The benefits of agentic artificial intelligence are real, but its use in logistics won’t be widespread for some time, says Jose Reyes, senior director, logistics and customer fulfillment analyst at Gartner.

Three things — the three “C’s” in Gartner parlance — are necessary if a company is to be prepared to launch an AI initiative, Reyes says: contextualization, connection and continuity.

“You have to get your data contextualized, making sure it's aligned specifically with the problem that AI needs to solve,” Reyes explains. “We talk about AI being connected, meaning your data needs to be connected across the supply chain from transportation, warehousing, providers and external providers. All that needs to be visible. AI can't act if it can't see the full picture. And then third, there needs to be a continuous effort to improve and monitor data because operations aren't static. They're dynamic.”

Low-quality data is the greatest hindrance to AI projects, he says. Pilots often seem to go well, but efforts to scale fail because companies’ data isn’t consistent across the enterprise.

Reyes says many processes in supply chain were designed for people making periodic decisions, but AI works best when processes are continuous, dynamic and adaptive. “If you're trying to apply AI to a static process designed for people that make periodic decisions, you will be unsuccessful.” 

Success rests upon redesigning workflows and re-engineering processes, he says. First, AI should be embedded in a workflow exactly where a decision is to be made. “You need to use Lean principles to find where the waste is, where manual decisions need to be automated and where people just do repetitive work that AI can do.”

Secondly, it isn’t necessary to transform every process. “Pick specific processes, make sure you apply a lens of optimization, eliminate those waste points, then re-engineer everything from end to end.”