Raphaël Hervé, Manhattan Associates: ‘The adoption rate of AI and agents is impressively high’

AI gives logistics companies the opportunity to boost efficiency. Warehousing and transportation processes can be made smarter, faster, and more precise—for example, thanks to AI agents. It helps when software providers lay the groundwork for this, so that businesses can adapt the agents to their own processes more quickly. That’s exactly what Manhattan does.

“In my career in IT, I’ve never seen technology adopted as quickly as AI is now—and specifically AI agents.” This statement comes from Raphaël Hervé, who is responsible for Technical & Support Services at Manhattan Associates. “Ten years ago, we developed the Manhattan Active® platform, built on a microservices architecture and an API-first approach. Back then, we couldn’t have foreseen that AI would become this big.” Thanks to the cloud-native platform, AI data now remains within the platform, which offers security benefits, among other advantages.

Significant reduction in analysis time

According to Hervé, what makes AI so interesting is that it establishes a link between real-time visibility into the supply chain and the ability to respond to an event in real time. AI provides the information, and agents then act according to the parameters set by companies. Hervé: “Currently, within warehousing and transportation operations, it’s mainly key users who possess a great deal of knowledge, analyze processes, and then make decisions. The time between an action and a decision can easily run into hours, and that’s sometimes simply too long. Based on the first agents developed by customers in pilot projects, we see that we can easily reduce this time by thirty to forty percent.”

Two Types of AI Agents

Manhattan developed two types of agents: prebuilt, standard agents and custom agents. The first are ready-to-use and based on well-known logistics processes. The second give companies the opportunity to customize the agents and add logic. Hervé: “For example, the agents analyze existing workflows and rules. If event A occurs, then B follows, and so on. That knowledge is built into our system, and AI agents can make use of it. Users can also adapt the basic agents to their own specific processes and configure them so that, for example, a planner can ask questions about the status of a trip or schedule.”

Fine-Tuning Order-Picking Wave Processes

With agents, logistics companies can, for example, fine-tune their order-picking wave processes, says Hervé. “There are many opportunities for efficiency here, such as getting answers to questions like why a particular order wasn’t selected. If you get a real-time answer to such a question, you can fix it right away—sometimes even before anything happens, so you can prevent things from going wrong. That’s especially important for logistics operations serving B2C customers. You spend less time on corrections and can set priorities more effectively. Someone can prioritize a process because they can be certain that no problems will arise later.” There are also opportunities for TMS processes, such as agents that automatically verify the data on carriers“ invoices and compare it with the data in the TMS. ”This saves a user up to half an hour per invoice. It’s also easy to build an agent that can check, for example, whether a truck has a minimum load factor of eighty percent.”

CPG client is already developing its own agents

Although Hervé is extremely enthusiastic about the possibilities of AI agents, he is still surprised by the speed at which Manhattan customers are adopting them. “I expect the few pilots we have now to grow to several dozen in the coming months. One of our major CPG clients has already developed several agents in the first six weeks of the pilot. Analyses are being completed faster, orders are arriving on time more often—this client is already making truly enormous strides in improvement.”