Smart contracts and AI agents for secure last-mile pharmaceutical delivery through crowdsourcing

Pharmaceutical last-mile delivery must combine rapid service with product-integrity controls and a verifiable chain of custody, a challenge that becomes harder when distributors use crowdsourced carriers. Existing blockchain systems preserve traceability but provide limited support for day-to-day operational decisions, whereas AI agents can coordinate complex tasks but may rely on incomplete or unsupported information. This paper addresses that gap through a ledger-grounded agent architecture in which blockchain provides the authoritative operational record and agents use verified platform data, graph-based retrieval, and approved analytical tools to support forecasting, routing, carrier selection, and dispatch. Smart contracts retain control of identity status, assignments, commitments, delivery evidence, and returns, enabling operational intelligence without making the language model the source of truth. The framework also specifies how a carrier’s approved real-world credentials can be linked to a cryptographic identity while keeping personal documents off-chain. Evaluation used five matched simulated operating days representing 1600 orders. Relative to the human-operated workflow, the agent-assisted flow reduced route time by 25.0% and planning time by 78.6%, with consistent improvements in distance, urgent-order handling, assignment readiness, and proof completeness. The findings provide controlled evidence that ledger-grounded agents can add operational intelligence to pharmaceutical traceability while preserving accountability. Production effectiveness, model reliability, identity governance, network performance, and organizational integration require prospective field validation.