The supply chain's last mile is complex and expensive. AI has the potential to fix its woes.

Navigating the complexities of last-mile delivery is a daunting task, often fraught with potential mishaps. From misdelivered packages to delays caused by traffic or adverse weather conditions, the final stretch of the supply chain poses significant challenges. Fred Cook, co-founder and CTO of last-mile delivery firm Veho, emphasizes the unpredictable nature of this process, stating, "You're dealing with humans and the real world and trucks and traffic." In a landscape traditionally dominated by major carriers like UPS, FedEx, and the USPS, innovative software solutions are emerging to address these persistent issues. Companies are now leveraging artificial intelligence (AI) to enhance delivery efficiency, improve accuracy, and elevate the overall customer experience. Erik Mattson, a partner at consulting firm AlixPartners, sees enormous potential for AI to modernize this sector, which is rapidly evolving as e-commerce sales soar, recently hitting $300 billion in the last two quarters alone. As the demand for last-mile delivery grows, it's ripe for technological disruption. A report by McKinsey highlights that over the past decade, logistics startups have attracted around $80 billion in venture capital, with a significant portion directed towards on-demand last-mile delivery services. The intricacy of these routes, which often involve numerous stops and individual parcels rather than bulk deliveries, contributes to the high operational costs—estimated to comprise about 41% of total logistics expenses, according to the Capgemini Research Institute. One of the pioneering uses of AI in this field was UPS's ORION (On-Road Integrated Optimization and Navigation) system, launched in 2013. An upgrade four years later enhanced the system's capabilities, allowing it to adjust routes dynamically based on real-time conditions, reducing travel distances by two to four miles per driver. Today’s AI models can respond instantaneously to changes, unlike earlier static systems that failed to account for evolving circumstances. Andrew Leone, CEO of Dispatch, a last-mile delivery platform, explains, "Compared to pre-AI methods that relied on static routing rules or dispatcher intuition, our platform now responds dynamically to real-world conditions at scale." By analyzing variables like traffic patterns and delivery timeframes, Dispatch's AI-driven technology improves route efficiency, thereby reducing fuel costs and increasing revenue potential for delivery providers. Amazon is significantly advancing the integration of AI in last-mile logistics. Recently, the company introduced Wellspring, an initiative utilizing generative AI to process satellite imagery, building layouts, and delivery data to optimize package drop-off strategies. This innovative approach successfully identified optimal parking locations for deliveries across four million addresses during a recent trial. Moreover, Veho employs AI for quality assurance, allowing the company to analyze delivery data and improve operational practices. Cook notes that while manually evaluating each delivery is impractical, AI can effectively identify patterns and potential issues, enhancing overall service quality. Additionally, real-time tracking has led to a remarkable 80% drop in customer service inquiries for Deliveright, another key player in the sector. AI's role extends to predicting and mitigating theft in last-mile deliveries, an issue highlighted by the $16 billion loss from 58 million stolen packages last year. UPS has developed an AI tool called DeliveryDefense to analyze historical data and identify vulnerable delivery locations. As McCandless points out, AI can forecast high-risk areas and times, allowing companies to adjust their delivery strategies proactively to reduce theft risks. In summary, AI is poised to transform last-mile delivery, offering solutions that not only streamline operations but also enhance security, ultimately revolutionizing how goods reach consumers' doorsteps.