AI-Powered Citizen Support System for ASAN Services

Abstract

This capstone evaluates whether a conversational‑AI contact centre can relieve mounting workload at Azerbaijan’s ASAN Service 108 hotline without sacrificing the initiative’s hallmark of citizen‑centred care. Adopting a mixed‑methods design, the study conducted five semi‑structured staff interviews and augments these data with international cases from New Orleans, San José, Singapore, Spain, Rwanda and the automotive industry. Three organisational scenarios—a fully automated virtual‑agent model, a hybrid AI‑plus‑human model and a traditional human‑only desk—are stress‑tested against the criteria of effectiveness, efficiency, equity, feasibility and flexibility. Findings reveal that full automation maximises throughput and suppresses marginal costs yet falters on empathy and nuanced judgement, while a purely human desk safeguards emotional intelligence but buckles under surge demand and rising labour expense. Only the hybrid architecture blends rapid, multilingual self‑service with seamless human escalation, halving routine workload, shortening wait times, preserving inclusivity for digitally hesitant users and requiring only staged capital outlay. By dovetailing with Azerbaijan’s 2025–2028 national AI roadmap, a phased, multilingual hybrid rollout can transform the 108 hotline into a 24/7 data‑driven gateway, reposition staff as high‑value advisers and offer a scalable blueprint for neighbouring e‑government services.