really far. Those that don’ t, end up getting stuck in the past.”
Venture capital investment
Raft’ s most recent fundraising round was in 2023, when investors committed $ 30 million in Series B funding. In 2022, a subsidiary of private equity advisory firm Apax led an $ 80 million investment in Xeneta, valuing the company at $ 265 million.
Venture capital is returning to supply chain technologies, according to a November report from PitchBook. It estimates the value of investments into such companies at $ 3 billion in the third quarter, up 26 % over the preceding three months“ as trade and tariff concerns eased.”
“ AI applications continue to garner attention, and established teams in supply chain tech have landed sizable rounds to build new AI solutions,” the report noted.“ Investors are backing experienced teams for AI-native startups with big checks.” Mehta estimates the return on investments that its customers have achieved from using its products and services is anywhere from 400 % to 900 %, depending on the client.
“ What drives VC money is the ROI opportunity,” Brocca noted.“ The logistics industry’ s data is fragmented because it’ s a very physical industry. Data quality is a big issue. Contracts are not enforceable and fragile. These problems have been the same for many, many decades.
“ The fact that those problems are still there, they’ re being solved in other industries but not in logistics— that’ s what drives confidence from a VC perspective, that there will be someone who will come and solve those problems. With AI, some of these problems can probably be solved in a different way.”
Media and academia are increasingly discussing an AI bubble following a multifold increase in AI-related infrastructure investment pledges and a tech-led stock market frenzy. This has propelled the S & P 500 to valuations last seen during the dot-com mania of the late 1990s and early 2000s. S & P Global is the parent company of Journal of Commerce.
“ You won’ t see an AI company in logistics growing to a $ 100 million [ valuation ] in a few weeks because it’ s still an industry based on relationships where we’ re moving things, where people matter and people make decisions,” Brocca said.“ So, it’ s a longer journey than in other industries.”
An October paper by Dutch AI researcher Servaas Storm argues humanity has reached“ peak GenAI” in terms of large language models( LLMs), noting that returns are diminishing rapidly. Yet for Mehta, the question of whether LLMs can advance much further is unimportant.
“ The capacity that these models have given us [ means ] there’ s tens of years of interesting work that can be done, even if capacity does not change,” he said.“ There is so much value that could be added which we haven’ t even scratched the surface of yet.”
Brocca acknowledged that from a logistics industry perspective, there was still much to determine in terms of how to handle so-called hallucinations— instances when LLMs provide incorrect or entirely invented information— and other potential pitfalls.
“ These models are good for certain things, but they’ re not good for everything. So, for me it’ s about figuring out where are you going to use more of a rule-based approach or more of a standard machine learning AI model versus where you need interpretation and context,” Brocca said.
“ We’ re getting really good at insight creation. What’ s next is how we automate more so that you wake up in the morning and you’ ve got a new provider or a new rate, or you negotiate an existing rate. www. joc. com December 1, 2025 | Journal of Commerce 45