Beyond automation: when AI becomes your supply chain colleague

 

Ask any logistics manager what fuels their sleepless nights and you'll get the same answer: too many decisions, too little time and the relentless pressure of getting it right. But what if AI could ease the burden; handling the routine decisions like carrier selection, route optimisation and exception management, and freeing humans to focus on strategy and relationships?

This isn't a distant future. In fact, Gartner predicts that by 2030, 50% of supply chain solutions will incorporate autonomous decision-making. This is a significant shift from executing tasks to pursuing outcomes.But we're not there yet. While 36% of shippers have moderate or basic AI capabilities in their transportation management systems, only 1% currently use advanced autonomous decision-making. But momentum is building, with 23% of organisations already scaling agentic AI systems and another 39% experimenting.

What makes agentic AI different

Traditional automation follows pre-programmed rules; say, if X happens, then do Y. Agentic AI is different. These autonomous systems plan and execute multiple workflow steps on their own. They're goal-oriented. They monitor situations. Make decisions. Take action within the boundaries you set for them.

In other words, automation executes: ‘Book this carrier at this rate’, whereas agentic AI pursues outcomes such as ‘Optimise freight costs while maintaining service levels’.

So, where are shippers looking to put Agentic AI to use? Spot buying, carrier vetting and real-time ETA monitoring and disruption management top the list of priorities. But once organisations prove that it works in these areas, it’s unlikely any part of the supply chain will remain untouched.

AI as a colleague: the new paradigm

The expression ‘AI as a tool’ used to be commonplace in the workplace, but it’s being replaced by ‘AI as a colleague’. This boost in confidence reflects the thinking: two-thirds of shippers and more than half of carriers see AI's primary role as automating repetitive tasks, thereby freeing people for higher-value work. This shift is already tangible. Agentic AI systems are becoming fully-fledged parts of the workforce.

As a result, companies are no longer asking whether AI can help. Instead, they’re increasingly asking: “Can AI do it and how quickly can it deliver?” 

But even as these systems become more autonomous, the “human-in-the-loop” approach remains the preferred one. That means treating agents like new colleagues, and not just software. In other words, like with any new hire, they need clear job descriptions, continuous feedback and ongoing evaluation to become effective, reliable workers and partners.

And roles are already evolving. Dispatchers and planners are shifting from handling every task manually to overseeing intelligent agents, still responsible for the decisions, but with AI handling the execution.

The infrastructure reality: data, networks and modularity

It’s no surprise that data quality remains the biggest obstacle to adoption. It has been our industry’s most talked-about topic for years. More than half of both shippers and carriers cite it as their primary barrier.

But quality data alone isn't enough if it stays siloed. Network connectivity is critical, as it amplifies AI’s potential: systems learn faster when connected across trading partners, drawing insights from shared real-time information rather than isolated datasets.

Modularity also matters. Companies must be able to integrate agentic AI into what they already have, not rebuild everything from scratch. This approach lets organisations adopt agentic capabilities incrementally, matching their pace to their resources and technical readiness.

Why governance is essential

The more decisions AI makes on its own, the more critical governance becomes. This means setting clear boundaries: what can your AI agents do and what's off limits? Those guardrails enable safe AI use that stays perfectly in line with your intentions.

The key is to establish those guardrails before you scale, not after things break down. You need to track how agents perform at each step of the workflow and not just examine the final results. This enables you to catch errors early and keep refining, giving you a level of visibility that becomes critical as you move beyond pilots. Working with market-validated platforms and a trusted network can help you keep your deployments on target.

The path to 2030

If 2025 was the year of AI experimentation, then getting to 50% Agentic AI adoption by 2030 means 2026 must be the year of AI acceleration. 

So what’s the roadmap to success? 

1. Assess data readiness
2. Pilot in sandboxed environments
3. Establish or adopt market-validated governance frameworks
4. Build for network connectivity 
5. Invest in human capability by training teams to collaborate with and oversee agents 

There’s no question that AI colleagues will be an integral part of tomorrow’s supply chain teams. The technology has already proven its reach: in the US, for example, research shows AI can already handle tasks representing 11.7% of the workforce. By the end of the decade, the global potential for efficiency gains and cost reduction will be substantially higher.

The measure of excellence won’t be automation levels alone, but also the business results that human-AI teams achieve together. The companies that build the right governance, infrastructure and, perhaps most importantly, the culture, to support that partnership, will shape the next era of supply chain leadership.