The future of logistics: Careful AI implementation

The logistics industry is increasingly turning to AI to automate processes, making it vital for companies to implement the tech thoughtfully, incrementally and responsibly. AI is here to stay and each business needs to determine the best ways for AI to serve it, panelists agreed at the AirCargo Conference in Dallas this week. [caption id="attachment_57149237" align="aligncenter" width="1920"] Industry experts discuss the use of AI in logistics at the AirCargo Conference 2025 in Dallas. (Photo/Cargo Facts)[/caption] “There's actually no choice but to adopt AI at some level,” Glenn Palanacki, vice president of solutions for brokers and forwarders at technology provider Descartes Systems Group , said during the session. Palanacki compared AI adoption to internet use. While many businesses thirty years ago said they would never rely on the internet, it is nearly impossible today to find a business that is not online, he said. Uses for AI in logistics AI can mean many things. Most people use “traditional AI” every day when they, for instance, check the emails that have been sent to their spam folder, Ben Kozy, chief operating officer at AI-based logistics company Airspace, said. “Many [logistics providers] do quotes and routes [and filling out documents] through more traditional AI,” Kozy said. However, large language models (LLM) have grown in popularity in the last two years, he said. LLMs like ChatGPT, for example, can understand and generate communication, allowing them to analyze, translate and summarize text. In the logistics sector, LLMs are being used to filter and respond to emails, analyze voicemails and more, Kozy said. While these applications are very broad, they will likely become more industry-specific going forward, Michael King, CEO of software company Crown Data Systems, said in the session. Logistics companies will be best served by using AI to automate repetitive tasks, such as analyzing emails and documents, compiling data into reports, and inputting information into forms and management systems, panelists agreed. Generative AI can complete these tasks, but businesses can take it a step further by using agentic AI, Krishna Vattipalli, founder and CEO of final-mile management system Fleet Enable, said in the session. Agentic AI takes autonomous action based on knowledge obtained through machine learning and gen AI. “That's going to be the future of AI,” Vattipalli said. Keeping humans in the loop AI allows companies to increase output without changing headcount, Gero Keil, co-founder and CEO of freight email software company Levity, said. Panelists insisted that AI implementation, when done correctly, does not lead to layoffs for employees, but rather allows employees to refocus their energies to more creative roles. “We take humans out of the loop, not to cut costs, but in fact, to keep them focused on our customers,” Kozy said. “We can improve customer service by keeping our frontline operators focused on various load and other issues that are, of course, going to come up in logistics.” — Ben Kozy , chief operating officer at Airspace Still, employees must not become overly reliant on AI, Kozy said. It remains necessary for employees to think critically and be familiar with operations should AI go down. Implementation When looking to implement AI , companies should start small by automating tasks that employees generally hate doing, Kozy said. Panelists also advised companies to work with software companies that can customize the AI for them. Airspace is “built on AI and we’re not doing LLMs in-house,” Kozy said. Palanacki told the audience that “partnering with the right organization, right technology provider, I think it's just so critical.” Each company must then ensure employees at all levels of the company are on board with AI implementation, Kozy said. “You can have all the cool tech, fancy machine learning models in the world, [but it won’t help] if your people don't want to use it,” Kozy said. Once a company has an AI program and employees on board, it should implement the program slowly. "You should be tracking the amount of times the human makes a change to the automation," Kozy said. “And then, once X amount of time goes by and you've reached [an error] threshold that you're happy with, that's when you can cut the human out of the loop.” And a small number of errors does not negate the value of AI automation, Keil said. “If you let a human enter a million shipments into the [transport management system], there will be mistakes at scale,” Keil said. “And so, the question is always: Is it at least as good as a human with regards to accuracy and quality of data inputs and other tasks?” If the answer is “yes,” then “it's a no-brainer to make the switch, but of course, it needs to be tested,” he said. Issues with AI Despite the value AI implementation can bring to a logistics company, there are downsides and risks. Two major risks are organizational and technical, Jonathan Beaver, IT operations manager at logistics firm Team Worldwide, said. When implementing AI, companies must determine clear lines of responsibility for if something goes wrong and processes to locate and correct the error, Beaver said. Technical risks, according to panelists, include: AI drawing conclusions from biased and poor-quality data; AI “hallucinating,” or making up information; Plagiarism from copyrighted sources; Cyberattacks ; Data poisoning, in which malicious data is injected into a training set; and Jailbreaking, in which data injections can override built-in ethical constraints and content filters. AI can also be a concern for companies that aren’t implementing it. “If you don't want to use it, that's fine, but I'm going to tell you now a lot of your employees want to use it.” — Jonathan Bea ver , IT operations manager at logistics firm Team Worldwide If employees are using outside AI providers without a company’s sanction, they could be giving proprietary information to a third party that will use the information in any number of ways. When using an AI model, in-house or otherwise, users should be aware of the product’s privacy policy, the data-processing agreement, terms of use and similar documents, Beaver said. Register now for Cargo Facts' essential regional events for industry stakeholders: Cargo Facts Asia 2025 on March 24-26 at the W Shanghai in China; and Cargo Facts EMEA 2025 on May 6-8 at the Shangri-La Bosphorus in Turkey.

The future of logistics: Careful AI implementation

AirCargo Conference 2025

Skyway founder Peralta joins Cargo Facts Asia 2025

SmartLynx plans A321 freighter market exit

Related Posts

Cyber Aviation Global Forum to explore advancing cybersecurity risks in aviation

Fraunhofer looks to bring autonomous tech to air cargo

More data scientists needed for increasingly digital industry

SmartLynx plans A321 freighter market exit

Cargo Facts Free Newsletters

Cargo Facts Connect Podcast

Follow Us

Welcome Back!

Retrieve your password