15 articles drawn from published research — each paper distilled into a concise, reader-friendly post covering key findings, contributions, and practical implications.
AI anomaly detectors in critical 5G energy IoT systems suffer from opacity, protocol-class bias, and privacy leakage. EAGF operationalizes the EU AI Act into four computable metrics (Transparency, Fairness, Privacy, Accountability), co-optimized into a composite Trust Index.
Digital twins simulate civil infrastructure; agentic AI drives them. This paper investigates how integrating agentic AI within smart-city digital twin frameworks enables real-time autonomous optimization of urban infrastructure, traffic systems, and emergency response with continuous simulation feedback and blockchain auditability.
Assisting neurodivergent individuals requires adaptive, multi-modal intelligence that respects sensory and cognitive preferences. The ADAPT framework deploys a four-agent hierarchy connected via Model Context Protocol (MCP) and a shared blackboard for health and daily routine management.
Disability-inclusive employment matching is hamstrung by data privacy restrictions across institutions. FedAgent-Chain combines differential-privacy federated learning, blockchain audit trails, and multi-agent governance to train fair, multilingual hiring pathways across international borders.
Most planning tools optimize one objective. FinNutriAgent (FNA) optimizes two simultaneously—household budget and personal nutrition—using constrained reasoning pipelines and preference-based alignment to produce plans that are both affordable and healthy.
Achieving UN SDGs 11 and 13 requires urban systems that can adapt dynamically to climate hazards. This PRISMA-guided review synthesizes 70 studies on agentic AI in urban climate resilience, proposing a Pareto-optimized Agentic AI–Digital Twin architecture.
Traffic congestion in urban centers cannot be solved by static timing or black-box RL alone. This paper presents a three-layer framework—Perception, Conceptualization, and Action—combining LangChain reasoning with Model Context Protocol (MCP) and digital twin simulations to optimize signal timings in real time.
AI tools are already in classrooms and research labs. Rather than ban them, this paper provides comprehensive guidelines for ethical AI use in academic settings—balancing genuine pedagogical benefits against integrity, fairness, and attribution risks—with a policy adoption roadmap for institutions.
Wildfire detection demands both speed and accountability. This paper introduces a blockchain-enforced, governance-aware agentic AI architecture that combines hierarchical multi-agent UAV coordination with mandatory human-authorization checkpoints—delivering cryptographically verifiable responsibility without sacrificing adaptive detection performance.
Static trust models collapse in long-duration underwater missions. This paper presents a behavioral trust-monitoring interrogator that dynamically scores underwater agents using a lightweight transformer—storing tamper-proof evidence on a permissioned blockchain—achieving a 21.7% improvement in detection accuracy over static baselines.
Food security meets AI: FinAgent is a price-aware multi-agent system that simultaneously optimizes personal finance and dietary nutrition, achieving a 12–18% cost reduction with over 95% nutritional adequacy in real-world Saudi household simulations.
Can blockchain's immutability solve the auditability crisis in agentic AI? This paper introduces a LangChain-based multi-agent system with Hyperledger Fabric governance, demonstrating trusted autonomous decision-making across healthcare, smart cities, and supply chain applications.
Supply chain disruptions cost businesses trillions annually. This paper presents an agentic AI pipeline that integrates real-time demand forecasting with autonomous procurement decisions, enabling intelligent inventory replenishment under conditions of supply uncertainty.
AI accessibility is about more than screen readers. This work deploys a multi-agent agentic AI system tailored to the daily routines, dietary needs, and cognitive patterns of individuals with disabilities and neurodivergent conditions—making personalized assistance a reality.
Traditional disaster systems treat prediction and response as separate processes. This paper unifies them in a closed-loop agentic AI framework—spanning sensing, forecasting, hydrological modeling, and coordinated evacuation—validated on multi-year radar data from northern Pakistan.
Follow on Google Scholar for the latest publication updates and citation metrics.
Follow on Google Scholar