The Logistics Bottleneck: Can AI Finally Solve The Last-Mile Crisis In eCommerce?

The last few miles of a delivery may seem simple, but in the world of e-commerce, this step can be one of the most complex and expensive parts of the supply chain. Companies worldwide are searching for ways to make this stage more efficient and less costly. And in this world, professionals like Udit Agarwal have seen how machine learning and AI can change the game of last-mile delivery.

Udit brings deep experience in building and scaling large-scale delivery networks. He was one of the early members of the last-mile delivery team at Walmart, where he helped expand Walmart’s delivery service to reach over 90% of the U.S. population through more than 3,000 stores. His work contributed to the launch and rapid growth of grocery delivery nationwide, making it easier and more affordable for millions of households to receive everyday essentials at their doorsteps.

During his time at Walmart, Udit worked on some of the most important projects in the company’s delivery initiatives. One key initiative was the launch of Delivery by Store Associates, which allowed in-store employees to fulfil and deliver orders directly to customers. This approach helped Walmart use its existing workforce and store network to reach more homes at a lower cost. Udit was also involved in autonomous delivery pilots with Nuro, testing driverless vehicles for grocery drop-offs.

The impact of these initiatives was significant. Under these initiatives and many more, grocery delivery scaled to 3,000+ stores across the U.S., dramatically increasing Walmart’s reach and efficiency. 

Udit also gained first-hand experience on the large-scale impact that AI can have in addressing last-mile headroom for these large retailers.

As Udit explains, non-AI methods often leave inefficiencies that drive up costs, but AI can evaluate numerous variables at once to create smarter routes and schedules across several parameters and constraints. Lowering delivery cost and making the service accessible are the primary success drivers given heightened competition globally in last-mile delivery across all its use cases (e-commerce 2 days or more, restaurant delivery, same-day delivery).

AI also helps balance catchment size, or the area a store serves. Expanding a store’s reach can boost revenue, but it also raises costs if deliveries are scattered too far apart. AI is uniquely suited to finding the sweet spot where revenue potential and cost efficiency can meet.

Volume is another key factor, where AI can help to find a balance. Udit points out that delivery costs drop with higher volume, which is why companies like Uber try to combine different services, rides, food, and package delivery, to keep costs low. Further, companies like Instacart, with high volume, leverage lower-cost stores in suburban areas to deliver groceries to higher-cost urban areas, such as in San Francisco, providing goods for city dwellers at low costs with an unparalleled last-mile service.

Looking at the current trends, Udit continues to see AI, automation, and machine learning as essentials for tackling last-mile challenges across eCommerce, grocery, and restaurant delivery.

For Udit, his work at Walmart shows that with the appropriate tools and strategies, AI can turn last-mile delivery from a costly bottleneck into a competitive advantage. As e-commerce continues to grow, these innovations will continue to determine how quickly and affordably customers can receive their orders, no matter where they are in the world.

Published on: Saturday, October 25, 2025, 03:31 PM IST