Agentic AI systems—autonomous, goal-driven software entities—are emerging as a transformative paradigm in AI deployment. At their core, agents comprise perception modules, reasoning/planning engines, memory, and action interfaces, often integrated through orchestrators or multi-agent frameworks. Effective agents require robust task understanding, context awareness, secure execution environments, and governance controls. Opportunities span automation of complex workflows, continuous problem-solving, and adaptive decision support; however, risks include hallucination-driven actions, value misalignment, security vulnerabilities, and loss of human oversight. Adoption of agentic systems requires rethinking workflows, introducing human-agent collaboration protocols, and managing cultural change toward trust in autonomy. General-purpose “out-of-the-box” agents offer rapid experimentation but can lack domain precision, whereas domain-specific agents, though costlier to develop, deliver higher reliability in regulated or specialized contexts. Integration capabilities—via APIs, enterprise systems, IoT, and other agents—amplify impact but introduce new interoperability and security challenges. Development paths vary: off-the-shelf frameworks enable faster prototyping, while custom builds optimize for proprietary processes. Deployment at scale demands orchestration, monitoring, and dynamic policy enforcement. Responsible AI principles must be embedded end-to-end with continuous monitoring. This workshop will gather researchers, practitioners, and industry leaders at the forefront of agentic AI to share experiences, evaluate emerging practices, and co-develop actionable guidance for safe, effective, and scalable deployment. Through cross-disciplinary discussion, it aims to distill lessons learned, surface open challenges, and chart a path toward responsible adoption of autonomous AI agents across domains.