AI in Logistics and Supply Chain: The Quiet Revolution

There's a reason you don't hear much about AI in logistics. It's not glamorous. Nobody's writing breathless articles about how AI is revolutionising pallet optimisation or route planning. But that's precisely why it's one of the most significant workforce impacts happening right now — it's quiet, it's steady, and it's affecting millions of workers who aren't getting the same media attention as white-collar knowledge workers.

i spent time looking at supply chain operations as part of various data projects, and what struck me was how much of the work is essentially decision-making under constraints. Which items go in which warehouse? What route should this lorry take? How much stock should we order for next month? When should we reorder? These are optimisation problems. And AI is very, very good at optimisation problems.

Amazon is the bellwether

You want to know where logistics is heading? Look at Amazon. Not because every company will become Amazon, but because Amazon has been the most aggressive adopter of AI and automation in logistics, and what they're doing now is what the rest of the industry will be doing in three to five years.

Amazon's fulfilment centres have progressively replaced manual processes with automated ones. Robotic picking and packing. AI-optimised warehouse layouts that change dynamically based on demand patterns. Route planning that factors in real-time traffic, weather, delivery windows, and driver efficiency. Demand forecasting that determines what inventory sits in which warehouse before anyone's even ordered it.

The result? Fewer humans per parcel processed. Not zero humans — the technology isn't there yet for full lights-out warehousing. But significantly fewer. And the humans who remain are doing different work. They're managing the robots, handling exceptions, dealing with non-standard items, and maintaining the systems.

Other logistics companies are following the same trajectory, just on a delayed timeline. DHL, Maersk, UPS — they're all investing heavily in AI across their operations. The smaller logistics firms will adopt these technologies as they become more accessible and affordable. It's not a question of whether, it's when.

What's being automated now

Demand forecasting. This was one of the first areas where AI made a significant impact. Traditional demand forecasting relied on historical data, seasonal patterns, and a fair amount of gut feeling from experienced planners. AI forecasting incorporates hundreds of variables — weather, social media trends, economic indicators, competitor pricing, local events — and produces forecasts that are consistently more accurate than human planners.

The demand planners aren't all gone. But there are fewer of them, and their role has shifted from creating forecasts to reviewing and adjusting AI-generated forecasts. The junior analyst who used to spend days building spreadsheet-based forecasts? That role is essentially gone at any company that's invested in modern forecasting tools.

Route optimisation. Every major logistics company now uses AI for route planning. The variables involved — delivery windows, vehicle capacity, driver hours, traffic patterns, fuel costs, road restrictions — create an optimisation problem that's far too complex for a human to solve optimally. AI doesn't just plan routes; it re-plans them in real time as conditions change.

The dispatchers and route planners who used to do this manually are being redeployed or made redundant. Some companies still have human oversight on route planning, but it's oversight, not creation. The AI plans the routes. The human checks for obvious issues and handles exceptions.

Warehouse management. AI determines optimal warehouse layouts, picking routes, storage locations, and workforce allocation. It decides which items should be stored near each other based on ordering patterns. It optimises the sequence of picks to minimise travel time. In more advanced facilities, robotic systems handle the physical movement of goods.

Inventory management. When to reorder, how much to order, where to hold safety stock, when to markdown slow-moving items. These decisions used to involve significant human judgement. AI makes them continuously, across thousands of SKUs, factoring in variables that no human could process simultaneously.

Freight and carrier management. Selecting carriers, negotiating rates, optimising container loading, managing shipping schedules. AI is increasingly handling the transactional elements of this, particularly for standard shipments. The freight broker who matched shippers with carriers through personal relationships and phone calls is being replaced by AI-powered freight platforms.

The middle management squeeze

Here's where it gets interesting for people who don't work on the warehouse floor. Logistics middle management is being compressed significantly, and it's not getting the attention it deserves.

The regional logistics manager who oversaw operations across multiple warehouses relied on a combination of experience, relationship management, and the ability to synthesise information from various sources into decisions. AI now provides a unified view of operations across all locations, with recommendations already made. The manager's role shrinks from decision-maker to decision-approver.

The supply chain planning manager who coordinated demand forecasts, inventory levels, and procurement across the business? AI-powered supply chain planning platforms do most of that coordination automatically. The manager is still there, but managing exceptions rather than managing the process.

The operations manager who balanced staffing levels against expected volumes? Workforce management AI handles the scheduling, accounting for historical patterns, predicted volume, individual worker capabilities, and regulatory requirements. The manager's job becomes handling the things the AI can't — personal issues, unexpected situations, quality problems.

i've seen this pattern across multiple industries but it's particularly stark in logistics because the decisions are so data-driven. When the underlying work is fundamentally about processing data and making optimisation decisions, AI can absorb a large portion of what middle managers actually did day to day.

Warehouse floor workers: it's complicated

The narrative that "robots are replacing warehouse workers" is both true and oversimplified. Robotic automation is eliminating certain warehouse tasks — particularly repetitive picking, packing, and sorting in large, modern facilities. But many warehouses aren't large or modern enough to justify full robotic systems. Older buildings with non-standard layouts, smaller operations, facilities handling non-standard goods — these are harder and more expensive to automate.

What's happening more broadly is that warehouse work is changing. Workers increasingly work alongside automated systems rather than being replaced by them. The job shifts from purely physical (pick this item, carry it there, pack it) to hybrid (manage this section of automated picking, handle the items the robot couldn't pick, troubleshoot when the system jams).

This creates a split. Large, modern warehouses in big logistics networks are automating aggressively and reducing headcount. Smaller, older, or more specialised facilities are changing more slowly. The overall trajectory is fewer warehouse jobs per unit of throughput, but the total volume of e-commerce and logistics is growing, which partially offsets the per-unit reduction.

The honest assessment: warehouse floor jobs are declining and will continue to decline. The rate depends on the specific operation, but the direction is clear. If you're a warehouse operative, the stability of your job depends heavily on where you work and what you handle.

The last-mile problem

Delivery drivers are in an interesting position. Last-mile delivery — getting the parcel from the local depot to your door — is the hardest part of the logistics chain to automate. Autonomous delivery vehicles exist but they're limited in capability. They work in controlled environments and specific geographies. Dense urban areas with stairs, locked buildings, narrow streets, and unpredictable obstacles are extraordinarily difficult for autonomous systems.

Delivery drivers are probably safer from automation than most logistics roles in the near to medium term. But their work is being AI-managed more intensively. The route is AI-optimised. The delivery sequence is AI-determined. The time window is AI-calculated. The driver's autonomy is shrinking even as their job persists. You're increasingly a human executing an AI's plan.

Long-haul trucking is different. Motorway driving is significantly easier to automate than urban delivery. Autonomous trucking on major routes is being piloted and is closer to commercial viability than most people in the industry think. The long-haul driver on a regular motorway route is more at risk than the delivery driver navigating a residential estate.

What's growing

Supply chain data analysts and scientists. The AI systems generate enormous amounts of data and someone needs to make sense of it at a strategic level. Understanding patterns, identifying opportunities, and connecting supply chain performance to business strategy.

Automation engineers and technicians. Someone has to install, maintain, calibrate, and repair the automated systems. This is skilled technical work and demand is growing.

Exception management specialists. The humans who handle everything AI can't. Non-standard shipments, unusual requirements, crisis management, supplier relationship issues. When the volcano erupts and air freight stops, AI doesn't have a plan. Humans do.

Sustainability and compliance specialists. Supply chain sustainability is increasingly regulated and complex. Carbon tracking, ethical sourcing, regulatory compliance across multiple jurisdictions. AI assists but the regulatory and ethical dimensions require human judgement.

What to do if you work in logistics

If you're in planning or forecasting: Learn the AI tools. Seriously. The planners who survive are the ones who can work with AI forecasting systems, understand their outputs, identify when they're wrong, and add the contextual knowledge that the AI lacks. If you're still building forecasts in spreadsheets, you're on borrowed time.

If you're in warehouse operations: Look at the automation trajectory of your specific facility. If your company is investing in robotics and automation, understand where that's heading. Consider moving towards automation management, maintenance, or operational roles that involve overseeing automated systems rather than doing the manual work yourself.

If you're in middle management: Your role is being compressed. The value you need to demonstrate is in handling exceptions, managing relationships, making judgement calls in ambiguous situations, and connecting operational decisions to business strategy. If your job is mainly reviewing reports and approving routine decisions, AI is coming for it.

If you're in transport: Short term, driving jobs are relatively secure, especially last-mile delivery. But watch the autonomous vehicle developments in your specific area. Long-haul motorway routes will automate first. Urban delivery last. Position accordingly.

Watch for the signs of restructuring in your organisation. Logistics companies tend to restructure operations region by region or function by function. If AI tools have been deployed in one area and your area is next, use that lead time wisely.

The one thing to do today: find out what AI and automation investments your company has made in the past year and what's planned for the next year. The investment pattern tells you where the headcount changes will happen.