The transformation of urban logistics toward sustainability and circularity has become a defining challenge of modern cities, particularly in emerging economies where infrastructural constraints hinder efficient delivery systems. This study, examines how artificial intelligence (AI) can be leveraged to enhance last-mile optimization, reduce environmental impact, and promote circular supply chain models. Within the context of peer-to-peer (P2P) delivery frameworks, AI-enabled tools such as predictive routing, micro-fulfillment centers, and dynamic ride-pooling algorithms were deployed to minimize idle time, improve resource utilization, and lower emissions per parcel. The research explores how decentralized data intelligence, when integrated with localized mobility patterns, can achieve efficient delivery in densely populated areas with informal infrastructure. Through adaptive learning models and real-time feedback loops, the system enabled dynamic resource allocation, reducing carbon intensity and operational redundancy. The findings demonstrate that when combined with circular design principles such as reusable packaging and reverse logistics AI-driven P2P networks can simultaneously improve service efficiency and environmental outcomes. This model represents a foundational step in developing urban logistics ecosystems that prioritize both technological advancement and ecological stewardship. By aligning AI-enabled optimization with the principles of the circular economy, it illustrates a pathway toward equitable, low-carbon logistics infrastructure. The work contributes to global discourse on sustainable logistics by highlighting the capacity of intelligent systems to integrate economic inclusion, environmental responsibility, and digital innovation in emerging urban markets.