How AI in the Supply Chain Will Impact Delivery in the Next Year

While the ultimate promise of AI is seductive—herds of autonomous delivery vehicles loaded by robots roaming far and wide under an umbrella of drones clutching packages—as with most tech, the real world applications are more practical and limited, especially in the short term. What could ultimately happen and what will actually happen in the next year are two very different things.

However, as with other technologies, businesses that fail to get on board are already missing considerable benefits and will be at a competitive disadvantage sooner rather than later. Last mile delivery organizations that delay a full commitment to digitalization are making their eventual learning and deployment curve steeper and steeper. 

AI has already transformed the last mile in important ways, and the next year won’t simply be more of the same as adoption and use cases expand.

If 2022 was the year of AI in the future, and 2023 was the year of AI hype, the next 12 months appear to be the time when AI is making measurable differences in a significant number of delivery organizations. If a delivery organization isn’t yet digitized and ready to implement new technologies, they’re behind the game, and the gap between them and their leading competitors is likely to widen. 

Caveat: Not all AI is created equal. Just because a developer says its product is “AI-powered,” doesn’t mean that AI is being used to its full potential, or even that it’s doing useful work inside the platform. Like the word “organic” in food, “AI” can mean anything from the engineers used ChatGPT to do some research to having a significant amount of AI and ML processes baked in. “Hype AI” and “Real AI” both exist in the marketplace, and here’s how “Real AI” is making an impact right now. 

1. Route Optimization 

While route optimization software is not new, AI is expanding the ways in which it makes routing more efficient. Rather than choosing between static and dynamic routes, sophisticated AI allows route engineers to create highly optimized routes that include both static stops dictated by customer mandates (fixed delivery days, windows or crews) with less constricted stops that can be scheduled around those static deliveries.

While less dramatic than drones dropping deliveries on customers’ doorsteps, the results of real AI route optimization are impressive: Stops are assigned to the most optimal trucks and crews; fleet capacity is maximized, enabling more stops with the same equipment and crews, lowering labor and capital costs; miles driven are reduced by up 10%+, resulting in less fuel burned and less carbon emitted. Routing that just reshuffles the stops until it finds the shortest distance can’t do that. It takes real AI to analyze the incoming stream of data about traffic, crew efficiency, available equipment and customer mandates to come up with optimal solutions in minutes, not hours. The best software uses the speed of AI to allow manual modifications to routes without degrading overall efficiency. 

When evaluating last mile software that claims to use AI, dig deeper. Ask how many different factors (such as predicted service time, individual truck characteristics, history of crew, customer preferences, weather, traffic, cost to serve) are considered in the solution. You’re looking for solutions that consider all of those, and more, and—importantly—use machine learning (ML) understand and predict the impact of those factors on last mile delivery arrival times. 

2. Customer Experience

The operational improvements made by AI are important, but the ultimate goal is a better experience for customers: Getting the right products delivered how and when they want them. 

Operators are finding that AI provides tools to improve the customer experience as well. Self scheduling is one of the most important. By rapidly comparing fleet capacity and existing deliveries against a customer’s expected delivery dates, AI-enabled tools can help suggest delivery windows that optimize capacity and routing. The customer can choose one of the suggested windows or ask for more options. Once chosen, that stop can be routed and a confirmation sent to the customer. When customers choose their delivery window, they are more likely to be at home, dramatically reducing first attempt failures and giving them control over the experience. 

In addition to increasing customer satisfaction, providing optimized windows increases efficiency for operators. The ability of AI to provide accurate ETAs—up to 98% accurate once ML has some history to work with—is a huge help for customers who get narrower windows and on-time delivery. 

AI also has the potential to allow operators to speed up and streamline communications between customers and the delivery organization, further reducing not-at-homes and boosting customer satisfaction. 

3. Analytics

Having huge volumes of data—loads, stops, routes, traffic, weather, crew service times, costs—is of no use if it can’t be analyzed. AI and ML are increasingly used to recognize patterns that pinpoint friction points in last mile delivery. When operators understand where their processes are slowing down or failing, they’re able to make improvements to increase efficiency and customer satisfaction. The ability of AI to quickly sort and collate reams of data into actionable reports is shortening improvement cycles and enabling faster process changes. 

Using AI to automate the review process—sending review requests, reporting results and highlighting noteworthy (positive or negative) reviews to human managers for action—has the power to fuel further rapid improvements. 

Upstream Effects of AI in the Supply Chain

What happens before a truck leaves the warehouse has an enormous impact on the last mile, and AI will continue to infiltrate upstream processes in the next year.

• Demand forecasting: Suppliers and distributors benefit from AI’s ability to forecast demand based on seasonal and other factors. 
• Warehouse automation: AI can improve connectivity across multiple warehouses or distribution centers to determine optimum stock levels and suggest transfers. 

What Isn’t Happening, Yet

Like flying cars and jet packs, some of the promising benefits of AI in the supply chain seem always just over the horizon. 

• Autonomous delivery trucks: While trials are popping up everywhere, the large scale rollout of autonomous delivery trucks is not just around the corner. The problem of mixing autonomous vehicles with everyday traffic is harder to solve than many people in the tech community thought. Some high profile accidents involving self-driving vehicles have created doubts among licensing authorities, and some vehicle producers have voluntary curbed or ended live trials. Smaller delivery robots confined to sidewalks may arrive in numbers well before autonomous delivery trucks. 
• Drone deliveries: Actually, according to a McKinsey report, there were an estimated half-million drone deliveries worldwide in the first half of 2023. Well publicized trials by Amazon, UPS, Walmart and others have shown great promise in cutting both delivery times and costs. However, only 15% of those drone deliveries were in North America because the U.S. lacks a national regulatory framework would facilitate large scale rollout of drone deliveries. This is not just a government issue; studies have shown that U.S. consumers are hesitant to embrace drone deliveries. Concerns about safety and covert surveillance are still strong in the U.S., whereas consumers in European countries with approved regulatory schemes have been more accepting. 

When it comes to AI, the toothpaste is out of the tube and it’s not going back in. It has already put its mark on last mile delivery and its influence is going to grow, and grow rapidly, in the next year. Here’s what businesses need to do to ensure they aren’t left at the starting line: 

• Identify areas where AI in logistics and its associated technologies can enhance operations, and evaluate which technologies best match your needs and objectives. 
• Explore how AI in the supply chain will integrate with existing processes and infrastructure.
• Pick technology partners carefully. The best tech won’t improve anything if it is hard to implement or the provider is unresponsive to requests for integration help, features upgrades or support.