Scalable and Domain Aware Agentic AI for Real Data Systems

Agentic artificial intelligence refers to systems that can pursue goals through capabilities such as planning, reasoning, tool use, memory, adaptation, and autonomous or semi-autonomous action. These capabilities may be implemented through a single agent or through multiple agents that coordinate specialized tasks. In real-world data environments, agentic systems must operate across heterogeneous, distributed, and evolving data while respecting domain rules, operational constraints, and regulatory requirements. Although recent advances in large language models, distributed computing, and knowledge integration have accelerated development, major challenges remain in scalability, reliability, observability, security, and failure recovery. This Research Topic focuses on agentic AI systems that combine domain knowledge with scalable deployment and rigorous operational safeguards. It seeks contributions that move beyond conceptual demonstrations toward measurable, reliable, and efficient systems for realistic data-intensive applications.This Research Topic aims to advance scalable, domain-aware, and reliable agentic AI for real-world data systems. It welcomes research on both single-agent and multi-agent architectures, provided that submissions clearly identify the capabilities that make the system agentic. Key goals include integrating domain knowledge, rules, constraints, ontologies, or expert input into planning and decision-making; supporting deployment across cloud, edge, distributed, and legacy infrastructures; and improving long-horizon performance, robustness, observability, and recovery from failures. The topic also seeks stronger methods for evaluating error propagation, secure tool use, human oversight, computational efficiency, and real-world effectiveness. A central objective is to bridge methodological innovation with operational requirements in high-value domains while encouraging reproducible evaluation, appropriate baselines, and transparent discussion of system limitations.The scope includes methods, architectures, benchmarks, and applications for scalable and domain-aware agentic AI systems. Relevant topics include single-agent planning and tool use, multi-agent coordination, memory and adaptation, domain knowledge integration, cloud and edge deployment, interoperability, secure interaction with tools and data, human oversight, observability, failure detection, and recovery. Submissions should demonstrate genuine agentic capabilities and include rigorous evaluation, such as long-horizon task performance, robustness, efficiency, scalability, or comparison with non-agentic baselines. Applications in healthcare, finance, smart infrastructure, and industrial systems are encouraged when they involve meaningful domain constraints and realistic operational settings. General chatbot studies, conventional prediction models, loosely connected multi-agent systems, and application papers without substantive domain integration or agent-level evaluation are outside the intended scope. Topic editors: Tala Talaei Khoei, Aditi Singh Source: https://www.frontiersin.org/research-topics/82366/scalable-and-domain-aware-agentic-ai-for-real-data-systems