To break through the confusion surrounding the application of artificial intelligence to supply chain planning, businesses must take into account current technology, support systems such as agentic AI, and the future development of AI, says John Lash, group vice president of strategy with e2open, a WiseTech Global company.

When it comes to the application of AI to supply chain planning, “there’s a lot of excitement, and a lot of hype, too,” says Lash. Potential users need to be able to distinguish one from the other.

The way to do that, Lash says, is to think of AI in three ways: the technology, accompanying systems, and the future.

On the technology side, there’s substantial interest in generative AI, based on large language models, which came on the scene in late 2022. But there’s also “traditional” AI, which has been around for some 20 years, in the form of machine learning and AI engines that run in the background.

The systems view comes into focus with the more recent introduction of agentic AI, which can receive a request, turn it into a “goal-oriented” query, then work with multiple subagents to accomplish the desired task. Above them sits another agent, performing the road of “orchestrator,” Lash says.

Tasks suitable for AI agents include changing suppliers, explaining shifts in revenue, and running multiple scenarios to determine the best path forward. “The agent puts that decision into action,” says Lash.

But not without some oversight from humans. With the proper guardrails installed, the agent might be allowed to take autonomous action where the impact on cost is relatively small. What sets agentic AI apart from previous iterations of AI, however, is that it isn’t deterministic — in the sense, say, of a robot performing the same actions every single time. “It’s not pre-canning what can be asked,” Lash says. “You work with the situation you find yourself in.”