The rapid rise of online shopping has intensified the complexities of last-mile delivery (LMD), causing both logistical and environmental challenges. High greenhouse gas emissions and inefficiencies in delivering goods directly to consumers’ doorsteps have prompted an urgent need for innovative solutions that reduce transportation costs, carbon footprints, and wasted resources. These factors underscore the motivation of this research, which aims to alleviate environmental harm while enhancing cost-effectiveness in e-commerce logistics.

To address these challenges, this study proposes leveraging Robotic Process Automation (RPA) in tandem with artificial intelligence (AI) to optimize LMD operations. By integrating RPA-driven workflows with generative AI capabilities, real-time customer interactions become possible, allowing for flexible route planning, tailored delivery times, and personalized engagement. The framework also utilizes social media data for advanced insight into consumer preferences, enabling highly targeted and timely logistics strategies.

Figure 1. Proposed Framework for application of Generative AI-Based RPA in Last-mile Delivery Operations

A central contribution of this work lies in its detailed framework that merges AI-enhanced e-commerce interfaces, automated social media engagement, and sustainable delivery options. This model not only empowers customers to choose more eco-friendly shipping methods but also consolidates orders to minimize underutilized vehicle capacity and reduce emissions. By embedding intelligence into the decision loop—prompting route adjustments based on consumer preferences—companies can create a powerful synergy between operational efficiency and environmental responsibility.

The study underscores how data privacy and secure, transparent user consent mechanisms must accompany these technological advancements. It highlights the potential of expanding RPA’s capabilities to balance multiple objectives such as cost, consumer experience, and sustainability and highlights further research into cognitive automation that can adapt to a range of shifting priorities in LMD.